NVIDIA CUDA™

Programming Guide

Version 2.3
7/1/2009
ii CUDA Programming Guide Version 2.3
 Table of Contents

Chapter 1. Introduction ..................................................................................... 1 

1.1  From Graphics Processing to General-Purpose Parallel Computing ................... 1 
1.2  CUDA™: a General-Purpose Parallel Computing Architecture ........................... 3 
1.3  CUDA’s Scalable Programming Model ............................................................. 4 
1.4  Document’s Structure ................................................................................... 5 
Chapter 2. Programming Model ......................................................................... 7 
2.1  Kernels ........................................................................................................ 7 
2.2  Thread Hierarchy.......................................................................................... 8 
2.3  Memory Hierarchy ...................................................................................... 10 
2.4  Host and Device ......................................................................................... 11 
2.5  Compute Capability .................................................................................... 14 
Chapter 3. Programming Interface ................................................................. 15 
3.1  Compilation with NVCC ............................................................................... 16 
3.1.1  __noinline__........................................................................................ 16 
3.1.2  #pragma unroll ................................................................................... 17 
3.2  C for CUDA ................................................................................................ 17 
3.2.1  Device Memory .................................................................................... 17 
3.2.2  Shared Memory ................................................................................... 20 
3.2.3  Multiple Devices................................................................................... 26 
3.2.4  Texture Memory .................................................................................. 27 
3.2.4.1  Texture Reference Declaration ....................................................... 28 
3.2.4.2  Runtime Texture Reference Attributes ............................................ 28 
3.2.4.3  Texture Binding ............................................................................ 29 
3.2.5  Page-Locked Host Memory ................................................................... 32 
3.2.5.1  Portable Memory ........................................................................... 32 
3.2.5.2  Write-Combining Memory .............................................................. 32 
3.2.5.3  Mapped Memory ........................................................................... 32 
3.2.6  Asynchronous Concurrent Execution ..................................................... 33 

CUDA Programming Guide Version 2.3 iii

 3.2.6.1  Stream ......................................................................................... 34 
3.2.6.2  Event ........................................................................................... 35 
3.2.6.3  Synchronous Calls ......................................................................... 35 
3.2.7  OpenGL Interoperability ....................................................................... 36 
3.2.8  Direct3D Interoperability ...................................................................... 38 
3.2.9  Error Handling ..................................................................................... 43 
3.2.10  Debugging using the Device Emulation Mode ........................................ 44 
3.3  Driver API .................................................................................................. 45 
3.3.1  Context ............................................................................................... 47 
3.3.2  Module ................................................................................................ 48 
3.3.3  Kernel Execution .................................................................................. 49 
3.3.4  Device Memory .................................................................................... 50 
3.3.5  Shared Memory ................................................................................... 53 
3.3.6  Multiple Devices................................................................................... 54 
3.3.7  Texture Memory .................................................................................. 55 
3.3.8  Page-Locked Host Memory ................................................................... 57 
3.3.9  Asynchronous Concurrent Execution ..................................................... 57 
3.3.9.1  Stream ......................................................................................... 57 
3.3.9.2  Event Management ....................................................................... 58 
3.3.9.3  Synchronous Calls ......................................................................... 59 
3.3.10  OpenGL Interoperability ....................................................................... 59 
3.3.11  Direct3D Interoperability ...................................................................... 61 
3.3.12  Error Handling ..................................................................................... 67 
3.4  Versioning and Compatibility ....................................................................... 67 
3.5  Compute Modes ......................................................................................... 68 
3.6  Mode Switches ........................................................................................... 69 
Chapter 4. Hardware Implementation ............................................................ 71 
4.1  A Set of SIMT Multiprocessors with On-Chip Shared Memory......................... 71 
4.2  Multiple Devices ......................................................................................... 75 
Chapter 5. Performance Guidelines ................................................................. 77 
5.1  Instruction Performance ............................................................................. 77 
5.1.1  Instruction Throughput ........................................................................ 77 
5.1.1.1  Arithmetic Instructions .................................................................. 77 

iv CUDA Programming Guide Version 2.3

 5.1.1.2  Control Flow Instructions ............................................................... 79 
5.1.1.3  Memory Instructions ..................................................................... 80 
5.1.1.4  Synchronization Instruction ........................................................... 80 
5.1.2  Memory Bandwidth .............................................................................. 80 
5.1.2.1  Global Memory.............................................................................. 81 
5.1.2.2  Local Memory ............................................................................... 88 
5.1.2.3  Constant Memory .......................................................................... 89 
5.1.2.4  Texture Memory ........................................................................... 89 
5.1.2.5  Shared Memory ............................................................................ 89 
5.1.2.6  Registers ...................................................................................... 97 
5.2  Execution Configuration .............................................................................. 97 
5.3  Data Transfer between Host and Device ...................................................... 98 
5.4  Warp-Level Synchronization ........................................................................ 99 
5.5  Overall Performance Optimization Strategies ................................................ 99 
Appendix A. Technical Specifications ............................................................ 101 
A.1  General Specifications ............................................................................... 101 
A.1.1  Specifications for Compute Capability 1.0 ............................................ 102 
A.1.2  Specifications for Compute Capability 1.1 ............................................ 103 
A.1.3  Specifications for Compute Capability 1.2 ............................................ 103 
A.1.4  Specifications for Compute Capability 1.3 ............................................ 103 
A.2  Floating-Point Standard ............................................................................ 103 
Appendix B. C Extensions .............................................................................. 105 
B.1  Function Type Qualifiers ........................................................................... 105 
B.1.1  __device__ ........................................................................................ 105 
B.1.2  __global__ ........................................................................................ 105 
B.1.3  __host__ ........................................................................................... 105 
B.1.4  Restrictions ....................................................................................... 106 
B.2  Variable Type Qualifiers ............................................................................ 106 
B.2.1  __device__ ........................................................................................ 106 
B.2.2  __constant__ .................................................................................... 106 
B.2.3  __shared__ ....................................................................................... 107 
B.2.4  Volatile.............................................................................................. 107 
B.2.5  Restrictions ....................................................................................... 108 

CUDA Programming Guide Version 2.3 v

 B.3  Built-in Vector Types ................................................................................ 108 
B.3.1  char1, uchar1, char2, uchar2, char3, uchar3, char4, uchar4, short1,
ushort1, short2, ushort2, short3, ushort3, short4, ushort4, int1, uint1, int2, uint2,
int3, uint3, int4, uint4, long1, ulong1, long2, ulong2, long3, ulong3, long4, ulong4,
longlong1, longlong2, float1, float2, float3, float4, double1, double2 .................. 108 
B.3.2  dim3 ................................................................................................. 110 
B.4  Built-in Variables ...................................................................................... 110 
B.4.1  gridDim ............................................................................................. 110 
B.4.2  blockIdx ............................................................................................ 110 
B.4.3  blockDim ........................................................................................... 110 
B.4.4  threadIdx .......................................................................................... 110 
B.4.5  warpSize ........................................................................................... 110 
B.4.6  Restrictions ....................................................................................... 110 
B.5  Memory Fence Functions .......................................................................... 110 
B.6  Synchronization Function .......................................................................... 112 
B.7  Mathematical Functions ............................................................................ 112 
B.8  Texture Functions..................................................................................... 113 
B.8.1  tex1Dfetch() ...................................................................................... 113 
B.8.2  tex1D() ............................................................................................. 114 
B.8.3  tex2D() ............................................................................................. 114 
B.8.4  tex3D() ............................................................................................. 114 
B.9  Time Function .......................................................................................... 114 
B.10  Atomic Functions ...................................................................................... 115 
B.10.1  Arithmetic Functions .......................................................................... 115 
B.10.1.1  atomicAdd()................................................................................ 115 
B.10.1.2  atomicSub() ................................................................................ 115 
B.10.1.3  atomicExch() .............................................................................. 115 
B.10.1.4  atomicMin() ................................................................................ 116 
B.10.1.5  atomicMax() ............................................................................... 116 
B.10.1.6  atomicInc() ................................................................................. 116 
B.10.1.7  atomicDec() ................................................................................ 116 
B.10.1.8  atomicCAS() ............................................................................... 116 
B.10.2  Bitwise Functions ............................................................................... 117 
B.10.2.1  atomicAnd()................................................................................ 117 

vi CUDA Programming Guide Version 2.3

 B.10.2.2  atomicOr() .................................................................................. 117 
B.10.2.3  atomicXor() ................................................................................ 117 
B.11  Warp Vote Functions ................................................................................ 117 
B.12  Execution Configuration ............................................................................ 118 
Appendix C. Mathematical Functions ............................................................ 119 
C.1  Standard Functions ................................................................................... 119 
C.1.1  Single-Precision Floating-Point Functions ............................................. 119 
C.1.2  Double-Precision Floating-Point Functions ........................................... 121 
C.1.3  Integer Functions .............................................................................. 123 
C.2  Intrinsic Functions .................................................................................... 124 
C.2.1  Single-Precision Floating-Point Functions ............................................. 124 
C.2.2  Double-Precision Floating-Point Functions ........................................... 125 
C.2.3  Integer Functions .............................................................................. 126 
Appendix D. C++ Language Constructs ........................................................ 129 
D.1  Polymorphism .......................................................................................... 130 
D.2  Default Parameters ................................................................................... 130 
D.3  Operator Overloading ............................................................................... 131 
D.4  Namespaces............................................................................................. 131 
D.5  Function Templates .................................................................................. 132 
Appendix E. Texture Fetching ........................................................................ 135 
E.1  Nearest-Point Sampling............................................................................. 136 
E.2  Linear Filtering ......................................................................................... 136 
E.3  Table Lookup ........................................................................................... 138 

CUDA Programming Guide Version 2.3 vii

 List of Figures

Figure 1-1.  Floating-Point Operations per Second and Memory Bandwidth for the CPU
and GPU 2 
Figure 1-2.  The GPU Devotes More Transistors to Data Processing ............................ 3 
Figure 1-3.  CUDA is Designed to Support Various Languages or Application
Programming Interfaces ...................................................................................... 4 
Figure 2-1.  Grid of Thread Blocks ........................................................................... 10 
Figure 2-2.  Memory Hierarchy ............................................................................... 11 
Figure 2-3.  Heterogeneous Programming ............................................................... 13 
Figure 3-1.  Matrix Multipliation without Shared Memory .......................................... 22 
Figure 3-2.  Matrix Multipliation with Shared Memory ............................................... 26 
Figure 3-3.  Library Context Management ................................................................ 48 
Figure 3-4.  The Driver API is Forward Compatible ................................................... 68 
Figure 4-1.  Automatic Scalability ............................................................................ 72 
Figure 4-2.  Hardware Model .................................................................................. 75 
Figure 5-1.  Examples of Coalesced Global Memory Access Patterns .......................... 84 
Figure 5-2.  Examples of Global Memory Access Patterns That Are Non-Coalesced for
Devices of Compute Capability 1.0 or 1.1 ........................................................... 85 
Figure 5-3.  Examples of Global Memory Access Patterns That Are Non-Coalesced for
Devices of Compute Capability 1.0 or 1.1 ........................................................... 86 
Figure 5-4.  Examples of Global Memory Access by Devices with Compute Capability
1.2 and Higher .................................................................................................. 87 
Figure 5-5.  Examples of Shared Memory Access Patterns without Bank Conflicts ..... 93 
Figure 5-6.  Example of a Shared Memory Access Pattern without Bank Conflicts ...... 94 
Figure 5-7.  Examples of Shared Memory Access Patterns with Bank Conflicts ........... 95 
Figure 5-8.  Example of Shared Memory Read Access Patterns with Broadcast........... 96 

viii CUDA Programming Guide Version 2.3

 Chapter 1.
Introduction

1.1 From Graphics Processing to

General-Purpose Parallel Computing
Driven by the insatiable market demand for realtime, high-definition 3D graphics,
the programmable Graphic Processor Unit or GPU has evolved into a highly
parallel, multithreaded, manycore processor with tremendous computational
horsepower and very high memory bandwidth, as illustrated by Figure 1-1.

CUDA Programming Guide Version 2.3 1

Chapter 1. Introduction

GT200

G80 G92
Ultra
G80

G71

G70
NV40 3.2 GHz
NV35 3.0 GHz Harpertown
NV30 Core2 Duo

Jan Jun Apr Jun Mar Nov May Jun

2003 2004 2005 2006 2007 2008

GT200 = GeForce GTX 280 G71 = GeForce 7900 GTX NV35 = GeForce FX 5950 Ultra

G92 = GeForce 9800 GTX G70 = GeForce 7800 GTX NV30 = GeForce FX 5800

G80 = GeForce 8800 GTX NV40 = GeForce 6800 Ultra

G80
Ultra

G80

G71

NV40
Harpertown
Woodcrest
NV30
Prescott EE
Northwood

Figure 1-1.Floating-Point Operations per Second and Memory

Bandwidth for the CPU and GPU

The reason behind the discrepancy in floating-point capability between the CPU and
the GPU is that the GPU is specialized for compute-intensive, highly parallel
computation – exactly what graphics rendering is about – and therefore designed
such that more transistors are devoted to data processing rather than data caching
and flow control, as schematically illustrated by Figure 1-2.

2 CUDA Programming Guide Version 2.3

 Chapter 1. Introduction

Control ALU ALU

ALU ALU

Cache

DRAM DRAM

CPU GPU

Figure 1-2.The GPU Devotes More Transistors to Data

Processing

More specifically, the GPU is especially well-suited to address problems that can be
expressed as data-parallel computations – the same program is executed on many
data elements in parallel – with high arithmetic intensity – the ratio of arithmetic
operations to memory operations. Because the same program is executed for each
data element, there is a lower requirement for sophisticated flow control; and
because it is executed on many data elements and has high arithmetic intensity, the
memory access latency can be hidden with calculations instead of big data caches.
Data-parallel processing maps data elements to parallel processing threads. Many
applications that process large data sets can use a data-parallel programming model
to speed up the computations. In 3D rendering, large sets of pixels and vertices are
mapped to parallel threads. Similarly, image and media processing applications such
as post-processing of rendered images, video encoding and decoding, image scaling,
stereo vision, and pattern recognition can map image blocks and pixels to parallel
processing threads. In fact, many algorithms outside the field of image rendering
and processing are accelerated by data-parallel processing, from general signal
processing or physics simulation to computational finance or computational biology.

1.2 CUDA™: a General-Purpose Parallel

Computing Architecture
In November 2006, NVIDIA introduced CUDA™, a general purpose parallel
computing architecture – with a new parallel programming model and instruction
set architecture – that leverages the parallel compute engine in NVIDIA GPUs to
solve many complex computational problems in a more efficient way than on a
CPU.
CUDA comes with a software environment that allows developers to use C as a
high-level programming language. As illustrated by Figure 1-3, other languages or
application programming interfaces will be supported in the future, such as
FORTRAN, C++, OpenCL, and DirectX Compute.

CUDA Programming Guide Version 2.3 3

Chapter 1. Introduction

Figure 1-3.CUDA is Designed to Support Various Languages

or Application Programming Interfaces

1.3 CUDA’s Scalable Programming Model

The advent of multicore CPUs and manycore GPUs means that mainstream
processor chips are now parallel systems. Furthermore, their parallelism continues
to scale with Moore’s law. The challenge is to develop application software that
transparently scales its parallelism to leverage the increasing number of processor
cores, much as 3D graphics applications transparently scale their parallelism to
manycore GPUs with widely varying numbers of cores.
CUDA’s parallel programming model is designed to overcome this challenge while
maintaining a low learning curve for programmers familiar with standard
programming languages such as C.
At its core are three key abstractions – a hierarchy of thread groups, shared
memories, and barrier synchronization – that are simply exposed to the programmer
as a minimal set of language extensions.
These abstractions provide fine-grained data parallelism and thread parallelism,
nested within coarse-grained data parallelism and task parallelism. They guide the
programmer to partition the problem into coarse sub-problems that can be solved
independently in parallel, and then into finer pieces that can be solved cooperatively
in parallel. Such a decomposition preserves language expressivity by allowing
threads to cooperate when solving each sub-problem, and at the same time enables
transparent scalability since each sub-problem can be scheduled to be solved on any
of the available processor cores: A compiled CUDA program can therefore execute
on any number of processor cores, and only the runtime system needs to know the
physical processor count.
This scalable programming model allows the CUDA architecture to span a wide
market range by simply scaling the number of processors and memory partitions:

4 CUDA Programming Guide Version 2.3

 Chapter 1. Introduction

from the high-performance enthusiast GeForce GTX 280 GPU and professional
Quadro and Tesla computing products to a variety of inexpensive, mainstream
GeForce GPUs (see Appendix A for a list of all CUDA-enabled GPUs).

1.4 Document’s Structure

This document is organized into the following chapters:
‰ Chapter 1 is a general introduction to CUDA.
‰ Chapter 2 outlines CUDA’s programming model.
‰ Chapter 3 describes the programming interface.
‰ Chapter 4 describes the hardware implementation.
‰ Chapter 5 gives some guidance on how to achieve maximum performance.
‰ Appendix A lists the technical specifications of various devices.
‰ Appendix B is a detailed description of all extensions to the C language.
‰ Appendix C lists the mathematical functions supported in CUDA.
‰ Appendix D lists the C++ constructs supported in device code.
‰ Appendix E gives more details on texture fetching.

CUDA Programming Guide Version 2.3 5

 Chapter 2.
Programming Model

This chapter introduces the main concepts that make up the CUDA programming
model by outlining how they are exposed in C. An extensive description of C for
CUDA is given in Section 3.2.

2.1 Kernels
C for CUDA extends C by allowing the programmer to define C functions, called
kernels, that, when called, are executed N times in parallel by N different CUDA
threads, as opposed to only once like regular C functions.
A kernel is defined using the __global__ declaration specifier and the number of
CUDA threads for each call is specified using a new <<<…>>> syntax:
// Kernel definition
__global__ void VecAdd(float* A, float* B, float* C)
{
}

int main()
{
// Kernel invocation
VecAdd<<<1, N>>>(A, B, C);
}
Each of the threads that execute a kernel is given a unique thread ID that is
accessible within the kernel through the built-in threadIdx variable. As an
illustration, the following sample code adds two vectors A and B of size N and
stores the result into vector C:
// Kernel definition
__global__ void VecAdd(float* A, float* B, float* C)
{
int i = threadIdx.x;
C[i] = A[i] + B[i];
}

int main()
{
// Kernel invocation
VecAdd<<<1, N>>>(A, B, C);

CUDA Programming Guide Version 2.3 7

Chapter 2. Programming Model

}
Each of the threads that execute VecAdd() performs one pair-wise addition.

2.2 Thread Hierarchy

For convenience, threadIdx is a 3-component vector, so that threads can be
identified using a one-dimensional, two-dimensional, or three-dimensional thread
index, forming a one-dimensional, two-dimensional, or three-dimensional thread
block. This provides a natural way to invoke computation across the elements in a
domain such as a vector, matrix, or field. As an example, the following code adds
two matrices A and B of size NxN and stores the result into matrix C:
// Kernel definition
__global__ void MatAdd(float A[N][N], float B[N][N],
float C[N][N])
{
int i = threadIdx.x;
int j = threadIdx.y;
C[i][j] = A[i][j] + B[i][j];
}

int main()
{
// Kernel invocation
dim3 dimBlock(N, N);
MatAdd<<<1, dimBlock>>>(A, B, C);
}
The index of a thread and its thread ID relate to each other in a straightforward
way: For a one-dimensional block, they are the same; for a two-dimensional block
of size (Dx, Dy), the thread ID of a thread of index (x, y) is (x + y Dx); for a three-
dimensional block of size (Dx, Dy, Dz), the thread ID of a thread of index (x, y, z) is
(x + y Dx + z Dx Dy).
Threads within a block can cooperate among themselves by sharing data through
some shared memory and synchronizing their execution to coordinate memory
accesses. More precisely, one can specify synchronization points in the kernel by
calling the __syncthreads() intrinsic function; __syncthreads() acts as a
barrier at which all threads in the block must wait before any is allowed to proceed.
Section 3.2.2 gives an example of using shared memory.
For efficient cooperation, the shared memory is expected to be a low-latency
memory near each processor core, much like an L1 cache, __syncthreads() is
expected to be lightweight, and all threads of a block are expected to reside on the
same processor core. The number of threads per block is therefore restricted by the
limited memory resources of a processor core. On current GPUs, a thread block
may contain up to 512 threads.
However, a kernel can be executed by multiple equally-shaped thread blocks, so that
the total number of threads is equal to the number of threads per block times the
number of blocks. These multiple blocks are organized into a one-dimensional or
two-dimensional grid of thread blocks as illustrated by Figure 2-1. The dimension of
the grid is specified by the first parameter of the <<<…>>> syntax. Each block
within the grid can be identified by a one-dimensional or two-dimensional index

8 CUDA Programming Guide Version 2.3

 Chapter 2: Programming Model

accessible within the kernel through the built-in blockIdx variable. The dimension
of the thread block is accessible within the kernel through the built-in blockDim
variable. The previous sample code becomes:
// Kernel definition
__global__ void MatAdd(float A[N][N], float B[N][N],
float C[N][N])
{
int i = blockIdx.x * blockDim.x + threadIdx.x;
int j = blockIdx.y * blockDim.y + threadIdx.y;
if (i < N && j < N)
C[i][j] = A[i][j] + B[i][j];
}

int main()
{
// Kernel invocation
dim3 dimBlock(16, 16);
dim3 dimGrid((N + dimBlock.x – 1) / dimBlock.x,
(N + dimBlock.y – 1) / dimBlock.y);
MatAdd<<<dimGrid, dimBlock>>>(A, B, C);
}
The thread block size of 16x16 = 256 threads was chosen somewhat arbitrarily, and
a grid is created with enough blocks to have one thread per matrix element as
before.
Thread blocks are required to execute independently: It must be possible to execute
them in any order, in parallel or in series. This independence requirement allows
thread blocks to be scheduled in any order across any number of cores, enabling
programmers to write code that scales with the number of cores.
The number of thread blocks in a grid is typically dictated by the size of the data
being processed rather than by the number of processors in the system, which it can
greatly exceed.

CUDA Programming Guide Version 2.3 9

Chapter 2. Programming Model

Grid

Block (0, 0) Block (1, 0) Block (2, 0)

Block (0, 1) Block (1, 1) Block (2, 1)

Block (1, 1)

Thread (0, 0) Thread (1, 0) Thread (2, 0) Thread (3, 0)

Thread (0, 1) Thread (1, 1) Thread (2, 1) Thread (3, 1)

Thread (0, 2) Thread (1, 2) Thread (2, 2) Thread (3, 2)

Figure 2-1.Grid of Thread Blocks

2.3 Memory Hierarchy

CUDA threads may access data from multiple memory spaces during their
execution as illustrated by Figure 2-2. Each thread has a private local memory. Each
thread block has a shared memory visible to all threads of the block and with the
same lifetime as the block. Finally, all threads have access to the same global
memory.
There are also two additional read-only memory spaces accessible by all threads: the
constant and texture memory spaces. The global, constant, and texture memory
spaces are optimized for different memory usages (see Sections 5.1.2.1, 5.1.2.3, and
5.1.2.4). Texture memory also offers different addressing modes, as well as data
filtering, for some specific data formats (see Section 3.2.4).
The global, constant, and texture memory spaces are persistent across kernel
launches by the same application.

10 CUDA Programming Guide Version 2.3

 Chapter 2: Programming Model

Thread
Per-thread local
memory

Thread Block
Per-block shared
memory

Grid 0

Block (0, 0) Block (1, 0) Block (2, 0)

Block (0, 1) Block (1, 1) Block (2, 1)

Grid 1
Global memory
Block (0, 0) Block (1, 0)

Block (0, 1) Block (1, 1)

Block (0, 2) Block (1, 2)

Figure 2-2.Memory Hierarchy

2.4 Host and Device

As illustrated by Figure 2-3, CUDA’s programming model assumes that the CUDA
threads execute on a physically separate device that operates as a coprocessor to the
host running the C program. This is the case, for example, when the kernels execute
on a GPU and the rest of the C program executes on a CPU.

CUDA Programming Guide Version 2.3 11

Chapter 2. Programming Model

CUDA’s programming model also assumes that both the host and the device
maintain their own DRAM, referred to as host memory and device memory, respectively.
Therefore, a program manages the global, constant, and texture memory spaces
visible to kernels through calls to the CUDA runtime (described in Chapter 3). This
includes device memory allocation and deallocation, as well as data transfer between
host and device memory.

12 CUDA Programming Guide Version 2.3

 Chapter 2: Programming Model

C Program
Sequential
Execution

Serial code Host

Parallel kernel Device

Kernel0<<<>>>() Grid 0

Block (0, 0) Block (1, 0) Block (2, 0)

Block (0, 1) Block (1, 1) Block (2, 1)

Serial code Host

Device
Parallel kernel
Grid 1
Kernel1<<<>>>()

Block (0, 0) Block (1, 0)

Block (0, 1) Block (1, 1)

Block (0, 2) Block (1, 2)

Serial code executes on the host while parallel code executes on the device.

Figure 2-3.Heterogeneous Programming

CUDA Programming Guide Version 2.3 13

Chapter 2. Programming Model

2.5 Compute Capability

The compute capability of a device is defined by a major revision number and a minor
revision number.
Devices with the same major revision number are of the same core architecture. The
devices listed in Appendix A are all of compute capability 1.x (Their major revision
number is 1).
The minor revision number corresponds to an incremental improvement to the core
architecture, possibly including new features.
The technical specifications of the various compute capabilities are given in
Appendix A.

14 CUDA Programming Guide Version 2.3

 Chapter 3.
Programming Interface

Two interfaces are currently supported to write CUDA programs: C for CUDA and
the CUDA driver API. They are mutually exclusive: A program must use either one
or the other.
C for CUDA exposes the CUDA programming model as a minimal set of
extensions to the C language. Any source file that contains some of these extensions
must be compiled with nvcc as outlined in Section 3.1. These extensions allow
programmers to define a kernel as a C function and use some new syntax to specify
the grid and block dimension each time the function is called.
The CUDA driver API is a lower-level C API that provides functions to load
kernels as modules of CUDA binary or assembly code, to inspect their parameters,
and to launch them. Binary or assembly code are usually obtained by compiling
kernels written in C.
C for CUDA comes with a runtime API and both the runtime API and the driver
API provide functions to allocate and deallocate device memory, transfer data
between host memory and device memory, manage systems with multiple devices,
etc.
The runtime API is built on top of the CUDA driver API. Initialization, context,
and module management are all implicit and resulting code is more concise. C for
CUDA also supports device emulation, which facilitates debugging (see
Section 3.2.9).
In contrast, the CUDA driver API requires more code, is harder to program and
debug, but offers a better level of control and is language-independent since it
handles binary or assembly code.
Section 3.2 continues the description of C for CUDA started in Chapter 2. It also
introduces concepts that are common to both C for CUDA and the driver API:
linear memory, CUDA arrays, shared memory, texture memory, page-locked host
memory, device enumeration, asynchronous execution, interoperability with
graphics APIs. Section 3.3 assumes knowledge of these concepts and describes how
they are exposed by the driver API.

CUDA Programming Guide Version 2.3 15

Chapter 3. Programming Interface

3.1 Compilation with NVCC

Kernels can be written using CUDA’s instruction set architecture, called PTX,
which is described in a separate document. It is however usually more effective to
use a high-level programming language such as C. In both cases, kernels must be
compiled into binary code by nvcc.
nvcc is a compiler driver that simplifies the process of compiling C for CUDA
code: It provides simple and familiar command line options and executes them by
invoking the collection of tools that implement the different compilation stages.
Source files can include a mix of host code (i.e. code that executes on the host) and
device code (i.e. code that executes on the device). nvcc’s basic workflow consists
in separating device code from host code and compiling the device code into an
assembly form (PTX code) or binary form (cubin object). The generated host code is
output either as C code that is left to be compiled using another tool or as object
code directly by invoking the host compiler during the last compilation stage.
Applications can either ignore the generated host code (if any) and load and execute
the PTX code or cubin object on the device using the CUDA driver API (see
Section 3.3), or they can link to the generated host code, which includes the cubin
object as a global initialized data array and contains a translation of the execution
configuration syntax described in Section B.12 into the necessary C for CUDA
runtime startup code to load and launch each compiled kernel.
The front end of the compiler processes CUDA source files according to C++
syntax rules. Full C++ is supported for the host code. However, only the C subset
of C++ is fully supported for the device code; C++ specific features such as classes,
inheritance, or declaration of variables within basic blocks are not. As a
consequence of the use of C++ syntax rules, void pointers (e.g. returned by
malloc()) cannot be assigned to non-void pointers without a typecast.
nvcc introduces two compiler directives described in the following sections.
Some PTX instructions are only supported on devices of higher compute
capabilities. For example, atomic instructions on global memory are only supported
on devices of compute capability 1.1 and above; double-precision instructions are
only supported on devices of compute capability 1.3 and above. The –arch
compiler option specifies the compute capability that is assumed when compiling to
PTX code. So, code that contains double-precision arithmetic, for example, must be
compiled with “-arch sm_13” (or higher compute capability), otherwise double-
precision arithmetic will get demoted to single-precision arithmetic.
A detailed description of nvcc’s workflow and command options can be found in a
separate document.

3.1.1 __noinline__
By default, a __device__ function is always inlined. The __noinline__
function qualifier however can be used as a hint for the compiler not to inline the
function if possible. The function body must still be in the same file where it is
called.

16 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

The compiler will not honor the __noinline__ qualifier for functions with
pointer parameters and for functions with large parameter lists.

3.1.2 #pragma unroll

By default, the compiler unrolls small loops with a known trip count. The #pragma
unroll directive however can be used to control unrolling of any given loop. It
must be placed immediately before the loop and only applies to that loop. It is
optionally followed by a number that specifies how many times the loop must be
unrolled.
For example, in this code sample:
#pragma unroll 5
for (int i = 0; i < n; ++i)
the loop will be unrolled 5 times. It is up to the programmer to make sure that
unrolling will not affect the correctness of the program (which it might, in the above
example, if n is smaller than 5).
#pragma unroll 1 will prevent the compiler from ever unrolling a loop.
If no number is specified after #pragma unroll, the loop is completely unrolled
if its trip count is constant, otherwise it is not unrolled at all.

3.2 C for CUDA

C for CUDA provides a simple path for users familiar with the C programming
language to easily write programs for execution by the device.
It consists of a minimal set of extensions to the C language and a runtime library.
The core language extensions have been introduced in Chapter 2. This section
continues with an introduction to the runtime. A complete description of all
extensions can be found in Appendix B and a complete description of the runtime
in the CUDA reference manual.
The runtime is implemented in the cudart dynamic library and all its entry points
are prefixed with cuda.
There is no explicit initialization function for the runtime; it initializes the first time
a runtime function is called. One needs to keep this in mind when timing runtime
function calls and when interpreting the error code from the first call into the
runtime.
On system with multiple devices, kernels are executed on device 0 by default as
detailed in Section 3.2.3.

3.2.1 Device Memory

As mentioned in Section 2.4, CUDA’s programming model assumes a system
composed of a host and a device, each with their own separate memory. Kernels
can only operate out of device memory, so the runtime provides functions to
allocate, deallocate, and copy device memory, as well as transfer data between host
memory and device memory.

CUDA Programming Guide Version 2.3 17

Chapter 3. Programming Interface

Device memory can be allocated either as linear memory or as CUDA arrays.

CUDA arrays are opaque memory layouts optimized for texture fetching. They are
described in Section 3.2.4.
Linear memory exists on the device in a 32-bit address space, so separately allocated
entities can reference one another via pointers, for example, in a binary tree.
Linear memory is typically allocated using cudaMalloc() and freed using
cudaFree() and data transfer between host memory and device memory are
typically done using cudaMemcpy(). In the vector addition code sample of
Section 2.1, the vectors need to be copied from host memory to device memory:
// Device code
__global__ void VecAdd(float* A, float* B, float* C)
{
int i = threadIdx.x;
if (i < N)
C[i] = A[i] + B[i];
}

// Host code
int main()
{
// Allocate vectors in device memory
size_t size = N * sizeof(float);
float* d_A;
cudaMalloc((void**)&d_A, size);
float* d_B;
cudaMalloc((void**)&d_B, size);
float* d_C;
cudaMalloc((void**)&d_C, size);

// Copy vectors from host memory to device memory

// h_A and h_B are input vectors stored in host memory
cudaMemcpy(d_A, h_A, size, cudaMemcpyHostToDevice);
cudaMemcpy(d_B, h_B, size, cudaMemcpyHostToDevice);

// Invoke kernel
int threadsPerBlock = 256;
int blocksPerGrid =
(N + threadsPerBlock – 1) / threadsPerBlock;
VecAdd<<<blocksPerGrid, threadsPerBlock>>>(d_A, d_B, d_C);

// Copy result from device memory to host memory

// h_C contains the result in host memory
cudaMemcpy(h_C, d_C, size, cudaMemcpyDeviceToHost);

// Free device memory

cudaFree(d_A);
cudaFree(d_B);
cudaFree(d_C);
}
Linear memory can also be allocated through cudaMallocPitch() and
cudaMalloc3D(). These functions are recommended for allocations of 2D or 3D
arrays as it makes sure that the allocation is appropriately padded to meet the
alignment requirements described in Section 5.1.2.1, therefore ensuring best

18 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

performance when accessing the row addresses or performing copies between 2D

arrays and other regions of device memory (using the cudaMemcpy2D() and
cudaMemcpy3D() functions). The returned pitch (or stride) must be used to access
array elements. The following code sample allocates a width×height 2D array of
floating-point values and shows how to loop over the array elements in device code:
// Host code
float* devPtr;
int pitch;
cudaMallocPitch((void**)&devPtr, &pitch,
width * sizeof(float), height);
myKernel<<<100, 512>>>(devPtr, pitch);

// Device code
__global__ void myKernel(float* devPtr, int pitch)
{
for (int r = 0; r < height; ++r) {
float* row = (float*)((char*)devPtr + r * pitch);
for (int c = 0; c < width; ++c) {
float element = row[c];
}
}
}
The following code sample allocates a width×height×depth 3D array of
floating-point values and shows how to loop over the array elements in device code:
// Host code
cudaPitchedPtr devPitchedPtr;
cudaExtent extent = make_cudaExtent(64, 64, 64);
cudaMalloc3D(&devPitchedPtr, extent);
myKernel<<<100, 512>>>(devPitchedPtr, extent);

// Device code
__global__ void myKernel(cudaPitchedPtr devPitchedPtr,
cudaExtent extent)
{
char* devPtr = devPitchedPtr.ptr;
size_t pitch = devPitchedPtr.pitch;
size_t slicePitch = pitch * extent.height;
for (int z = 0; z < extent.depth; ++z) {
char* slice = devPtr + z * slicePitch;
for (int y = 0; y < extent.height; ++y) {
float* row = (float*)(slice + y * pitch);
for (int x = 0; x < extent.width; ++x) {
float element = row[x];
}
}
}
}
The reference manual lists all the various functions used to copy memory between
linear memory allocated with cudaMalloc(), linear memory allocated with
cudaMallocPitch() or cudaMalloc3D(), CUDA arrays, and memory
allocated for variables declared in global or constant memory space.
The following code sample copies the 2D array to the CUDA array allocated in the
previous code samples:

CUDA Programming Guide Version 2.3 19

Chapter 3. Programming Interface

cudaMemcpy2DToArray(cuArray, 0, 0, devPtr, pitch,

width * sizeof(float), height,
cudaMemcpyDeviceToDevice);
The following code sample copies some host memory array to constant memory:
__constant__ float constData[256];
float data[256];
cudaMemcpyToSymbol(constData, data, sizeof(data));
cudaGetSymbolAddress() is used to retrieve the address pointing to the
memory allocated for a variable declared in global memory space. The size of the
allocated memory is obtained through cudaGetSymbolSize().

3.2.2 Shared Memory

As detailed in Section B.2 shared memory is allocated using the __shared__
qualifier.
Shared memory is expected to be much faster than global memory as mentioned in
Section 2.2 and detailed in Section 5.1.2.5. Any opportunity to replace global
memory accesses by shared memory accesses should therefore be exploited as
illustrated by the following matrix multiplication example.
The following code sample is a straightforward implementation of matrix
multiplication that does not take advantage of shared memory. Each thread reads
one row of A and one column of B and computes the corresponding element of C
as illustrated in Figure 3-1. A is therefore read B.width times from global memory
and B is read A.height times.
// Matrices are stored in row-major order:
// M(row, col) = *(M.elements + row * M.width + col)
typedef struct {
int width;
int height;
float* elements;
} Matrix;

// Thread block size

#define BLOCK_SIZE 16

// Forward declaration of the matrix multiplication kernel

__global__ void MatMulKernel(const Matrix, const Matrix, Matrix);

// Matrix multiplication - Host code

// Matrix dimensions are assumed to be multiples of BLOCK_SIZE
void MatMul(const Matrix A, const Matrix B, Matrix C)
{
// Load A and B to device memory
Matrix d_A;
d_A.width = A.width; d_A.height = A.height;
size_t size = A.width * A.height * sizeof(float);
cudaMalloc((void**)&d_A.elements, size);
cudaMemcpy(d_A.elements, A.elements, size,
cudaMemcpyHostToDevice);
Matrix d_B;
d_B.width = B.width; d_B.height = B.height;
size = B.width * B.height * sizeof(float);

20 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

cudaMalloc((void**)&d_B.elements, size);
cudaMemcpy(d_B.elements, B.elements, size,
cudaMemcpyHostToDevice);

// Allocate C in device memory

Matrix d_C;
d_C.width = C.width; d_C.height = C.height;
size = C.width * C.height * sizeof(float);
cudaMalloc((void**)&d_C.elements, size);

// Invoke kernel
dim3 dimBlock(BLOCK_SIZE, BLOCK_SIZE);
dim3 dimGrid(B.width / dimBlock.x, A.height / dimBlock.y);
MatMulKernel<<<dimGrid, dimBlock>>>(d_A, d_B, d_C);

// Read C from device memory

cudaMemcpy(C.elements, Cd.elements, size,
cudaMemcpyDeviceToHost);

// Free device memory

cudaFree(d_A.elements);
cudaFree(d_B.elements);
cudaFree(d_C.elements);
}

// Matrix multiplication kernel called by MatMul()

__global__ void MatMulKernel(Matrix A, Matrix B, Matrix C)
{
// Each thread computes one element of C
// by accumulating results into Cvalue
float Cvalue = 0;
int row = blockIdx.y * blockDim.y + threadIdx.y;
int col = blockIdx.x * blockDim.x + threadIdx.x;
for (int e = 0; e < A.width; ++e)
Cvalue += A.elements[row * A.width + e]
* B.elements[e * B.width + col];
C.elements[row * C.width + col] = Cvalue;

CUDA Programming Guide Version 2.3 21

Chapter 3. Programming Interface

B.width-1
0 col

B.height
0
A C

A.height
row

A.width B.width

A.height-1

Figure 3-1.Matrix Multipliation without Shared Memory

The following code sample is an implementation of matrix multiplication that does
take advantage of shared memory. In this implementation, each thread block is
responsible for computing one square sub-matrix Csub of C and each thread within
the block is responsible for computing one element of Csub. As illustrated in Figure
3-2, Csub is equal to the product of two rectangular matrices: the sub-matrix of A of
dimension (A.width, block_size) that has the same line indices as Csub, and the sub-
matrix of B of dimension (block_size, A.width) that has the same column indices as
Csub. In order to fit into the device’s resources, these two rectangular matrices are
divided into as many square matrices of dimension block_size as necessary and Csub is
computed as the sum of the products of these square matrices. Each of these
products is performed by first loading the two corresponding square matrices from
global memory to shared memory with one thread loading one element of each
matrix, and then by having each thread compute one element of the product. Each
thread accumulates the result of each of these products into a register and once
done writes the result to global memory.

22 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

By blocking the computation this way, we take advantage of fast shared memory
and save a lot of global memory bandwidth since A is only read (B.width / block_size)
times from global memory and B is read (A.height / block_size) times.
The Matrix type from the previous code sample is augmented with a stride field, so
that sub-matrices can be efficiently represented with the same type. __device__
functions (see Section B.1.1) are used to get and set elements and build any sub-
matrix from a matrix.
// Matrices are stored in row-major order:
// M(row, col) = *(M.elements + row * M.stride + col)
typedef struct {
int width;
int height;
int stride;
float* elements;
} Matrix;

// Get a matrix element

__device__ float GetElement(const Matrix A, int row, int col)
{
return A.elements[row * A.stride + col];
}

// Set a matrix element

__device__ void SetElement(Matrix A, int row, int col,
float value)
{
A.elements[row * A.stride + col] = value;
}

// Get the BLOCK_SIZExBLOCK_SIZE sub-matrix Asub of A that is

// located col sub-matrices to the right and row sub-matrices down
// from the upper-left corner of A
__device__ Matrix GetSubMatrix(Matrix A, int row, int col)
{
Matrix Asub;
Asub.width = BLOCK_SIZE;
Asub.height = BLOCK_SIZE;
Asub.stride = A.stride;
Asub.elements = &A.elements[A.stride * BLOCK_SIZE * row
+ BLOCK_SIZE * col];
return Asub;
}

// Thread block size

#define BLOCK_SIZE 16

// Forward declaration of the matrix multiplication kernel

__global__ void MatMulKernel(const Matrix, const Matrix, Matrix);

// Matrix multiplication - Host code

// Matrix dimensions are assumed to be multiples of BLOCK_SIZE
void MatMul(const Matrix A, const Matrix B, Matrix C)
{
// Load A and B to device memory
Matrix d_A;

CUDA Programming Guide Version 2.3 23

Chapter 3. Programming Interface

d_A.width = d_A.stride = A.width; d_A.height = A.height;

size_t size = A.width * A.height * sizeof(float);
cudaMalloc((void**)&d_A.elements, size);
cudaMemcpy(d_A.elements, A.elements, size,
cudaMemcpyHostToDevice);
Matrix d_B;
d_B.width = d_B.stride = B.width; d_B.height = B.height;
size = B.width * B.height * sizeof(float);
cudaMalloc((void**)&d_B.elements, size);
cudaMemcpy(d_B.elements, B.elements, size,
cudaMemcpyHostToDevice);

// Allocate C in device memory

Matrix d_C;
d_C.width = d_C.stride = C.width; d_C.height = C.height;
size = C.width * C.height * sizeof(float);
cudaMalloc((void**)&d_C.elements, size);

// Invoke kernel
dim3 dimBlock(BLOCK_SIZE, BLOCK_SIZE);
dim3 dimGrid(B.width / dimBlock.x, A.height / dimBlock.y);
MatMulKernel<<<dimGrid, dimBlock>>>(d_A, d_B, d_C);

// Read C from device memory

cudaMemcpy(C.elements, d_C.elements, size,
cudaMemcpyDeviceToHost);

// Free device memory

cudaFree(d_A.elements);
cudaFree(d_B.elements);
cudaFree(d_C.elements);
}

// Matrix multiplication kernel called by MatMul()

__global__ void MatMulKernel(Matrix A, Matrix B, Matrix C)
{
// Block row and column
int blockRow = blockIdx.y;
int blockCol = blockIdx.x;

// Each thread block computes one sub-matrix Csub of C

Matrix Csub = GetSubMatrix(C, blockRow, blockCol);

// Each thread computes one element of Csub

// by accumulating results into Cvalue
float Cvalue = 0;

// Thread row and column within Csub

int row = threadIdx.y;
int col = threadIdx.x;

// Loop over all the sub-matrices of A and B that are

// required to compute Csub
// Multiply each pair of sub-matrices together
// and accumulate the results
for (int m = 0; m < (A.width / BLOCK_SIZE); ++m) {

24 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

// Get sub-matrix Asub of A

Matrix Asub = GetSubMatrix(A, blockRow, m);

// Get sub-matrix Bsub of B

Matrix Bsub = GetSubMatrix(B, m, blockCol);

// Shared memory used to store Asub and Bsub respectively

__shared__ float As[BLOCK_SIZE][BLOCK_SIZE];
__shared__ float Bs[BLOCK_SIZE][BLOCK_SIZE];

// Load Asub and Bsub from device memory to shared memory

// Each thread loads one element of each sub-matrix
As[row][col] = GetElement(Asub, row, col);
Bs[row][col] = GetElement(Bsub, row, col);

// Synchronize to make sure the sub-matrices are loaded

// before starting the computation
__syncthreads();

// Multiply Asub and Bsub together

for (int e = 0; e < BLOCK_SIZE; ++e)
Cvalue += As[row][e] * Bs[e][col];

// Synchronize to make sure that the preceding

// computation is done before loading two new
// sub-matrices of A and B in the next iteration
__syncthreads();
}

// Write Csub to device memory

// Each thread writes one element
SetElement(Csub, row, col, Cvalue);
}

CUDA Programming Guide Version 2.3 25

Chapter 3. Programming Interface

blockCol

BLOCK_SIZE
B

B.height
BLOCK_SIZE
BLOCK_SIZE-1
A C

0 col

BLOCK_SIZE
Csub
blockRow

A.height
row
BLOCK_SIZE-1

BLOCK_SIZE BLOCK_SIZE BLOCK_SIZE

A.width B.width

Figure 3-2.Matrix Multipliation with Shared Memory

3.2.3 Multiple Devices

A host system can have multiple devices. These devices can be enumerated, their
properties can be queried, and one of them can be selected for kernel executions.
Several host threads can execute device code on the same device, but by design, a
host thread can execute device code on only one device at any given time. As a
consequence, multiple host threads are required to execute device code on multiple
devices. Also, any CUDA resources created through the runtime in one host thread
cannot be used by the runtime from another host thread.
The following code sample enumerates all devices in the system and retrieves their
properties. It also determines the number of CUDA-enabled devices.
int deviceCount;
cudaGetDeviceCount(&deviceCount);
int device;
for (device = 0; device < deviceCount; ++device) {
cudaDeviceProp deviceProp;
cudaGetDeviceProperties(&deviceProp, device);
if (dev == 0) {

26 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

if (deviceProp.major == 9999 && deviceProp.minor == 9999)

printf("There is no device supporting CUDA.\n");
else if (deviceCount == 1)
printf("There is 1 device supporting CUDA\n");
else
printf("There are %d devices supporting CUDA\n",
deviceCount);
}
}
By default, the device associated to the host thread is implicitly selected as device 0
as soon as a non-device management runtime function is called (see Section 3.5 for
exceptions). Any other device can be selected by calling cudaSetDevice() first.
After a device has been selected, either implicitly or explicitly, any subsequent
explicit call to cudaSetDevice() will fail up until cudaThreadExit() is called.
cudaThreadExit() cleans up all runtime-related resources associated with the
calling host thread. Any subsequent API call reinitializes the runtime.

3.2.4 Texture Memory

CUDA supports a subset of the texturing hardware that the GPU uses for graphics
to access texture memory. Reading data from texture memory instead of global
memory can have several performance benefits as described in Section 5.1.2.4.
Texture memory is read from kernels using device functions called texture fetches,
described in Section B.8. The first parameter of a texture fetch specifies an object
called a texture reference.
A texture reference defines which part of texture memory is fetched. As detailed in
Section 3.2.4.3, it must be bound through runtime functions to some region of
memory, called a texture, before it can be used by a kernel. Several distinct texture
references might be bound to the same texture or to textures that overlap in
memory.
A texture reference has several attributes. One of them is its dimensionality that
specifies whether the texture is addressed as a one-dimensional array using one
texture coordinate, a two-dimensional array using two texture coordinates, or a three-
dimensional array using three texture coordinates. Elements of the array are called
texels, short for “texture elements.”
Other attributes define the input and output data types of the texture fetch, as well
as how the input coordinates are interpreted and what processing should be done.
A texture can be any region of linear memory or a CUDA array.
CUDA arrays are opaque memory layouts optimized for texture fetching. They are
one-dimensional, two-dimensional, or three-dimensional and composed of
elements, each of which has 1, 2 or 4 components that may be signed or unsigned
8-, 16- or 32-bit integers, 16-bit floats (currently only supported through the driver
API), or 32-bit floats. CUDA arrays are only readable by kernels through texture
fetching and may only be bound to texture references with the same number of
packed components.

CUDA Programming Guide Version 2.3 27

Chapter 3. Programming Interface

3.2.4.1 Texture Reference Declaration

Some of the attributes of a texture reference are immutable and must be known at
compile time; they are specified when declaring the texture reference. A texture
reference is declared at file scope as a variable of type texture:
texture<Type, Dim, ReadMode> texRef;
where:
‰ Type specifies the type of data that is returned when fetching the texture; Type
is restricted to the basic integer and single-precision floating-point types and any
of the 1-, 2-, and 4-component vector types defined in Section B.3.1;
‰ Dim specifies the dimensionality of the texture reference and is equal to 1, 2, or
3; Dim is an optional argument which defaults to 1;
‰ ReadMode is equal to cudaReadModeNormalizedFloat or
cudaReadModeElementType; if it is cudaReadModeNormalizedFloat
and Type is a 16-bit or 8-bit integer type, the value is actually returned as
floating-point type and the full range of the integer type is mapped to [0.0, 1.0]
for unsigned integer type and [-1.0, 1.0] for signed integer type; for example, an
unsigned 8-bit texture element with the value 0xff reads as 1; if it is
cudaReadModeElementType, no conversion is performed; ReadMode is an
optional argument which defaults to cudaReadModeElementType.

3.2.4.2 Runtime Texture Reference Attributes

The other attributes of a texture reference are mutable and can be changed at
runtime through the host runtime. They specify whether texture coordinates are
normalized or not, the addressing mode, and texture filtering, as detailed below.
By default, textures are referenced using floating-point coordinates in the range
[0, N) where N is the size of the texture in the dimension corresponding to the
coordinate. For example, a texture that is 64×32 in size will be referenced with
coordinates in the range [0, 63] and [0, 31] for the x and y dimensions, respectively.
Normalized texture coordinates cause the coordinates to be specified in the range
[0.0, 1.0) instead of [0, N), so the same 64×32 texture would be addressed by
normalized coordinates in the range [0, 1) in both the x and y dimensions.
Normalized texture coordinates are a natural fit to some applications’ requirements,
if it is preferable for the texture coordinates to be independent of the texture size.
The addressing mode defines what happens when texture coordinates are out of
range. When using unnormalized texture coordinates, texture coordinates outside
the range [0, N) are clamped: Values below 0 are set to 0 and values greater or equal
to N are set to N-1. Clamping is also the default addressing mode when using
normalized texture coordinates: Values below 0.0 or above 1.0 are clamped to the
range [0.0, 1.0). For normalized coordinates, the “wrap” addressing mode also may
be specified. Wrap addressing is usually used when the texture contains a periodic
signal. It uses only the fractional part of the texture coordinate; for example, 1.25 is
treated the same as 0.25 and -1.25 is treated the same as 0.75.
Linear texture filtering may be done only for textures that are configured to return
floating-point data. It performs low-precision interpolation between neighboring
texels. When enabled, the texels surrounding a texture fetch location are read and
the return value of the texture fetch is interpolated based on where the texture
coordinates fell between the texels. Simple linear interpolation is performed for one-

28 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

dimensional textures and bilinear interpolation is performed for two-dimensional

textures.
Appendix E gives more details on texture fetching.

3.2.4.3 Texture Binding

As explained in the reference manual, the runtime API has a low-level C-style
interface and a high-level C++-style interface. The texture type is defined in the
high-level API as a structure publicly derived from the textureReference type
defined in the low-level API as such:
struct textureReference {
int normalized;
enum cudaTextureFilterMode filterMode;
enum cudaTextureAddressMode addressMode[3];
struct cudaChannelFormatDesc channelDesc;
}
‰ normalized specifies whether texture coordinates are normalized or not; if it is
non-zero, all elements in the texture are addressed with texture coordinates in
the range [0,1] rather than in the range [0,width-1], [0,height-1], or
[0,depth-1] where width, height, and depth are the texture sizes;
‰ filterMode specifies the filtering mode, that is how the value returned when
fetching the texture is computed based on the input texture coordinates;
filterMode is equal to cudaFilterModePoint or
cudaFilterModeLinear; if it is cudaFilterModePoint, the returned
value is the texel whose texture coordinates are the closest to the input texture
coordinates; if it is cudaFilterModeLinear, the returned value is the linear
interpolation of the two (for a one-dimensional texture), four (for a
two-dimensional texture), or eight (for a three-dimensional texture) texels whose
texture coordinates are the closest to the input texture coordinates;
cudaFilterModeLinear is only valid for returned values of floating-point
type;
‰ addressMode specifies the addressing mode, that is how out-of-range texture
coordinates are handled; addressMode is an array of size three whose first,
second, and third elements specify the addressing mode for the first, second, and
third texture coordinates, respectively; the addressing mode is equal to either
cudaAddressModeClamp, in which case out-of-range texture coordinates are
clamped to the valid range, or cudaAddressModeWrap, in which case out-of-
range texture coordinates are wrapped to the valid range;
cudaAddressModeWrap is only supported for normalized texture coordinates;
‰ channelDesc describes the format of the value that is returned when fetching
the texture; channelDesc is of the following type:
struct cudaChannelFormatDesc {
int x, y, z, w;
enum cudaChannelFormatKind f;
};
where x, y, z, and w are equal to the number of bits of each component of the
returned value and f is:
¾ cudaChannelFormatKindSigned if these components are of signed
integer type,

CUDA Programming Guide Version 2.3 29

Chapter 3. Programming Interface

¾ cudaChannelFormatKindUnsigned if they are of unsigned integer

type,
¾ cudaChannelFormatKindFloat if they are of floating point type.
normalized, addressMode, and filterMode may be directly modified in host
code.
Before a kernel can use a texture reference to read from texture memory, the texture
reference must be bound to a texture using cudaBindTexture() or
cudaBindTextureToArray(). cudaUnbindTexture() is used to unbind a
texture reference.
The following code samples bind a texture reference to linear memory pointed to by
devPtr:
‰ Using the low-level API:
texture<float, 2, cudaReadModeElementType> texRef;
textureReference* texRefPtr;
cudaGetTextureReference(&texRefPtr, “texRef”);
cudaChannelFormatDesc channelDesc =
cudaCreateChannelDesc<float>();
cudaBindTexture2D(0, texRefPtr, devPtr, &channelDesc,
width, height, pitch);
‰ Using the high-level API:
texture<float, 2, cudaReadModeElementType> texRef;
cudaChannelFormatDesc channelDesc =
cudaCreateChannelDesc<float>();
cudaBindTexture2D(0, texRef, devPtr, &channelDesc,
width, height, pitch);
The following code samples bind a texture reference to a CUDA array cuArray:
‰ Using the low-level API:
texture<float, 2, cudaReadModeElementType> texRef;
textureReference* texRefPtr;
cudaGetTextureReference(&texRefPtr, “texRef”);
cudaChannelFormatDesc channelDesc;
cudaGetChannelDesc(&channelDesc, cuArray);
cudaBindTextureToArray(texRef, cuArray, &channelDesc);
‰ Using the high-level API:
texture<float, 2, cudaReadModeElementType> texRef;
cudaBindTextureToArray(texRef, cuArray);
The format specified when binding a texture to a texture reference must match the
parameters specified when declaring the texture reference; otherwise, the results of
texture fetches are undefined.
The following code sample applies some simple transformation kernel to a
// 2D float texture
texture<float, 2, cudaReadModeElementType> texRef;

// Simple transformation kernel

__global__ void transformKernel(float* output,
int width, int height, float theta)
{
// Calculate normalized texture coordinates
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;

30 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;

float u = x / (float)width;
float v = y / (float)height;

// Transform coordinates
u -= 0.5f;
v -= 0.5f;
float tu = u * cosf(theta) – v * sinf(theta) + 0.5f;
float tv = v * cosf(theta) + u * sinf(theta) + 0.5f;

// Read from texture and write to global memory

output[y * width + x] = tex2D(tex, tu, tv);
}

// Host code
int main()
{
// Allocate CUDA array in device memory
cudaChannelFormatDesc channelDesc =
cudaCreateChannelDesc(32, 0, 0, 0,
cudaChannelFormatKindFloat);
cudaArray* cuArray;
cudaMallocArray(&cuArray, &channelDesc, width, height);

// Copy to device memory some data located at address h_data

// in host memory
cudaMemcpyToArray(cuArray, 0, 0, h_data, size,
cudaMemcpyHostToDevice);

// Set texture parameters

texRef.addressMode[0] = cudaAddressModeWrap;
texRef.addressMode[1] = cudaAddressModeWrap;
texRef.filterMode = cudaFilterModeLinear;
texRef.normalized = true;

// Bind the array to the texture

cudaBindTextureToArray(texRef, cuArray, channelDesc);

// Allocate result of transformation in device memory

float* output;
cudaMalloc((void**)&output, width * height * sizeof(float));

// Invoke kernel
dim3 dimBlock(16, 16);
dim3 dimGrid((width + dimBlock.x – 1) / dimBlock.x,
(height + dimBlock.y – 1) / dimBlock.y);
transformKernel<<<dimGrid, dimBlock>>>(output, width, height,
angle);

// Free device memory

cudaFreeArray(cuArray);
cudaFree(output);
}

CUDA Programming Guide Version 2.3 31

Chapter 3. Programming Interface

3.2.5 Page-Locked Host Memory

The runtime also provides functions to allocate and free page-locked (also known as
pinned) host memory – as opposed to regular pageable host memory allocated by
malloc(): cudaHostAlloc() and cudaFreeHost().
Using page-locked host memory has several benefits:
‰ Bandwidth between host memory and device memory is higher if host memory
is allocated as page-locked and even higher if in addition it is allocated as write-
combining as described in Section 3.2.5.2;
‰ Copies between page-locked host memory and device memory can be performed
concurrently with kernel execution for some devices as mentioned in
Section 3.2.6;
‰ On some devices, page-locked host memory can be mapped into the device’s
address space, eliminating the need to copy it to or from device memory as
detailed in Section 3.2.5.3.
Page-locked host memory is a scarce resource however, so allocations in page-
locked memory will start failing long before allocations in pageable memory. In
addition, by reducing the amount of physical memory available to the operating
system for paging, allocating too much page-locked memory reduces overall system
performance.
The simple zero-copy SDK sample comes with a detailed document on the page-
locked memory APIs.

3.2.5.1 Portable Memory

A block of page-locked memory can be used by any host threads, but by default, the
benefits of using page-locked memory described above are only available for the
thread that allocates it. To make these advantages available to all threads, it needs to
be allocated by passing flag cudaHostAllocPortable to cudaHostAlloc().

3.2.5.2 Write-Combining Memory

By default page-locked host memory is allocated as cacheable. It can optionally be
allocated as write-combining instead by passing flag
cudaHostAllocWriteCombined to cudaHostAlloc(). Write-combining
memory frees up L1 and L2 cache resources, making more cache available to the
rest of the application. In addition, write-combining memory is not snooped during
transfers across the PCI Express bus, which can improve transfer performance by
up to 40%.
Reading from write-combining memory from the host is prohibitively slow, so
write-combining memory should in general be used for memory that the host only
writes to.

3.2.5.3 Mapped Memory

On some devices, a block of page-locked host memory can also be mapped into the
device’s address space by passing flag cudaHostAllocMapped to
cudaHostAlloc(). Such a block has therefore two addresses: one in host
memory and one in device memory. The host memory pointer is returned by
cudaHostAlloc() and the device memory pointer can be retrieved using

32 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

cudaHostGetDevicePointer()and used to access the block from within a

kernel.
Accessing host memory directly from within a kernel has several advantages:
‰ There is no need to allocate a block in device memory and copy data between
this block and the block in host memory; data transfers are implicitly performed
as needed by the kernel;
‰ There is no need to use streams (see Section 3.2.6.1) to overlap data transfers
with kernel execution; the kernel-originated data transfers automatically overlap
with kernel execution.
Since mapped page-locked memory is shared between host and device however, the
application must synchronize memory accesses using streams or events (see
Section 3.2.6) to avoid any potential read-after-write, write-after-read, or write-after-
write hazards.
A block of page-locked host memory can be allocated as both mapped and portable
(see Section 3.2.5.1), in which case each host thread that needs to map the block to
its device address space must call cudaHostGetDevicePointer() to retrieve a
device pointer, as device pointers will generally differ from one host thread to the
other.
To be able to retrieve the device pointer to any mapped page-locked memory within
a given host thread, page-locked memory mapping must be enabled by calling
cudaSetDeviceFlags() with the cudaDeviceMapHost flag before any other
CUDA calls is performed by the thread. Otherwise,
cudaHostGetDevicePointer() will return an error.
cudaHostGetDevicePointer() also returns an error if the device does not
support mapped page-locked host memory.
Applications may query whether a device supports mapped page-locked host
memory or not by calling cudaGetDeviceProperties() and checking the
canMapHostMemory property.

3.2.6 Asynchronous Concurrent Execution

In order to facilitate concurrent execution between host and device, some functions
are asynchronous: Control is returned to the host thread before the device has
completed the requested task. These are:
‰ Kernel launches;
‰ The functions that perform memory copies and are suffixed with Async;
‰ The functions that perform device ↔ device memory copies;
‰ The functions that set memory.
Some devices can also perform copies between page-locked host memory and
device memory concurrently with kernel execution. Applications may query this
capability by calling cudaGetDeviceProperties() and checking the
deviceOverlap property. This capability is currently supported only for memory
copies that do not involve CUDA arrays or 2D arrays allocated through
cudaMallocPitch() (see Section 3.2.1).

CUDA Programming Guide Version 2.3 33

Chapter 3. Programming Interface

3.2.6.1 Stream
Applications manage concurrency through streams. A stream is a sequence of
commands that execute in order. Different streams, on the other hand, may execute
their commands out of order with respect to one another or concurrently.
A stream is defined by creating a stream object and specifying it as the stream
parameter to a sequence of kernel launches and host ↔ device memory copies. The
following code sample creates two streams and allocates an array hostPtr of
float in page-locked memory.
cudaStream_t stream[2];
for (int i = 0; i < 2; ++i)
cudaStreamCreate(&stream[i]);
float* hostPtr;
cudaMallocHost((void**)&hostPtr, 2 * size);
Each of these streams is defined by the following code sample as a sequence of one
memory copy from host to device, one kernel launch, and one memory copy from
device to host:
for (int i = 0; i < 2; ++i)
cudaMemcpyAsync(inputDevPtr + i * size, hostPtr + i * size,
size, cudaMemcpyHostToDevice, stream[i]);
for (int i = 0; i < 2; ++i)
myKernel<<<100, 512, 0, stream[i]>>>
(outputDevPtr + i * size, inputDevPtr + i * size, size);
for (int i = 0; i < 2; ++i)
cudaMemcpyAsync(hostPtr + i * size, outputDevPtr + i * size,
size, cudaMemcpyDeviceToHost, stream[i]);
cudaThreadSynchronize();
Each stream copies its portion of input array hostPtr to array inputDevPtr in
device memory, processes inputDevPtr on the device by calling myKernel(), and
copies the result outputDevPtr back to the same portion of hostPtr. Processing
hostPtr using two streams allows for the memory copies of one stream to overlap
with the kernel execution of the other stream. hostPtr must point to page-locked
host memory for any overlap to occur. cudaThreadSynchronize() is called in the
end to make sure all streams are finished before proceeding further.
cudaStreamSynchronize() can be used to synchronize the host with a specific
stream, allowing other streams to continue executing on the device. Streams are
released by calling cudaStreamDestroy().
for (int i = 0; i < 2; ++i)
cudaStreamDestroy(&stream[i]);
Any kernel launch, memory set, or memory copy function without a stream
parameter or with a zero stream parameter begins only after all preceding
commands are done, including commands that are part of streams, and no
subsequent command may begin until it is done.
cudaStreamQuery() provides applications with a way to know if all preceding
commands in a stream have completed. cudaStreamSynchronize() provides a
way to explicitly force the runtime to wait until all preceding commands in a stream
have completed.
Similarly, with cudaThreadSynchronize() forces the runtime to wait until all
preceding device tasks in all streams have completed. To avoid unnecessary
slowdowns, these functions are best used for timing purposes or to isolate a launch

34 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

or memory copy that is failing. cudaStreamDestroy() waits for all preceding

tasks in the given stream to complete before destroying the stream and returning
control to the host thread.
Two commands from different streams cannot run concurrently if either a page-
locked host memory allocation, a device memory allocation, a device memory set, a
device ↔ device memory copy, or any CUDA command to stream 0 is called in-
between them by the host thread.
Programmers can globally disable asynchronous execution for all CUDA
applications running on a system by setting the CUDA_LAUNCH_BLOCKING
environment variable to 1. This feature is provided for debugging purposes only and
should never be used as a way to make production software run reliably.

3.2.6.2 Event
The runtime also provides a way to closely monitor the device’s progress, as well as
perform accurate timing, by letting the application asynchronously record events at
any point in the program and query when these events are actually recorded. An
event is recorded when all tasks – or optionally, all commands in a given stream –
preceding the event have completed. Events in stream zero are recorded after all
preceding tasks/commands from all streams are completed by the device.
The following code sample creates two events:
cudaEvent_t start, stop;
cudaEventCreate(&start);
cudaEventCreate(&stop);
These events can be used to time the code sample of the previous section the
following way:
cudaEventRecord(start, 0);
for (int i = 0; i < 2; ++i)
cudaMemcpyAsync(inputDev + i * size, inputHost + i * size,
size, cudaMemcpyHostToDevice, stream[i]);
for (int i = 0; i < 2; ++i)
myKernel<<<100, 512, 0, stream[i]>>>
(outputDev + i * size, inputDev + i * size, size);
for (int i = 0; i < 2; ++i)
cudaMemcpyAsync(outputHost + i * size, outputDev + i * size,
size, cudaMemcpyDeviceToHost, stream[i]);
cudaEventRecord(stop, 0);
cudaEventSynchronize(stop);
float elapsedTime;
cudaEventElapsedTime(&elapsedTime, start, stop);
They are destroyed this way:
cudaEventDestroy(start);
cudaEventDestroy(stop);

3.2.6.3 Synchronous Calls

When a synchronous function is called, control is not returned to the host thread
before the device has completed the requested task. Whether the host thread will
then yield, block, or spin can be specified by calling cudaSetDeviceFlags()with
some specific flags (see reference manual for details) before any other CUDA calls is
performed by the host thread.

CUDA Programming Guide Version 2.3 35

Chapter 3. Programming Interface

3.2.7 OpenGL Interoperability

OpenGL buffer objects may be mapped into the address space of CUDA, either to
enable CUDA to read data written by OpenGL or to enable CUDA to write data
for consumption by OpenGL.
Interoperability with OpenGL requires that the CUDA device be specified by
cudaGLSetGLDevice() before any other runtime calls.
A buffer object must be registered to CUDA before it can be mapped. This is done
with cudaGLRegisterBufferObject():
GLuint bufferObj;
cudaGLRegisterBufferObject(bufferObj);
Once it is registered, a buffer object can be read from or written to by kernels using
the device memory address returned by cudaGLMapBufferObject():
GLuint bufferObj;
float* devPtr;
cudaGLMapBufferObject((void**)&devPtr, bufferObj);
Unmapping is done with cudaGLUnmapBufferObject() and unregistering with
cudaGLUnregisterBufferObject().
The following code sample uses a kernel to dynamically modify a 2D
width x height grid of vertices stored in a vertex buffer object:
GLuint positionsVBO;
int main()
{
// Explicitly set device
cudaGLSetGLDevice(0);

// Initialize OpenGL and GLUT

...
glutDisplayFunc(display);

// Create buffer object and register it with CUDA

glGenBuffers(1, positionsVBO);
glBindBuffer(GL_ARRAY_BUFFER, &vbo);
unsigned int size = width * height * 4 * sizeof(float);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, 0);
cudaGLRegisterBufferObject(positionsVBO);

// Launch rendering loop

glutMainLoop();
}

void display()
{
// Map buffer object for writing from CUDA
float4* positions;
cudaGLMapBufferObject((void**)&positions, positionsVBO);

// Execute kernel
dim3 dimBlock(16, 16, 1);
dim3 dimGrid(width / dimBlock.x, height / dimBlock.y, 1);
createVertices<<<dimGrid, dimBlock>>>(positions, time,

36 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

width, height);

// Unmap buffer object

cudaGLUnmapBufferObject(positionsVBO);

// Render from buffer object

glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
glBindBuffer(GL_ARRAY_BUFFER, positionsVBO);
glVertexPointer(4, GL_FLOAT, 0, 0);
glEnableClientState(GL_VERTEX_ARRAY);
glDrawArrays(GL_POINTS, 0, width * height);
glDisableClientState(GL_VERTEX_ARRAY);

// Swap buffers
glutSwapBuffers();
glutPostRedisplay();
}

void deleteVBO()
{
cudaGLUnregisterBufferObject(positionsVBO);
glDeleteBuffers(1, &positionsVBO);
}

__global__ void createVertices(float4* positions, float time,

unsigned int width, unsigned int height)
{
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;
unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;

// Calculate uv coordinates
float u = x / (float)width;
float v = y / (float)height;
u = u * 2.0f - 1.0f;
v = v * 2.0f - 1.0f;

// calculate simple sine wave pattern

float freq = 4.0f;
float w = sinf(u * freq + time)
* cosf(v * freq + time) * 0.5f;

// Write positions
positions[y * width + x] = make_float4(u, w, v, 1.0f);
}
On Windows and for Quadro GPUs, cudaWGLGetDevice() can be used to
retrieve the CUDA device associated to the handle returned by
WGL_NV_gpu_affinity(). Quadro GPUs offer higher performance OpenGL
interoperability than GeForce and Tesla GPUs in a multi-GPU configuration where
OpenGL rendering is performed on the Quadro GPU and CUDA computations are
performed on other GPUs in the system.

CUDA Programming Guide Version 2.3 37

Chapter 3. Programming Interface

3.2.8 Direct3D Interoperability

Direct3D resources may be mapped into the address space of CUDA, either to
enable CUDA to read data written by Direct3D or to enable CUDA to write data
for consumption by Direct3D.
Direct3D interoperability is supported for Direct3D 9.0 and Direct3D 10.0.
There are restrictions on which resources can be mapped as detailed in the reference
manual for cudaD3D9RegisterResource() (resp.
cudaD3D10RegisterResource()).
A CUDA context may interoperate with only one Direct3D device at a time and the
CUDA context and Direct3D device must be created on the same GPU. Moreover,
the Direct3D device must be created with the
D3DCREATE_HARDWARE_VERTEXPROCESSING flag.
Interoperability with Direct3D requires that the Direct3D device be specified by
cudaD3D9SetDirect3DDevice() (resp.
cudaD3D10SetDirect3DDevice()) before any other runtime calls.
Direct3D resources can then be registered to CUDA using
cudaD3D9RegisterResource() (resp. cudaD3D10RegisterResource()):
IDirect3DVertexBuffer9* buffer;
cudaD3D9RegisterResource(buffer, cudaD3D9RegisterFlagsNone);
IDirect3DSurface9* surface;
cudaD3D9RegisterResource(surface, cudaD3D9RegisterFlagsNone);

ID3D10Buffer* buffer;
cudaD3D10RegisterResource(buffer, cudaD3D10RegisterFlagsNone);
ID3D10Texture2D* tex2D;
cudaD3D10RegisterResource(tex2D, cudaD3D10RegisterFlagsNone);
cudaD3D9RegisterResource() (resp. cudaD3D10RegisterResource()) is
potentially high-overhead and typically called only once per resource. Unregistering
is done with cudaD3D9UnregisterVertexBuffer() (resp.
cudaD3D10UnregisterVertexBuffer()).
Once a resource is registered to CUDA, it can be mapped and unmapped as many
times as necessary using cudaD3D9MapResources() (resp.
cudaD3D10MapResources()) and cudaD3D9UnmapResources() (resp.
cudaD3D10UnmapResources()).
A mapped resource can be read from or written to by kernels using the device
memory address returned by cudaD3D9ResourceGetMappedPointer() (resp.
cudaD3D10ResourceGetMappedPointer()) and the size and pitch
information returned by cudaD3D9ResourceGetMappedSize() (resp.
cudaD3D10ResourceGetMappedSize()),
cudaD3D9ResourceGetMappedPitch() (resp.
cudaD3D10ResourceGetMappedPitch()), and
cudaD3D9ResourceGetMappedPitchSlice() (resp.
cudaD3D10ResourceGetMappedPitchSlice()).
When applicable, a CUDA array can also be obtained from a mapped resource using
cudaD3D9ResourceGetMappedArray() (resp.
cudaD3D10ResourceGetMappedArray()).

38 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

Accessing a resource through Direct3D while it is mapped produces undefined

results.
The following code sample uses a kernel to dynamically modify a 2D
width x height grid of vertices stored in a vertex buffer object:
IDirect3D9* D3D;
IDirect3DDevice9* device;
struct CUSTOMVERTEX {
FLOAT x, y, z;
DWORD color;
};
IDirect3DVertexBuffer9* positionsVB;

int main()
{
// Initialize Direct3D
D3D = Direct3DCreate9(D3D_SDK_VERSION);

// Get a CUDA-enabled adapter

unsigned int adapter = 0;
for (; adapter < g_pD3D->GetAdapterCount(); adapter++) {
D3DADAPTER_IDENTIFIER9 adapterId;
g_pD3D->GetAdapterIdentifier(adapter, 0, &adapterId);
int dev;
cudaD3D9GetDevice(&dev, adapterId.DeviceName);
if (cudaSuccess == cudaGetLastError())
break;
}

// Create device
...
D3D->CreateDevice(adapter, D3DDEVTYPE_HAL, hWnd,
D3DCREATE_HARDWARE_VERTEXPROCESSING,
&params, &device);

// Register device with CUDA

cudaD3D9SetDirect3DDevice(device);

// Create vertex buffer and register it with CUDA

unsigned int size = width * height * sizeof(CUSTOMVERTEX);
device->CreateVertexBuffer(size, 0, D3DFVF_CUSTOMVERTEX,
D3DPOOL_DEFAULT, &positionsVB, 0);
cudaD3D9RegisterResource(positionsVB,
cudaD3D9RegisterFlagsNone);
cudaD3D9ResourceSetMapFlags(positionsVB,
cudaD3D9MapFlagsWriteDiscard);

// Launch rendering loop

while (...) {
...
Render();
...
}
}

void Render()

CUDA Programming Guide Version 2.3 39

Chapter 3. Programming Interface

{
// Map vertex buffer for writing from CUDA
float4* positions;
cudaD3D9MapResources(1, (IDirect3DResource9**)&positionsVB);
cudaD3D9ResourceGetMappedPointer((void**)&positions,
positionsVB, 0, 0);

// Execute kernel
dim3 dimBlock(16, 16, 1);
dim3 dimGrid(width / dimBlock.x, height / dimBlock.y, 1);
createVertices<<<dimGrid, dimBlock>>>(positions, time,
width, height);

// Unmap vertex buffer

cudaD3D9UnmapResources(1, (IDirect3DResource9**)&positionsVB);

// Draw and present

...
}

void releaseVB()
{
cudaD3D9UnregisterResource(positionsVB);
positionsVB->Release();
}

__global__ void createVertices(float4* positions, float time,

unsigned int width, unsigned int height)
{
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;
unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;

// Calculate uv coordinates
float u = x / (float)width;
float v = y / (float)height;
u = u * 2.0f - 1.0f;
v = v * 2.0f - 1.0f;

// Calculate simple sine wave pattern

float freq = 4.0f;
float w = sinf(u * freq + time)
* cosf(v * freq + time) * 0.5f;

// Write positions
positions[y * width + x] =
make_float4(u, w, v, __int_as_float(0xff00ff00));
}

ID3D10Device* device;
struct CUSTOMVERTEX {
FLOAT x, y, z;
DWORD color;
};
ID3D10Buffer* positionsVB;

int main()

40 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

{
// Get a CUDA-enabled adapter
IDXGIFactory* factory;
CreateDXGIFactory(__uuidof(IDXGIFactory), (void**)&factory);
IDXGIAdapter* adapter = 0;
for (unsigned int i = 0; !adapter; ++i) {
if (FAILED(factory->EnumAdapters(i, &adapter))
break;
int dev;
cudaD3D10GetDevice(&dev, adapter);
if (cudaSuccess == cudaGetLastError())
break;
adapter->Release();
}
factory->Release();

// Create swap chain and device

...
D3D10CreateDeviceAndSwapChain(adapter,
D3D10_DRIVER_TYPE_HARDWARE, 0,
D3D10_CREATE_DEVICE_DEBUG,
D3D10_SDK_VERSION,
&swapChainDesc &swapChain,
&device);
adapter->Release();

// Register device with CUDA

cudaD3D10SetDirect3DDevice(device);

// Create vertex buffer and register it with CUDA

unsigned int size = width * height * sizeof(CUSTOMVERTEX);
D3D10_BUFFER_DESC bufferDesc;
bufferDesc.Usage = D3D10_USAGE_DEFAULT;
bufferDesc.ByteWidth = size;
bufferDesc.BindFlags = D3D10_BIND_VERTEX_BUFFER;
bufferDesc.CPUAccessFlags = 0;
bufferDesc.MiscFlags = 0;
device->CreateBuffer(&bufferDesc, 0, &positionsVB);
cudaD3D10RegisterResource(positionsVB,
cudaD3D10RegisterFlagsNone);
cudaD3D10ResourceSetMapFlags(positionsVB,
cudaD3D10MapFlagsWriteDiscard);

// Launch rendering loop

while (...) {
...
Render();
...
}
}

void Render()
{
// Map vertex buffer for writing from CUDA
float4* positions;
cudaD3D10MapResources(1, (ID3D10Resource**)&positionsVB);
cudaD3D10ResourceGetMappedPointer((void**)&positions,

CUDA Programming Guide Version 2.3 41

Chapter 3. Programming Interface

positionsVB, 0);

// Execute kernel
dim3 dimBlock(16, 16, 1);
dim3 dimGrid(width / dimBlock.x, height / dimBlock.y, 1);
createVertices<<<dimGrid, dimBlock>>>(positions, time,
width, height);

// Unmap vertex buffer

cudaD3D10UnmapResources(1, (ID3D10Resource**)&positionsVB);

// Draw and present

...
}

void releaseVB()
{
cudaD3D10UnregisterResource(positionsVB);
positionsVB->Release();
}

__global__ void createVertices(float4* positions, float time,

unsigned int width, unsigned int height)
{
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;
unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;

// Calculate uv coordinates
float u = x / (float)width;
float v = y / (float)height;
u = u * 2.0f - 1.0f;
v = v * 2.0f - 1.0f;

// Calculate simple sine wave pattern

float freq = 4.0f;
float w = sinf(u * freq + time)
* cosf(v * freq + time) * 0.5f;

// Write positions
positions[y * width + x] =
make_float4(u, w, v, __int_as_float(0xff00ff00));
}
In the following code sample, each thread accesses one pixel of a 2D surface of size
(width, height) and pixel format float4:
// host code
void* devPtr;
cudaD3D9ResourceGetMappedPointer(&devPtr, surface, 0, 0);
size_t pitch;
cudaD3D9ResourceGetMappedPitch(&pitch, 0, surface, 0, 0);
dim3 Db = dim3(16, 16);
dim3 Dg = dim3((width+Db.x–1)/Db.x, (height+Db.y–1)/Db.y);
myKernel<<<Dg, Db>>>((unsigned char*)devPtr,
width, height, pitch);

// device code
__global__ void myKernel(unsigned char* surface,

42 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

int width, int height, size_t pitch)

{
int x = blockIdx.x * blockDim.x + threadIdx.x;
int y = blockIdx.y * blockDim.y + threadIdx.y;
if (x >= width || y >= height) return;
float* pixel = (float*)(surface + y * pitch) + 4 * x;
}

// host code
void* devPtr;
cudaD3D10ResourceGetMappedPointer(&devPtr, surface, 0);
size_t pitch;
cudaD3D10ResourceGetMappedPitch(&pitch, 0, surface, 0);
dim3 Db = dim3(16, 16);
dim3 Dg = dim3((width+Db.x–1)/Db.x, (height+Db.y–1)/Db.y);
myKernel<<<Dg, Db>>>((unsigned char*)devPtr,
width, height, pitch);

// device code
__global__ void myKernel(unsigned char* surface,
int width, int height, size_t pitch)
{
int x = blockIdx.x * blockDim.x + threadIdx.x;
int y = blockIdx.y * blockDim.y + threadIdx.y;
if (x >= width || y >= height) return;
float* pixel = (float*)(surface + y * pitch) + 4 * x;
}

3.2.9 Error Handling

All runtime functions return an error code, but for an asynchronous function (see
Section 3.2.6), this error code cannot possibly report any of the asynchronous errors
that could occur on the device since the function returns before the device has
completed the task; the error code only reports errors that occur on the host prior
to executing the task, typically related to parameter validation; if an asynchronous
error occurs, it will be reported by some subsequent unrelated runtime function call.
The only way to check for asynchronous errors just after some asynchronous
function call is therefore to synchronize just after the call by calling
cudaThreadSynchronize() (or by using any other synchronization
mechanisms described in Section 3.2.6) and checking the error code returned by
cudaThreadSynchronize().
The runtime maintains an error variable for each host thread that is initialized to
cudaSuccess and is overwritten by the error code every time an error occurs (be
it a parameter validation error or an asynchronous error). cudaGetLastError()
returns this variable and resets it to cudaSuccess.
Kernel launches do not return any error code, so cudaGetLastError() must be
called just after the kernel launch to retrieve any pre-launch errors. To ensure that
any error returned by cudaGetLastError() does not originate from calls prior
to the kernel launch, one has to make sure that the runtime error variable is set to
cudaSuccess just before the kernel launch, for example, by calling
cudaGetLastError() just before the kernel launch. Kernel launches are

CUDA Programming Guide Version 2.3 43

Chapter 3. Programming Interface

asynchronous, so to check for asynchronous errors, the application must

synchronize in-between the kernel launch and the call to cudaGetLastError().

3.2.10 Debugging using the Device Emulation Mode

CUDA-GDB can be used to debug devices of compute capability greater than 1.0.
(see the CUDA-GDB user manual for supported platforms). The compiler and
runtime also supports an emulation mode for the purpose of debugging that can be
used even in the absence of any CUDA-enabled device. When compiling an
application in this mode (using the -deviceemu option), the device code is
compiled for and runs on the host, allowing the programmer to use the host’s native
debugging support to debug the application as if it were a host application. The
preprocessor macro __DEVICE_EMULATION__ is defined in this mode. All code
for an application, including any libraries used, must be compiled consistently either
for device emulation or for device execution. Linking code compiled for device
emulation with code compiled for device execution causes the following runtime
error to be returned upon initialization: cudaErrorMixedDeviceExecution.
When running an application in device emulation mode, the programming model is
emulated by the runtime. For each thread in a thread block, the runtime creates a
thread on the host. The programmer needs to make sure that:
‰ The host is able to run up to the maximum number of threads per block, plus
one for the master thread.
‰ Enough memory is available to run all threads, knowing that each thread gets
256 KB of stack.
Many features provided through the device emulation mode make it a very effective
debugging tool:
‰ By using the host’s native debugging support programmers can use all features
that the debugger supports, like setting breakpoints and inspecting data.
‰ Since device code is compiled to run on the host, the code can be augmented
with code that cannot run on the device, like input and output operations to files
or to the screen (printf(), etc.).
‰ Since all data resides on the host, any device- or host-specific data can be read
from either device or host code; similarly, any device or host function can be
called from either device or host code.
‰ In case of incorrect usage of the synchronization intrinsic function, the runtime
detects dead lock situations.
Programmers must keep in mind that device emulation mode is emulating the
device, not simulating it. Therefore, device emulation mode is very useful in finding
algorithmic errors, but certain errors are hard to find:
‰ Race conditions are less likely to manifest themselves in device-emulation mode,
since the number of threads executing simultaneously is much smaller than on
an actual device.
‰ When dereferencing a pointer to global memory on the host or a pointer to host
memory on the device, device execution almost certainly fails in some undefined
way, whereas device emulation can produce correct results.
‰ Most of the time the same floating-point computation will not produce exactly
the same result when performed on the device as when performed on the host in

44 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

device emulation mode. This is expected since in general, all you need to get
different results for the same floating-point computation are slightly different
compiler options, let alone different compilers, different instruction sets, or
different architectures.
In particular, some host platforms store intermediate results of single-precision
floating-point calculations in extended precision registers, potentially resulting in
significant differences in accuracy when running in device emulation mode.
When this occurs, programmers can try any of the following methods, none of
which is guaranteed to work:
¾ Declare some floating-point variables as volatile to force single-precision
storage;
¾ Use the –ffloat-store compiler option of gcc,
¾ Use the /Op or /fp compiler options of the Visual C++ compiler,
¾ Use _FPU_GETCW() and _FPU_SETCW() on Linux or _controlfp()
on Windows to force single-precision floating-point computation for a
portion of the code by surrounding it with
unsigned int originalCW;
_FPU_GETCW(originalCW);
unsigned int cw = (originalCW & ~0x300) | 0x000;
_FPU_SETCW(cw);

or
unsigned int originalCW = _controlfp(0, 0);
_controlfp(_PC_24, _MCW_PC);

at the beginning, to store the current value of the control word and change
it to force the mantissa to be stored in 24 bits using, and with
_FPU_SETCW(originalCW);

or
_controlfp(originalCW, 0xfffff);
at the end, to restore the original control word.
Also, for single-precision floating-point numbers, unlike compute devices (see
Appendix A), host platforms usually support denormalized numbers. This can
lead to dramatically different results between device emulation and device
execution modes since some computation might produce a finite result in one
case and an infinite result in the other.
‰ The warp size is equal to 1 in device emulation mode (see Section 4.1 for the
definition of a warp). Therefore, the warp vote functions (described in
Section B.11) produce different results than in device execution mode.

3.3 Driver API

The driver API is a handle-based, imperative API: Most objects are referenced by
opaque handles that may be specified to functions to manipulate the objects.

CUDA Programming Guide Version 2.3 45

Chapter 3. Programming Interface

The objects available in the driver API are summarized in Table 3-1.

Table 3-1. Objects Available in the CUDA Driver API

Object Handle Description
Device CUdevice CUDA-enabled device
Context CUcontext Roughly equivalent to a CPU process
Module CUmodule Roughly equivalent to a dynamic library
Function CUfunction Kernel
Heap memory CUdeviceptr Pointer to device memory
CUDA array CUarray Opaque container for one-dimensional or two-dimensional
data on the device, readable via texture references
Texture reference CUtexref Object that describes how to interpret texture memory data

The driver API is implemented in the nvcuda dynamic library and all its entry
points are prefixed with cu.
The driver API must be initialized with cuInit() before any function from the
driver API is called. A CUDA context must then be created that is attached to a
specific device and made current to the calling host thread as detailed in
Section 3.3.1.
Within a CUDA context, kernels are explicitly loaded as PTX or binary objects by
the host code as described in Section 3.3.2. Kernels written in C must therefore be
compiled separately into PTX or binary objects. Kernels are launched using API
entry points as described in Section 3.3.3.
Any application that wants to run on future device architectures must load PTX, not
binary code. This is because binary code is architecture-specific and therefore
incompatible with future architectures, whereas PTX code is compiled to binary
code at load time by the driver.
Here is the host code of the sample from Section 2.1 written using the driver API:
int main()
{
// Initialize
if (cuInit(0) != CUDA_SUCCESS)
exit (0);

// Get number of devices supporting CUDA

int deviceCount = 0;
cuDeviceGetCount(&deviceCount);
if (deviceCount == 0) {
printf("There is no device supporting CUDA.\n");
exit (0);
}

// Get handle for device 0

CUdevice cuDevice = 0;
cuDeviceGet(&cuDevice, 0);

// Create context
CUcontext cuContext;
cuCtxCreate(&cuContext, 0, cuDevice);

46 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

// Create module from binary file

CUmodule cuModule;
cuModuleLoad(&cuModule, “VecAdd.ptx”);

// Get function handle from module

CUfunction vecAdd;
cuModuleGetFunction(&vecAdd, cuModule, "VecAdd");

// Invoke kernel
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr;
ptr = (void*)(size_t)A;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)B;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)C;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
cuParamSetSize(vecAdd, offset);
int threadsPerBlock = 256;
int blocksPerGrid =
(N + threadsPerBlock – 1) / threadsPerBlock;
cuFuncSetBlockShape(vecAdd, threadsPerBlock, 1, 1);
cuLaunchGrid(vecAdd, blocksPerGrid, 1);
}

3.3.1 Context
A CUDA context is analogous to a CPU process. All resources and actions
performed within the driver API are encapsulated inside a CUDA context, and the
system automatically cleans up these resources when the context is destroyed.
Besides objects such as modules and texture references, each context has its own
distinct 32-bit address space. As a result, CUdeviceptr values from different
contexts reference different memory locations.
A host thread may have only one device context current at a time. When a context is
created with cuCtxCreate(), it is made current to the calling host thread. CUDA
functions that operate in a context (most functions that do not involve device
enumeration or context management) will return
CUDA_ERROR_INVALID_CONTEXT if a valid context is not current to the thread.
Each host thread has a stack of current contexts. cuCtxCreate() pushes the new
context onto the top of the stack. cuCtxPopCurrent() may be called to detach
the context from the host thread. The context is then "floating" and may be pushed
as the current context for any host thread. cuCtxPopCurrent() also restores the
previous current context, if any.

CUDA Programming Guide Version 2.3 47

Chapter 3. Programming Interface

A usage count is also maintained for each context. cuCtxCreate() creates a

context with a usage count of 1. cuCtxAttach() increments the usage count and
cuCtxDetach() decrements it. A context is destroyed when the usage count goes
to 0 when calling cuCtxDetach() or cuCtxDestroy().
Usage count facilitates interoperability between third party authored code operating
in the same context. For example, if three libraries are loaded to use the same
context, each library would call cuCtxAttach() to increment the usage count and
cuCtxDetach() to decrement the usage count when the library is done using the
context. For most libraries, it is expected that the application will have created a
context before loading or initializing the library; that way, the application can create
the context using its own heuristics, and the library simply operates on the context
handed to it. Libraries that wish to create their own contexts – unbeknownst to their
API clients who may or may not have created contexts of their own – would use
cuCtxPushCurrent() and cuCtxPopCurrent() as illustrated in Figure 3-3.

Library Initialization Call

Initialize
cuCtxCreate() context cuCtxPopCurrent()

Library Call

Use
cuCtxPushCurrent() context cuCtxPopCurrent()

Figure 3-3.Library Context Management

3.3.2 Module
Modules are dynamically loadable packages of device code and data, akin to DLLs in
Windows, that are output by nvcc (see Section 3.1). The names for all symbols,
including functions, global variables, and texture references, are maintained at
module scope so that modules written by independent third parties may interoperate
in the same CUDA context.
This code sample loads a module and retrieves a handle to some kernel:
CUmodule cuModule;
cuModuleLoad(&cuModule, “myModule.ptx”);
CUfunction myKernel;
cuModuleGetFunction(&myKernel, cuModule, “myKernel”);
This code sample compiles and loads a new module from PTX code and parses
compilation errors:
#define ERROR_BUFFER_SIZE 100
CUmodule cuModule;
CUptxas_option options[3];
void* values[3];

48 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

char* PTXCode = “some PTX code”;

options[0] = CU_ASM_ERROR_LOG_BUFFER;
values[0] = (void*)malloc(ERROR_BUFFER_SIZE);
options[1] = CU_ASM_ERROR_LOG_BUFFER_SIZE_BYTES;
values[1] = (void*)ERROR_BUFFER_SIZE;
options[2] = CU_ASM_TARGET_FROM_CUCONTEXT;
values[2] = 0;
cuModuleLoadDataEx(&cuModule, PTXCode, 3, options, values);
for (int i = 0; i < values[1]; ++i) {
// Parse error string here
}

3.3.3 Kernel Execution

cuFuncSetBlockShape() sets the number of threads per block for a given
function, and how their threadIDs are assigned.
cuFuncSetSharedSize() sets the size of shared memory for the function.
The cuParam*() family of functions is used to specify the parameters that will be
provided to the kernel the next time cuLaunchGrid() or cuLaunch() is
invoked to launch the kernel.
The second argument of each of the cuParam*() functions specifies the offset of
the parameter in the parameter stack. This offset must match the alignment
requirement for the parameter type in device code. Alignment requirements in
device code for the built-in vector types are listed in Table B-1. For all other basic
types, the alignment requirement in device code matches the alignment requirement
in host code and can therefore be obtained using __alignof(). The only
exception is when the host compiler aligns double and long long (and long on
a 64-bit system) on a one-word boundary instead of a two-word boundary (for
example, using gcc’s compilation flag -mno-align-double) since in device code
these types are always aligned on a two-word boundary. Also, CUdeviceptr is an
integer, but represents a pointer, so its alignment requirement is
__alignof(void*). The following code sample uses a macro to adjust the offset
of each parameter to meet its alignment requirement.
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;

int i;
ALIGN_UP(offset, __alignof(i));
cuParamSeti(cuFunction, offset, i);
offset += sizeof(i);

float4 f4;
ALIGN_UP(offset, 16); // float4’s alignment is 16
cuParamSetv(cuFunction, offset, &f4, sizeof(f4));
offset += sizeof(f4);

char c;
ALIGN_UP(offset, __alignof(c));
cuParamSeti(cuFunction, offset, c);
offset += sizeof(c);

CUDA Programming Guide Version 2.3 49

Chapter 3. Programming Interface

float f;
ALIGN_UP(offset, __alignof(f));
cuParamSeti(cuFunction, offset, f);
offset += sizeof(f);

CUdeviceptr dptr;
// void* should be used to determine CUdeviceptr’s alignment
void* ptr = (void*)(size_t)dptr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);

float2 f2;
ALIGN_UP(offset, 8); // float2’s alignment is 8
cuParamSetv(cuFunction, offset, &f2, sizeof(f2));
offset += sizeof(f2);

cuParamSetSize(cuFunction, offset);

cuFuncSetBlockShape(cuFunction, blockWidth, blockHeight, 1);

cuLaunchGrid(cuFunction, gridWidth, gridHeight);
The alignment requirement of a structure is equal to the maximum of the alignment
requirements of its fields. The alignment requirement of a structure that contains
built-in vector types, CUdeviceptr, or non-aligned double and long long,
might therefore differ between device code and host code. Such a structure might
also be padded differently. The following structure, for example, is not padded at all
in host code, but it is padded in device code with 12 bytes after field f since the
alignment requirement for field f4 is 16.
typedef struct {
float f;
float4 f4;
} myStruct;
Any parameter of type myStruct must therefore be passed using separate calls to
cuParam*(), such as:
myStruct s;
int offset = 0;

cuParamSetv(cuFunction, offset, &s.f, sizeof(s.f));

offset += sizeof(s.f);

ALIGN_UP(offset, 16); // float4’s alignment is 16

cuParamSetv(cuFunction, offset, &s.f4, sizeof(s.f4));
offset += sizeof(s.f4);

3.3.4 Device Memory

Linear memory is allocated using cuMemAlloc() or cuMemAllocPitch() and
freed using cuMemFree().
Here is the host code of the sample from Section 3.2.1 written using the driver API:
// Host code
int main()
{

50 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

// Initialize
if (cuInit(0) != CUDA_SUCCESS)
exit (0);

// Get number of devices supporting CUDA

int deviceCount = 0;
cuDeviceGetCount(&deviceCount);
if (deviceCount == 0) {
printf("There is no device supporting CUDA.\n");
exit (0);
}

// Get handle for device 0

CUdevice cuDevice = 0;
cuDeviceGet(&cuDevice, 0);

// Create context
CUcontext cuContext;
cuCtxCreate(&cuContext, 0, cuDevice);

// Create module from binary file

CUmodule cuModule;
cuModuleLoad(&cuModule, “VecAdd.ptx”);

// Get function handle from module

CUfunction vecAdd;
cuModuleGetFunction(&vecAdd, cuModule, "VecAdd");

// Allocate vectors in device memory

size_t size = N * sizeof(float);
CUdeviceptr d_A;
cuMemAlloc(&d_A, size);
CUdeviceptr d_B;
cuMemAlloc(&d_B, size);
CUdeviceptr d_C;
cuMemAlloc(&d_C, size);

// Copy vectors from host memory to device memory

// h_A and h_B are input vectors stored in host memory
cuMemcpyHtoD(d_A, h_A, size);
cuMemcpyHtoD(d_B, h_B, size);

// Invoke kernel
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr;
ptr = (void*)(size_t)d_A;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)d_B;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)d_C;
ALIGN_UP(offset, __alignof(ptr));

CUDA Programming Guide Version 2.3 51

Chapter 3. Programming Interface

cuParamSetv(vecAdd, offset, &ptr, sizeof(ptr));

offset += sizeof(ptr);
cuParamSetSize(VecAdd, offset);
int threadsPerBlock = 256;
int blocksPerGrid =
(N + threadsPerBlock – 1) / threadsPerBlock;
cuFuncSetBlockShape(vecAdd, threadsPerBlock, 1, 1);
cuLaunchGrid(VecAdd, blocksPerGrid, 1);

// Copy result from device memory to host memory

// h_C contains the result in host memory
cuMemcpyDtoH(h_C, d_C, size);

// Free device memory

cuMemFree(d_A);
cuMemFree(d_B);
cuMemFree(d_C);
}
Linear memory can also be allocated through cuMemAllocPitch(). This function
is recommended for allocations of 2D arrays as it makes sure that the allocation is
appropriately padded to meet the alignment requirements described in
Section 5.1.2.1, therefore ensuring best performance when accessing the row
addresses or performing copies between 2D arrays and other regions of device
memory (using the cuMemcpy2D()). The returned pitch (or stride) must be used to
access array elements. The following code sample allocates a width×height 2D
array of floating-point values and shows how to loop over the array elements in
device code:
// Host code (assuming cuModule has been loaded)
CUdeviceptr devPtr;
int pitch;
cuMemAllocPitch(&devPtr, &pitch,
width * sizeof(float), height, 4);
CUfunction myKernel;
cuModuleGetFunction(&myKernel, cuModule, “myKernel”);
void* ptr = (void*)(size_t)devPtr;
cuParamSetv(myKernel, 0, &ptr, sizeof(ptr));
cuParamSetSize(myKernel, sizeof(ptr));
cuFuncSetBlockShape(myKernel, 512, 1, 1);
cuLaunchGrid(myKernel, 100, 1);

// Device code
__global__ void myKernel(float* devPtr)
{
for (int r = 0; r < height; ++r) {
float* row = (float*)((char*)devPtr + r * pitch);
for (int c = 0; c < width; ++c) {
float element = row[c];
}
}
}
The following code sample allocates a width×height CUDA array of one 32-bit
floating-point component:
CUDA_ARRAY_DESCRIPTOR desc;
desc.Format = CU_AD_FORMAT_FLOAT;

52 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

desc.NumChannels = 1;
desc.Width = width;
desc.Height = height;
CUarray cuArray;
cuArrayCreate(&cuArray, &desc);
The reference manual lists all the various functions used to copy memory between
linear memory allocated with cuMemAlloc(), linear memory allocated with
cuMemAllocPitch(), and CUDA arrays.
The following code sample copies the 2D array to the CUDA array allocated in the
previous code samples:
CUDA_MEMCPY2D copyParam;
memset(&copyParam, 0, sizeof(copyParam));
copyParam.dstMemoryType = CU_MEMORYTYPE_ARRAY;
copyParam.dstArray = cuArray;
copyParam.srcMemoryType = CU_MEMORYTYPE_DEVICE;
copyParam.srcDevice = devPtr;
copyParam.srcPitch = pitch;
copyParam.WidthInBytes = width * sizeof(float);
copyParam.Height = height;
cuMemcpy2D(&copyParam);
The following code sample copies some host memory array to constant memory:
__constant__ float constData[256];
float data[256];
CUdeviceptr devPtr;
unsigned int bytes;
cuModuleGetGlobal(&devPtr, &bytes, cuModule, “constData”);
cuMemcpyHtoD(devPtr, data, bytes);

3.3.5 Shared Memory

The following code sample is the driver version of the host code of the sample from
Section 3.2.2. Note how the Matrix type must be declared differently in host code
since device pointers are represented as a handle of type CUdeviceptr.
In this sample, shared memory is statically allocated within the kernel as opposed to
allocated at runtime through cuFuncSetSharedSize().
// Matrices are stored in row-major order:
// M(row, col) = *(M.elements + row * M.stride + col)
typedef struct {
int width;
int height;
int stride;
#ifdef __CUDACC__
float* elements;
#else
CUdeviceptr elements;
#endif
} Matrix;

// Matrix multiplication - Host code

// Matrix dimensions are assumed to be multiples of BLOCK_SIZE
void MatMul(const Matrix A, const Matrix B, Matrix C)
{

CUDA Programming Guide Version 2.3 53

Chapter 3. Programming Interface

// Load A and B to device memory

Matrix d_A;
d_A.width = d_A.stride = A.width; d_A.height = A.height;
size_t size = A.width * A.height * sizeof(float);
cuMemAlloc(&d_A.elements, size);
cuMemcpyHtoD(d_A.elements, A.elements, size);
Matrix d_B;
d_B.width = d_B.stride = B.width; d_B.height = B.height;
size = B.width * B.height * sizeof(float);
cuMemAlloc(&d_B.elements, size);
cuMemcpyHtoD(d_B.elements, B.elements, size);

// Allocate C in device memory

Matrix d_C;
d_C.width = d_C.stride = C.width; d_C.height = C.height;
size = C.width * C.height * sizeof(float);
cuMemAlloc(&d_C.elements, size);

// Invoke kernel (assuming cuModule has been loaded)

CUfunction matMulKernel;
cuModuleGetFunction(&matMulKernel, cuModule, "MatMulKernel");
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr;
ptr = (void*)(size_t)d_A;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(matMulKernel, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)d_B;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(matMulKernel, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)d_C;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(matMulKernel, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
cuParamSetSize(matMulKernel, offset);
cuFuncSetBlockShape(matMulKernel, BLOCK_SIZE, BLOCK_SIZE, 1);
cuLaunchGrid(matMulKernel,
B.width / dimBlock.x, A.height / dimBlock.y);

// Read C from device memory

cuMemcpyDtoH(C.elements, d_C.elements, size);

// Free device memory

cuMemFree(d_A.elements);
cuMemFree(d_B.elements);
cuMemFree(d_C.elements);
}

3.3.6 Multiple Devices

cuDeviceGetCount() and cuDeviceGet() provide a way to enumerate the
devices present in the system and other functions (described in the reference
manual) to retrieve their properties:

54 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

int deviceCount;
cuDeviceGetCount(&deviceCount);
int device;
for (int device = 0; device < deviceCount; ++device) {
CUdevice cuDevice;
cuDeviceGet(&cuDevice, device);
int major, minor;
cuDeviceComputeCapability(&major, &minor, cuDevice);
}

3.3.7 Texture Memory

Texure binding is done using cuTexRefSetAddress() for linear memory and
cuTexRefSetArray() for CUDA arrays.
If a module cuModule contains some texture reference texRef defined as
texture<float, 2, cudaReadModeElementType> texRef;
the following code sample retrieves texRef‘s handle:
CUtexref cuTexRef;
cuModuleGetTexRef(&cuTexRef, cuModule, “texRef”);
The following code sample binds texRef to some linear memory pointed to by
devPtr:
CUDA_ARRAY_DESCRIPTOR desc;
cuTexRefSetAddress2D(cuTexRef, &desc, devPtr, pitch);
The following code samples bind texRef to a CUDA array cuArray:
cuTexRefSetArray(cuTexRef, cuArray, CU_TRSA_OVERRIDE_FORMAT);
The reference manual lists various functions used to set address mode, filter mode,
format, and other flags for some texture reference. The format specified when
binding a texture to a texture reference must match the parameters specified when
declaring the texture reference; otherwise, the results of texture fetches are
undefined.
The following code sample is the driver version of the host code of the sample from
Section 3.2.4.3.
// Host code
int main()
{
// Allocate CUDA array in device memory
CUarray cuArray;
CUDA_ARRAY_DESCRIPTOR desc;
desc.Format = CU_AD_FORMAT_FLOAT;
desc.NumChannels = 1;
desc.Width = width;
desc.Height = height;
cuArrayCreate(&cuArray, &desc);

// Copy to device memory some data located at address h_data

// in host memory
CUDA_MEMCPY2D copyParam;
memset(&copyParam, 0, sizeof(copyParam));
copyParam.dstMemoryType = CU_MEMORYTYPE_ARRAY;
copyParam.dstArray = cuArray;

CUDA Programming Guide Version 2.3 55

Chapter 3. Programming Interface

copyParam.srcMemoryType = CU_MEMORYTYPE_HOST;
copyParam.srcHost = h_data;
copyParam.srcPitch = width * sizeof(float);
copyParam.WidthInBytes = copyParam.srcPitch;
copyParam.Height = height;
cuMemcpy2D(&copyParam);

// Set texture parameters

CUtexref texRef;
cuModuleGetTexRef(&texRef, cuModule, "texRef"));
cuTexRefSetAddressMode(texRef, 0, CU_TR_ADDRESS_MODE_WRAP);
cuTexRefSetAddressMode(texRef, 1, CU_TR_ADDRESS_MODE_WRAP);
cuTexRefSetFilterMode(texRef, CU_TR_FILTER_MODE_LINEAR);
cuTexRefSetFlags(texRef, CU_TRSF_NORMALIZED_COORDINATES);
cuTexRefSetFormat(texRef, CU_AD_FORMAT_FLOAT, 1);

// Bind the array to the texture

cuTexRefSetArray(texRef, cuArray, CU_TRSA_OVERRIDE_FORMAT);

// Allocate result of transformation in device memory

CUdeviceptr output;
cuMemAlloc((void**)&output, width * height * sizeof(float));

// Invoke kernel (assuming cuModule has been loaded)

CUfunction transformKernel;
cuModuleGetFunction(&transformKernel,
cuModule, "transformKernel");
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)output;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(transformKernel, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(transformKernel, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(transformKernel, offset, height);
offset += sizeof(height);
ALIGN_UP(offset, __alignof(angle));
cuParamSetf(transformKernel, offset, angle);
offset += sizeof(angle);
cuParamSetSize(transformKernel, offset));
cuParamSetTexRef(transformKernel,
CU_PARAM_TR_DEFAULT, texRef);
cuFuncSetBlockShape(transformKernel, 16, 16, 1);
cuLaunchGrid(transformKernel,
(width + dimBlock.x – 1) / dimBlock.x,
(height + dimBlock.y – 1) / dimBlock.y);

// Free device memory

cuArrayDestroy(cuArray);
cuMemFree(output);
}

56 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

3.3.8 Page-Locked Host Memory

Page-locked host memory can be allocated using cuMemHostAlloc() with
optional mutually non-exclusive flags:
‰ CU_MEMHOSTALLOC_PORTABLE to allocate memory that is portable across
CUDA contexts (see Section 3.2.5.1) 3.2.5.2;
‰ CU_MEMHOSTALLOC_WRITECOMBINED to allocate memory as write-combining
(see Section 3.2.5.2);
‰ CU_MEMHOSTALLOC_DEVICEMAP to allocate mapped page-locked memory (see
Section 3.2.5.3).
Page-locked host memory is freed using cuMemFreeHost().
Page-locked memory mapping is enabled for a CUDA context by creating the
context with the CU_CTX_MAP_HOST flag and device pointers to mapped page-
locked memory are retrieved using cuMemHostGetDevicePointer().
Applications may query whether a device supports mapped page-locked host
memory or not by checking the
CU_DEVICE_ATTRIBUTE_CAN_MAP_HOST_MEMORY attribute using
cuDeviceGetAttribute().

3.3.9 Asynchronous Concurrent Execution

Applications may query if a device can perform copies between page-locked host
memory and device memory concurrently with kernel execution by checking the
CU_DEVICE_ATTRIBUTE_GPU_OVERLAP attribute using
cuDeviceGetAttribute().

3.3.9.1 Stream
The driver API provides functions similar to the runtime API to manage streams.
The following code sample is the driver version of the code sample from
Section 3.2.6.1.
CUstream stream[2];
for (int i = 0; i < 2; ++i)
cuStreamCreate(&stream[i], 0);
float* hostPtr;
cuMemAllocHost((void**)&hostPtr, 2 * size);

for (int i = 0; i < 2; ++i)

cuMemcpyHtoDAsync(inputDevPtr + i * size, hostPtr + i * size,
size, stream[i]);
for (int i = 0; i < 2; ++i) {
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr;
ptr = (void*)(size_t)outputDevPtr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)inputDevPtr;

CUDA Programming Guide Version 2.3 57

Chapter 3. Programming Interface

ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(size));
cuParamSeti(cuFunction, offset, size);
offset += sizeof(int);
cuParamSetSize(cuFunction, offset);
cuFuncSetBlockShape(cuFunction, 512, 1, 1);
cuLaunchGridAsync(cuFunction, 100, 1, stream[i]);
}
for (int i = 0; i < 2; ++i)
cuMemcpyDtoHAsync(hostPtr + i * size, outputDevPtr + i * size,
size, stream[i]);
cuCtxSynchronize();

for (int i = 0; i < 2; ++i)

cuStreamDestroy(&stream[i]);

3.3.9.2 Event Management

The driver API provides functions similar to the runtime API to manage events.
The following code sample is the driver version of the code sample from
Section 3.2.6.2.
CUevent start, stop;
cuEventCreate(&start);
cuEventCreate(&stop);

cuEventRecord(start, 0);
for (int i = 0; i < 2; ++i)
cuMemcpyHtoDAsync(inputDevPtr + i * size, hostPtr + i * size,
size, stream[i]);
for (int i = 0; i < 2; ++i) {
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr;
ptr = (void*)(size_t)outputDevPtr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ptr = (void*)(size_t)inputDevPtr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(size));
cuParamSeti(cuFunction, offset, size);
offset += sizeof(size);
cuParamSetSize(cuFunction, offset);
cuFuncSetBlockShape(cuFunction, 512, 1, 1);
cuLaunchGridAsync(cuFunction, 100, 1, stream[i]);
}
for (int i = 0; i < 2; ++i)
cuMemcpyDtoHAsync(hostPtr + i * size, outputDevPtr + i * size,
size, stream[i]);
cuEventRecord(stop, 0);

58 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

cuEventSynchronize(stop);
float elapsedTime;
cuEventElapsedTime(&elapsedTime, start, stop);
They are destroyed this way:
cuEventDestroy(start);
cuEventDestroy(stop);

3.3.9.3 Synchronous Calls

Whether the host thread will yield, block, or spin on a synchronous function call can
be specified by calling cuCtxCreate() with some specific flags as described in the
reference manual.

3.3.10 OpenGL Interoperability

The driver API provides functions similar to the runtime API to manage OpenGL
interoperability.
Interoperability with OpenGL requires that the CUDA context be specifically
created using cuGLCtxCreate() instead of cuCtxCreate().
A buffer object must be registered to CUDA before it can be mapped. This is done
with cuGLRegisterBufferObject():
GLuint bufferObj;
cuGLRegisterBufferObject(bufferObj);
Once it is registered, a buffer object can be read from or written to by kernels using
the device memory address returned by cuGLMapBufferObject():
GLuint bufferObj;
float* devPtr;
cuGLMapBufferObject((void**)&devPtr, bufferObj);
Unmapping is done with cuGLUnmapBufferObject() and unregistering with
cuGLUnregisterBufferObject().
The following code sample is the driver version of the code sample from
Section 3.2.6.3.
CUfunction createVertices;
int main()
{
// Initialize driver API
...

// Get handle for device 0

CUdevice cuDevice = 0;
cuDeviceGet(&cuDevice, 0);

// Create context
CUcontext cuContext;
cuGLCtxCreate(&cuContext, 0, cuDevice);

// Create module from binary file

CUmodule cuModule;
cuModuleLoad(&cuModule, “createVertices.ptx”);

// Get function handle from module

CUDA Programming Guide Version 2.3 59

Chapter 3. Programming Interface

cuModuleGetFunction(&createVertices,
cuModule, "createVertices");

// Initialize OpenGL and GLUT

...
glutDisplayFunc(display);

// Create buffer object and register it with CUDA

glGenBuffers(1, positionsVBO);
glBindBuffer(GL_ARRAY_BUFFER, &vbo);
unsigned int size = width * height * 4 * sizeof(float);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, 0);
cuGLRegisterBufferObject(positionsVBO);

// Launch rendering loop

glutMainLoop();
}

void display()
{
// Map OpenGL buffer object for writing from CUDA
CUdeviceptr positions;
unsigned int size;
cuGLMapBufferObject(&positions, &size, positionsVBO);

// Execute kernel
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)positions;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(createVertices, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(time));
cuParamSetf(createVertices, offset, time);
offset += sizeof(time);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(createVertices, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(createVertices, offset, height);
offset += sizeof(height);
cuParamSetSize(createVertices, offset);
int threadsPerBlock = 16;
cuFuncSetBlockShape(createVertices,
threadsPerBlock, threadsPerBlock, 1);
cuLaunchGrid(createVertices,
width / threadsPerBlock, height / threadsPerBlock);

// Unmap buffer object

cuGLUnmapBufferObject(positionsVBO);

// Render from buffer object

glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
glBindBuffer(GL_ARRAY_BUFFER, positionsVBO);
glVertexPointer(4, GL_FLOAT, 0, 0);

60 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

glEnableClientState(GL_VERTEX_ARRAY);
glDrawArrays(GL_POINTS, 0, width * height);
glDisableClientState(GL_VERTEX_ARRAY);

// Swap buffers
glutSwapBuffers();
glutPostRedisplay();
}

void deleteVBO()
{
cuGLUnregisterBufferObject(positionsVBO);
glDeleteBuffers(1, &positionsVBO);
}
On Windows and for Quadro GPUs, cuWGLGetDevice() can be used to retrieve
the CUDA device associated to the handle returned by
WGL_NV_gpu_affinity().

3.3.11 Direct3D Interoperability

The driver API provides functions similar to the runtime API to manage Direct3D
interoperability in the same way.
Interoperability with Direct3D requires that the Direct3D device be specified when
the CUDA context is created. This is done by creating the CUDA context using
cuD3D9CtxCreate() (resp. cuD3D10CtxCreate()) instead of
cuCtxCreate().
Direct3D resources can then be registered to CUDA using
cuD3D9RegisterResource() (resp. cuD3D10RegisterResource()):
IDirect3DVertexBuffer9* buffer;
cuD3D9RegisterResource(buffer, CU_D3D9_REGISTER_FLAGS_NONE);
IDirect3DSurface9* surface;
cuD3D9RegisterResource(surface, CU_D3D9_REGISTER_FLAGS_NONE);

ID3D10Buffer* buffer;
cuD3D10RegisterResource(buffer, CU_D3D10_REGISTER_FLAGS_NONE);
ID3D10Texture2D* tex2D;
cuD3D10RegisterResource(tex2D, CU_D3D10_REGISTER_FLAGS_NONE);
cuD3D9RegisterResource() (resp. cuD3D10RegisterResource()) is
potentially high-overhead and typically called only once per resource. Unregistering
is done with cuD3D9UnregisterVertexBuffer() (resp.
cuD3D10UnregisterVertexBuffer()).
Once a resource is registered to CUDA, it can be mapped and unmapped as many
times as necessary using cuD3D9MapResources() (resp.
cuD3D10MapResources()) and cuD3D9UnmapResources() (resp.
cuD3D10UnmapResources()).
A mapped resource can be read from or written to by kernels using the device
memory address returned by cuD3D9ResourceGetMappedPointer() (resp.
cuD3D10ResourceGetMappedPointer()) and the size and pitch information
returned by cuD3D9ResourceGetMappedSize() (resp.
cuD3D10ResourceGetMappedSize()),

CUDA Programming Guide Version 2.3 61

Chapter 3. Programming Interface

cuD3D9ResourceGetMappedPitch() (resp.
cuD3D10ResourceGetMappedPitch()), and
cuD3D9ResourceGetMappedPitchSlice() (resp.
cuD3D10ResourceGetMappedPitchSlice()).
When applicable, a CUDA array can also be obtained from a mapped resource using
cuD3D9ResourceGetMappedArray() (resp.
cuD3D10ResourceGetMappedArray()).
Accessing a resource through Direct3D while it is mapped produces undefined
results.
The following code sample is the driver version of the host code of the sample from
Section 3.2.8.
IDirect3D9* D3D;
IDirect3DDevice9* device;
struct CUSTOMVERTEX {
FLOAT x, y, z;
DWORD color;
};
IDirect3DVertexBuffer9* positionsVB;

int main()
{
// Initialize Direct3D
D3D = Direct3DCreate9(D3D_SDK_VERSION);

// Get a CUDA-enabled adapter

unsigned int adapter = 0;
for (; adapter < g_pD3D->GetAdapterCount(); adapter++) {
D3DADAPTER_IDENTIFIER9 adapterId;
g_pD3D->GetAdapterIdentifier(adapter, 0, &adapterId);
int dev;
cuD3D9GetDevice(&dev, adapterId.DeviceName);
if (cudaSuccess == cudaGetLastError())
break;
}

// Create device
...
D3D->CreateDevice(adapter, D3DDEVTYPE_HAL, hWnd,
D3DCREATE_HARDWARE_VERTEXPROCESSING,
&params, &device);

// Initialize driver API

...

// Create context
CUdevice cuDevice;
CUcontext cuContext;
cuD3D9CtxCreate(&cuContext, &cuDevice, 0, &device);

// Create module from binary file

CUmodule cuModule;
cuModuleLoad(&cuModule, “createVertices.ptx”);

62 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

// Get function handle from module

cuModuleGetFunction(&createVertices,
cuModule, "createVertices");

// Create vertex buffer and register it with CUDA

unsigned int size = width * height * sizeof(CUSTOMVERTEX);
device->CreateVertexBuffer(size, 0, D3DFVF_CUSTOMVERTEX,
D3DPOOL_DEFAULT, &positionsVB, 0);
cuD3D9RegisterResource(positionsVB,
CU_D3D9_REGISTER_FLAGS_NONE);
cuD3D9ResourceSetMapFlags(positionsVB,
CU_D3D9_MAPRESOURCE_FLAGS_WRITEDISCARD);

// Launch rendering loop

while (...) {
...
Render();
...
}
}

void Render()
{
// Map vertex buffer for writing from CUDA
float4* positions;
cuD3D9MapResources(1, (IDirect3DResource9**)&positionsVB);
cuD3D9ResourceGetMappedPointer((void**)&positions,
positionsVB, 0, 0);

// Execute kernel
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)positions;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(createVertices, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(time));
cuParamSetf(createVertices, offset, time);
offset += sizeof(time);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(createVertices, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(createVertices, offset, height);
offset += sizeof(height);
cuParamSetSize(createVertices, offset);
int threadsPerBlock = 16;
cuFuncSetBlockShape(createVertices,
threadsPerBlock, threadsPerBlock, 1);
cuLaunchGrid(createVertices,
width / threadsPerBlock, height / threadsPerBlock);

// Unmap vertex buffer

cuD3D9UnmapResources(1, (IDirect3DResource9**)&positionsVB);

// Draw and present

CUDA Programming Guide Version 2.3 63

Chapter 3. Programming Interface

...
}

void releaseVB()
{
cuD3D9UnregisterResource(positionsVB);
positionsVB->Release();
}

ID3D10Device* device;
struct CUSTOMVERTEX {
FLOAT x, y, z;
DWORD color;
};
ID3D10Buffer* positionsVB;

int main()
{
// Get a CUDA-enabled adapter
IDXGIFactory* factory;
CreateDXGIFactory(__uuidof(IDXGIFactory), (void**)&factory);
IDXGIAdapter* adapter = 0;
for (unsigned int i = 0; !adapter; ++i) {
if (FAILED(factory->EnumAdapters(i, &adapter))
break;
int dev;
cuD3D10GetDevice(&dev, adapter);
if (cudaSuccess == cudaGetLastError())
break;
adapter->Release();
}
factory->Release();

// Create swap chain and device

...
D3D10CreateDeviceAndSwapChain(adapter,
D3D10_DRIVER_TYPE_HARDWARE, 0,
D3D10_CREATE_DEVICE_DEBUG,
D3D10_SDK_VERSION,
&swapChainDesc &swapChain,
&device);
adapter->Release();

// Initialize driver API

...

// Create context
CUdevice cuDevice;
CUcontext cuContext;
cuD3D10CtxCreate(&cuContext, &cuDevice, 0, &device);

// Create module from binary file

CUmodule cuModule;
cuModuleLoad(&cuModule, “createVertices.ptx”);

// Get function handle from module

64 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

cuModuleGetFunction(&createVertices,
cuModule, "createVertices");

// Create vertex buffer and register it with CUDA

unsigned int size = width * height * sizeof(CUSTOMVERTEX);
D3D10_BUFFER_DESC bufferDesc;
bufferDesc.Usage = D3D10_USAGE_DEFAULT;
bufferDesc.ByteWidth = size;
bufferDesc.BindFlags = D3D10_BIND_VERTEX_BUFFER;
bufferDesc.CPUAccessFlags = 0;
bufferDesc.MiscFlags = 0;
device->CreateBuffer(&bufferDesc, 0, &positionsVB);
cuD3D10RegisterResource(positionsVB,
CU_D3D10_REGISTER_FLAGS_NONE);
cuD3D10ResourceSetMapFlags(positionsVB,
CU_D3D10_MAPRESOURCE_FLAGS_WRITEDISCARD);

// Launch rendering loop

while (...) {
...
Render();
...
}
}

void Render()
{
// Map vertex buffer for writing from CUDA
float4* positions;
cuD3D10MapResources(1, (ID3D10Resource**)&positionsVB);
cuD3D10ResourceGetMappedPointer((void**)&positions,
positionsVB, 0);

// Execute kernel
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)positions;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(createVertices, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(time));
cuParamSetf(createVertices, offset, time);
offset += sizeof(time);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(createVertices, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(createVertices, offset, height);
offset += sizeof(height);
cuParamSetSize(createVertices, offset);
int threadsPerBlock = 16;
cuFuncSetBlockShape(createVertices,
threadsPerBlock, threadsPerBlock, 1);
cuLaunchGrid(createVertices,
width / threadsPerBlock, height / threadsPerBlock);

CUDA Programming Guide Version 2.3 65

Chapter 3. Programming Interface

// Unmap vertex buffer

cuD3D10UnmapResources(1, (ID3D10Resource**)&positionsVB);

// Draw and present

...
}

void releaseVB()
{
cuD3D10UnregisterResource(positionsVB);
positionsVB->Release();
}
In the following code sample, each thread accesses one pixel of a 2D surface of size
(width, height) and pixel format float4:
// host code
CUdeviceptr devPtr;
cuD3D9ResourceGetMappedPointer(&devPtr, surface, 0, 0);
size_t pitch;
cuD3D9ResourceGetMappedPitch(&pitch, 0, surface, 0, 0);
cuModuleGetFunction(&cuFunction, cuModule, “myKernel”);
#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)devPtr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(cuFunction, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(cuFunction, offset, height);
offset += sizeof(height);
ALIGN_UP(offset, __alignof(pitch));
cuParamSeti(cuFunction, offset, pitch);
offset += sizeof(pitch);
cuParamSetSize(cuFunction, offset);
cuFuncSetBlockShape(cuFunction, 16, 16, 1);
cuLaunchGrid(cuFunction,
(width+Db.x–1)/Db.x, (height+Db.y–1)/Db.y);

// device code
__global__ void myKernel(unsigned char* surface,
int width, int height, size_t pitch)
{
int x = blockIdx.x * blockDim.x + threadIdx.x;
int y = blockIdx.y * blockDim.y + threadIdx.y;
if (x >= width || y >= height) return;
float* pixel = (float*)(surface + y * pitch) + 4 * x;
}

// host code
CUdeviceptr devPtr;
cuD3D10ResourceGetMappedPointer(&devPtr, surface, 0);
size_t pitch;
cuD3D10ResourceGetMappedPitch(&pitch, 0, surface, 0);

66 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

cuModuleGetFunction(&cuFunction, cuModule, “myKernel”);

#define ALIGN_UP(offset, alignment) \
(offset) = ((offset) + (alignment) – 1) & ~((alignment) – 1)
int offset = 0;
void* ptr = (void*)(size_t)devPtr;
ALIGN_UP(offset, __alignof(ptr));
cuParamSetv(cuFunction, offset, &ptr, sizeof(ptr));
offset += sizeof(ptr);
ALIGN_UP(offset, __alignof(width));
cuParamSeti(cuFunction, offset, width);
offset += sizeof(width);
ALIGN_UP(offset, __alignof(height));
cuParamSeti(cuFunction, offset, height);
offset += sizeof(height);
ALIGN_UP(offset, __alignof(pitch));
cuParamSeti(cuFunction, offset, pitch);
offset += sizeof(pitch);
cuParamSetSize(cuFunction, offset);
cuFuncSetBlockShape(cuFunction, 16, 16, 1);
cuLaunchGrid(cuFunction,
(width+Db.x–1)/Db.x, (height+Db.y–1)/Db.y);

// device code
__global__ void myKernel(unsigned char* surface,
int width, int height, size_t pitch)
{
int x = blockIdx.x * blockDim.x + threadIdx.x;
int y = blockIdx.y * blockDim.y + threadIdx.y;
if (x >= width || y >= height) return;
float* pixel = (float*)(surface + y * pitch) + 4 * x;
}

3.3.12 Error Handling

All driver functions return an error code, but for an asynchronous function (see
Section 3.2.6), this error code cannot possibly report any of the asynchronous errors
that could occur on the device since the function returns before the device has
completed the task; the error code only reports errors that occur on the host prior
to executing the task, typically related to parameter validation; if an asynchronous
error occurs, it will be reported by some subsequent unrelated runtime function call.
The only way to check for asynchronous errors just after some asynchronous
function call is therefore to synchronize just after the call by calling
cuCtxSynchronize() (or by using any other synchronization mechanisms
described in Section 3.3.9) and checking the error code returned by
cuCtxSynchronize().

3.4 Versioning and Compatibility

There are two version numbers that developers should care about when developing
a CUDA application: The compute capability that describes the general
specifications and features of the compute device (see Section 2.5) and the version

CUDA Programming Guide Version 2.3 67

Chapter 3. Programming Interface

of the CUDA driver API that describes the features supported by the driver API
and runtime.
The version of the driver API is defined in the driver header file as
CUDA_VERSION. It allows developers to check whether their application requires a
newer driver than the one currently installed. This is important, because the driver
API is backward compatible, meaning that applications, plug-ins, and libraries
(including the C runtime) compiled against a particular version of the driver API will
continue to work on subsequent driver releases as illustrated in Figure 3-4. The
driver API is not forward compatible, which means that applications, plug-ins, and
libraries (including the C runtime) compiled against a particular version of the driver
API will not work on previous versions of the driver.

Figure 3-4.The Driver API is Backward, but Not Forward

Compatible
It is important to note that mixing and matching versions is not supported;
specifically:
‰ All applications, plug-ins, and libraries on a system must use the same version of
the CUDA driver API, since only one version of the CUDA driver can be
installed on a system.
‰ All plug-ins and libraries used by an application must use the same version of the
runtime.
‰ All plug-ins and libraries used by an application must use the same version of
any libraries that use the runtime (such as CUFFT, CUBLAS, …).

3.5 Compute Modes

On Tesla solutions running Linux, one can set any device in a system in one of the
three following modes using NVIDIA’s System Management Interface (nvidia-smi),
which is a tool distributed as part of the Linux driver:

68 CUDA Programming Guide Version 2.3

 Chapter 3. Programming Interface

‰ Default compute mode: Multiple host threads can use the device (by calling
cudaSetDevice() on this device, when using the runtime API, or by making
current a context associated to the device, when using the driver API) at the
same time.
‰ Exclusive compute mode: Only one host thread can use the device at any given
time.
‰ Prohibited compute mode: No host thread can use the device.
This means, in particular, that a host thread using the runtime API without explicitly
calling cudaSetDevice() might be associated with a device other than device 0 if
device 0 turns out to be in prohibited compute mode or in exclusive compute mode
and used by another host thread. cudaSetValidDevices() can be used to set a
device from a prioritized list of devices.
Applications may query the compute mode of a device by calling
cudaGetDeviceProperties() and checking the computeMode property or
checking the CU_DEVICE_COMPUTE_MODE attribute using
cuDeviceGetAttribute().

3.6 Mode Switches

GPUs dedicate some DRAM memory to the so-called primary surface, which is used
to refresh the display device whose output is viewed by the user. When users initiate
a mode switch of the display by changing the resolution or bit depth of the display
(using NVIDIA control panel or the Display control panel on Windows), the
amount of memory needed for the primary surface changes. For example, if the user
changes the display resolution from 1280x1024x32-bit to 1600x1200x32-bit, the
system must dedicate 7.68 MB to the primary surface rather than 5.24 MB. (Full-
screen graphics applications running with anti-aliasing enabled may require much
more display memory for the primary surface.) On Windows, other events that may
initiate display mode switches include launching a full-screen DirectX application,
hitting Alt+Tab to task switch away from a full-screen DirectX application, or
hitting Ctrl+Alt+Del to lock the computer.
If a mode switch increases the amount of memory needed for the primary surface,
the system may have to cannibalize memory allocations dedicated to CUDA
applications. Therefore, a mode switch results in any call to the CUDA runtime to
fail and return an invalid context error.

CUDA Programming Guide Version 2.3 69

 Chapter 4.
Hardware Implementation

4.1 A Set of SIMT Multiprocessors with On-Chip

Shared Memory
The CUDA architecture is built around a scalable array of multithreaded Streaming
Multiprocessors (SMs). When a CUDA program on the host CPU invokes a kernel
grid, the blocks of the grid are enumerated and distributed to multiprocessors with
available execution capacity as illustrated in Figure 4-1. The threads of a thread
block execute concurrently on one multiprocessor. As thread blocks terminate, new
blocks are launched on the vacated multiprocessors.

CUDA Programming Guide Version 2.3 71

Chapter 4. Hardware Implementation

Kernel Grid

Block 0 Block 1 Block 2 Block 3

Block 4 Block 5 Block 6 Block 7

Block 5 Block 6

Device with 2 SMs Device with 4 SMs

SM 0 SM 1 SM 0 SM 1 SM 2 SM 3

Block 0 Block 1 Block 0 Block 1 Block 2 Block 3

Block 2 Block 3 Block 4 Block 5 Block 6 Block 7

Block 4 Block 5

Block 6 Block 7

A device with more multiprocessors will automatically execute a kernel grid in less time than a device
with fewer multiprocessors.

Figure 4-1.Automatic Scalability

A multiprocessor consists of eight Scalar Processor (SP) cores, two special function
units for transcendentals, a multithreaded instruction unit, and on-chip shared
memory. The multiprocessor creates, manages, and executes concurrent threads in
hardware with zero scheduling overhead. It implements the __syncthreads()
barrier synchronization intrinsic with a single instruction. Fast barrier
synchronization together with lightweight thread creation and zero-overhead thread
scheduling efficiently support very fine-grained parallelism, allowing, for example, a
low granularity decomposition of problems by assigning one thread to each data
element (such as a pixel in an image, a voxel in a volume, a cell in a grid-based
computation).
To manage hundreds of threads running several different programs, the
multiprocessor employs a new architecture we call SIMT (single-instruction,
multiple-thread). The multiprocessor maps each thread to one scalar processor core,
and each scalar thread executes independently with its own instruction address and
register state. The multiprocessor SIMT unit creates, manages, schedules, and
executes threads in groups of 32 parallel threads called warps. (This term originates
from weaving, the first parallel thread technology. A half-warp is either the first or
second half of a warp.) Individual threads composing a SIMT warp start together at

72 CUDA Programming Guide Version 2.3

 Chapter 4: Hardware Implementation

the same program address but are otherwise free to branch and execute
independently.
When a multiprocessor is given one or more thread blocks to execute, it splits them
into warps that get scheduled by the SIMT unit. The way a block is split into warps
is always the same; each warp contains threads of consecutive, increasing thread IDs
with the first warp containing thread 0. Section 2.2 describes how thread IDs relate
to thread indices in the block.
Every instruction issue time, the SIMT unit selects a warp that is ready to execute
and issues the next instruction to the active threads of the warp. A warp executes
one common instruction at a time, so full efficiency is realized when all 32 threads
of a warp agree on their execution path. If threads of a warp diverge via a data-
dependent conditional branch, the warp serially executes each branch path taken,
disabling threads that are not on that path, and when all paths complete, the threads
converge back to the same execution path. Branch divergence occurs only within a
warp; different warps execute independently regardless of whether they are
executing common or disjointed code paths.
SIMT architecture is akin to SIMD (Single Instruction, Multiple Data) vector
organizations in that a single instruction controls multiple processing elements. A
key difference is that SIMD vector organizations expose the SIMD width to the
software, whereas SIMT instructions specify the execution and branching behavior
of a single thread. In contrast with SIMD vector machines, SIMT enables
programmers to write thread-level parallel code for independent, scalar threads, as
well as data-parallel code for coordinated threads. For the purposes of correctness,
the programmer can essentially ignore the SIMT behavior; however, substantial
performance improvements can be realized by taking care that the code seldom
requires threads in a warp to diverge. In practice, this is analogous to the role of
cache lines in traditional code: Cache line size can be safely ignored when designing
for correctness but must be considered in the code structure when designing for
peak performance. Vector architectures, on the other hand, require the software to
coalesce loads into vectors and manage divergence manually.
As illustrated by Figure 4-2, each multiprocessor has on-chip memory of the four
following types:
‰ One set of local 32-bit registers per processor,
‰ A parallel data cache or shared memory that is shared by all scalar processor cores
and is where the shared memory space resides,
‰ A read-only constant cache that is shared by all scalar processor cores and speeds
up reads from the constant memory space, which is a read-only region of device
memory,
‰ A read-only texture cache that is shared by all scalar processor cores and speeds up
reads from the texture memory space, which is a read-only region of device
memory; each multiprocessor accesses the texture cache via a texture unit that
implements the various addressing modes and data filtering mentioned in
Section 3.2.4.
The local and global memory spaces are read-write regions of device memory and
are not cached.
The number of blocks a multiprocessor can process at once – referred to as the
number of active blocks per multiprocessor – depends on how many registers per

CUDA Programming Guide Version 2.3 73

Chapter 4. Hardware Implementation

thread and how much shared memory per block are required for a given kernel since
the multiprocessor’s registers and shared memory are split among all the threads of
the active blocks. If there are not enough registers or shared memory available per
multiprocessor to process at least one block, the kernel will fail to launch. The
maximum number of active blocks per multiprocessor, as well as the maximum
number of active warps and maximum number of active threads are given in
Appendix A.
If a non-atomic instruction executed by a warp writes to the same location in global
or shared memory for more than one of the threads of the warp, the number of
serialized writes that occur to that location and the order in which they occur is
undefined, but one of the writes is guaranteed to succeed. If an atomic instruction
(see Section B.10) executed by a warp reads, modifies, and writes to the same
location in global memory for more than one of the threads of the warp, each read,
modify, write to that location occurs and they are all serialized, but the order in
which they occur is undefined.

74 CUDA Programming Guide Version 2.3

 Chapter 4: Hardware Implementation

Device

Multiprocessor N

Multiprocessor 2

Multiprocessor 1

Shared Memory

Registers Registers Registers

Instruction

Processor 1 Processor 2 … Processor M

Unit

Constant
Cache

Texture
Cache

Device Memory

A set of SIMT multiprocessors with on-chip shared memory.

Figure 4-2.Hardware Model

4.2 Multiple Devices

The use of multiple GPUs as CUDA devices by an application running on a multi-
GPU system is only guaranteed to work if these GPUs are of the same type.
When the system is in SLI mode, all GPUs are accessible via the CUDA driver and
runtime as separate devices, but there are special considerations as described below.
First, an allocation in one CUDA device on one GPU will consume memory on
other GPUs. Because of this, allocations may fail earlier than otherwise expected.

CUDA Programming Guide Version 2.3 75

Chapter 4. Hardware Implementation

Second, when a Direct3D application runs in SLI Alternate Frame Rendering mode,
the Direct3D device(s) created by that application can be used for CUDA-Direct3D
interoperability (i.e. passed as a parameter to
cudaD3D[9|10]SetDirect3DDevice() when using the runtime API), but only
one CUDA device can be created at a time from one of these Direct3D devices.
This CUDA device only executes the CUDA work on one of the GPUs in the SLI
configuration. As a consequence, real interoperability only happens with the copy of
a Direct3D resource in that GPU (Note: in AFR mode Direct3D resources that
must be in GPU memory are duplicated in the GPU memory of each GPU in the
SLI configuration). In some cases this is not the desired behavior and an application
may need to forfeit CUDA-Direct3D interoperability and manually copy the output
of its CUDA work to Direct3D resources using the existing CUDA and Direct3D
API.

76 CUDA Programming Guide Version 2.3

 Chapter 5.
Performance Guidelines

5.1 Instruction Performance

To process an instruction for a warp of threads, a multiprocessor must:
‰ Read the instruction operands for each thread of the warp,
‰ Execute the instruction,
‰ Write the result for each thread of the warp.
Therefore, the effective instruction throughput depends on the nominal instruction
throughput as well as the memory latency and bandwidth. It is maximized by:
‰ Minimizing the use of instructions with low throughput (see Section 5.1.1),
‰ Maximizing the use of the available memory bandwidth for each category of
memory (see Section 5.1.2),
‰ Allowing the thread scheduler to overlap memory transactions with
mathematical computations as much as possible, which requires that:
¾ The program executed by the threads is of high arithmetic intensity, that is,
has a high number of arithmetic operations per memory operation;
¾ There are many active threads per multiprocessor, as detailed in Section 5.2.

5.1.1 Instruction Throughput

In this section, throughputs are given in number of operations per clock cycle per
multiprocessor. For a warp size of 32, an instruction is made of 32 operations.
Therefore, if T is the number of operations per clock cycle, the instruction
throughput is one instruction every 32/T clock cycles.
All throughputs are for one multiprocessor. They must be multiplied by the number
of multiprocessors in the device to get throughput for the whole device.

5.1.1.1 Arithmetic Instructions

For single-precision floating-point code, we highly recommend use of the float
type and the single-precision floating-point mathematical functions. When
compiling for devices without native double-precision floating-point support, such
as devices of compute capability 1.2 and lower, each double variable gets
converted to single-precision floating-point format (but retains its size of 64 bits)

CUDA Programming Guide Version 2.3 77

Chapter 5. Performance Guidelines

and double-precision floating-point arithmetic gets demoted to single-precision

floating-point arithmetic.
Single-Precision Floating-Point Basic Arithmetic
Throughput of single-precision floating-point add, multiply, and multiply-add is 8
operations per clock cycle.
Throughput of reciprocal is 2 operations per clock cycle.
Throughput of single-precision floating-point division is 0.88 operations per clock
cycle, but __fdividef(x, y) (see Section C.2) provides a faster version with a
throughput of 1.6 operations per clock cycle.
Single-Precision Floating-Point Square Root and Reciprocal Square
Root
Throughput of reciprocal square root is 2 operations per clock cycle.
Single-precision floating-point square root is implemented as a reciprocal square
root followed by a reciprocal instead of a reciprocal square root followed by a
multiplication, so that it gives correct results for 0 and infinity. Therefore, its
throughput is 1 operation per clock cycle.
Single-Precision Floating-Point Logarithm
Throughput of __logf(x) (see Section C.2) is 2 operations per clock cycle.
Sine and Cosine
Throughput of __sinf(x), __cosf(x), __expf(x) (see Section C.2) is 1
operation per clock cycle.
sinf(x), cosf(x), tanf(x), sincosf(x) and corresponding double-precision
instructions are much more expensive and even more so if the absolute value of x
needs to be reduced.
More precisely, the argument reduction code (see math_functions.h for
implementation) comprises two code paths referred to as the fast path and the slow
path, respectively.
The fast path is used for arguments sufficiently small in magnitude and essentially
consists of a few multiply-add operations. The slow path is used for arguments large
in magnitude, and consists of lengthy computations required to achieve correct
results over the entire argument range.
At present, the argument reduction code for the trigonometric functions selects the
fast path for arguments whose magnitude is less than 48039.0f for the single-
precision functions, and less than 2147483648.0 for the double-precision functions.
As the slow path requires more registers than the fast path, an attempt has been
made to reduce register pressure in the slow path by storing some intermediate
variables in local memory, which may affect performance because of local memory
high latency and bandwidth (see Section 5.1.2.2). At present, 28 bytes of local
memory are used by single-precision functions, and 44 bytes are used by double-
precision functions. However, the exact amount is subject to change.

78 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Due to the lengthy computations and use of local memory in the slow path, the
trigonometric functions throughput is lower by one order of magnitude when the
slow path reduction is used as opposed to the fast path reduction.
Integer Arithmetic
Throughput of integer add is 8 operations per clock cycle.
Throughput of 32-bit integer multiplication is 2 operations per clock cycle, but
__mul24 and __umul24 (see Section C.2) provide signed and unsigned 24-bit
integer multiplication with a troughput of 8 operations per clock cycle. On future
architectures however, __[u]mul24 will be slower than 32-bit integer
multiplication, so we recommend to provide two kernels, one using __[u]mul24
and the other using generic 32-bit integer multiplication, to be called appropriately
by the application.
Integer division and modulo operation are particularly costly and should be avoided
if possible or replaced with bitwise operations whenever possible: If n is a power of
2, (i/n) is equivalent to (i>>log2(n)) and (i%n) is equivalent to (i&(n-1));
the compiler will perform these conversions if n is literal.
Comparison
Throughput of compare, min, max is 8 operations per clock cycle.
Bitwise Operations
Throughput of any bitwise operation is 8 operations per clock cycle.
Type Conversion
Throughput of type conversion operations is 8 operations per clock cycle.
Sometimes, the compiler must insert conversion instructions, introducing additional
execution cycles. This is the case for:
‰ Functions operating on char or short whose operands generally need to be
converted to int,
‰ Double-precision floating-point constants (defined without any type suffix) used
as input to single-precision floating-point computations.
This last case can be avoided by using single-precision floating-point constants,
defined with an f suffix such as 3.141592653589793f, 1.0f, 0.5f.

5.1.1.2 Control Flow Instructions

Any flow control instruction (if, switch, do, for, while) can significantly
impact the effective instruction throughput by causing threads of the same warp to
diverge, that is, to follow different execution paths. If this happens, the different
executions paths have to be serialized, increasing the total number of instructions
executed for this warp. When all the different execution paths have completed, the
threads converge back to the same execution path.
To obtain best performance in cases where the control flow depends on the thread
ID, the controlling condition should be written so as to minimize the number of
divergent warps. This is possible because the distribution of the warps across the
block is deterministic as mentioned in Section 4.1. A trivial example is when the
controlling condition only depends on (threadIdx / warpSize) where

CUDA Programming Guide Version 2.3 79

Chapter 5. Performance Guidelines

warpSize is the warp size. In this case, no warp diverges since the controlling
condition is perfectly aligned with the warps.
Sometimes, the compiler may unroll loops or it may optimize out if or switch
statements by using branch predication instead, as detailed below. In these cases, no
warp can ever diverge. The programmer can also control loop unrolling using the
#pragma unroll directive (see Section 3.1.2).
When using branch predication none of the instructions whose execution depends
on the controlling condition gets skipped. Instead, each of them is associated with a
per-thread condition code or predicate that is set to true or false based on the
controlling condition and although each of these instructions gets scheduled for
execution, only the instructions with a true predicate are actually executed.
Instructions with a false predicate do not write results, and also do not evaluate
addresses or read operands.
The compiler replaces a branch instruction with predicated instructions only if the
number of instructions controlled by the branch condition is less or equal to a
certain threshold: If the compiler determines that the condition is likely to produce
many divergent warps, this threshold is 7, otherwise it is 4.

5.1.1.3 Memory Instructions

Memory instructions include any instruction that reads from or writes to shared,
local or global memory. Local memory accesses only occur for some automatic
variables as detailed in Section B.2.4.
Throughput of memory operations is 8 operations per clock cycle. When accessing
local or global memory, there are, in addition, 400 to 600 clock cycles of memory
latency.
As an example, the throughput for the assignment operator in the following sample
code:
__shared__ float shared[32];
__device__ float device[32];
shared[threadIdx.x] = device[threadIdx.x];
is 8 operations per clock cycle for the read from global memory, 8 operations per
clock cycle for the write to shared memory, but above all, there is a latency of 400 to
600 clock cycles to read data from global memory.
Much of this global memory latency can be hidden by the thread scheduler if there
are sufficient independent arithmetic instructions that can be issued while waiting
for the global memory access to complete.

5.1.1.4 Synchronization Instruction

Throughput for __syncthreads is 8 operations per clock cycle in the case where
no thread has to wait for any other threads.

5.1.2 Memory Bandwidth

The effective bandwidth of each memory space depends significantly on the
memory access pattern as detailed in the following sub-sections.
Since device memory is of much higher latency and lower bandwidth than on-chip
memory, device memory accesses should be minimized. A typical programming

80 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

pattern is to stage data coming from device memory into shared memory; in other
words, to have each thread of a block:
‰ Load data from device memory to shared memory,
‰ Synchronize with all the other threads of the block so that each thread can safely
read shared memory locations that were written by different threads,
‰ Process the data in shared memory,
‰ Synchronize again if necessary to make sure that shared memory has been
updated with the results,
‰ Write the results back to device memory.

5.1.2.1 Global Memory

The global memory space is not cached, so it is all the more important to follow the
right access pattern to get maximum memory bandwidth, especially given how
costly accesses to device memory are.
First, the device is capable of reading 4-byte, 8-byte, or 16-byte words from global
memory into registers in a single instruction. To have assignments such as:
__device__ type device[32];
type data = device[tid];
compile to a single load instruction, type must be such that sizeof(type) is
equal to 4, 8, or 16 and variables of type type must be aligned to sizeof(type)
bytes (that is, have their address be a multiple of sizeof(type)).
The alignment requirement is automatically fulfilled for built-in types of
Section B.3.1 like float2 or float4.
For structures, the size and alignment requirements can be enforced by the compiler
using the alignment specifiers __align__(8) or __align__(16), such as
struct __align__(8) {
float a;
float b;
};
or
struct __align__(16) {
float a;
float b;
float c;
};
For structures larger than 16 bytes, the compiler generates several load instructions.
To ensure that it generates the minimum number of instructions, such structures
should be defined with __align__(16) , such as
struct __align__(16) {
float a;
float b;
float c;
float d;
float e;
};
which is compiled into two 16-byte load instructions instead of five 8-byte load
instructions.

CUDA Programming Guide Version 2.3 81

Chapter 5. Performance Guidelines

Any address of a variable residing in global memory or returned by one of the

memory allocation routines from the driver or runtime API is always aligned to at
least 256 bytes.
Reading mis-aligned 8-byte or 16-byte words produces incorrect results (off by a
few words), so special care must be taken to maintain alignment of the starting
address of any value or array of values of these types. A typical case where this
might be easily overlooked is when using some custom global memory allocation
scheme, whereby the allocations of multiple arrays (with multiple calls to
cudaMalloc() or cuMemAlloc()) is replaced by the allocation of a single large
block of memory partitioned into multiple arrays, in which case the starting address
of each array is offset from the block’s starting address.
Second, global memory bandwidth is used most efficiently when the simultaneous
memory accesses by threads in a half-warp (during the execution of a single read or
write instruction) can be coalesced into a single memory transaction of 32, 64, or 128
bytes.
The rest of this section describes the various requirements for memory accesses to
coalesce based on the compute capability of the device. If a half-warp fulfills these
requirements, coalescing is achieved even if the warp is divergent and some threads
of the half-warp do not actually access memory.
For the purpose of the following discussion, global memory is considered to be
partitioned into segments of size equal to 32, 64, or 128 bytes and aligned to this
size.
Coalescing on Devices with Compute Capability 1.0 and 1.1
The global memory access by all threads of a half-warp is coalesced into one or two
memory transactions if it satisfies the following three conditions:
‰ Threads must access
‰ Either 4-byte words, resulting in one 64-byte memory transaction,
‰ Or 8-byte words, resulting in one 128-byte memory transaction,
‰ Or 16-byte words, resulting in two 128-byte memory transactions;
‰ All 16 words must lie in the same segment of size equal to the memory
transaction size (or twice the memory transaction size when accessing 16-byte
words);
‰ Threads must access the words in sequence: The kth thread in the half-warp must
access the kth word.
If a half-warp does not fulfill all the requirements above, a separate memory
transaction is issued for each thread and throughput is significantly reduced.
Figure 5-1 shows some examples of coalesced memory accesses, while Figure 5-2
and Figure 5-3 show some examples of memory accesses that are non-coalesced for
devices of compute capability 1.0 or 1.1.
Coalesced 8-byte accesses deliver a little lower bandwidth than coalesced 4-byte
accesses and coalesced 16-byte accesses deliver a noticeably lower bandwidth than
coalesced 4-byte accesses. But, while bandwidth for non-coalesced accesses is
around an order of magnitude lower than for coalesced accesses when these
accesses are 4-byte, it is only around four times lower when they are 8-byte and
around two times when they are 16-byte.

82 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Coalescing on Devices with Compute Capability 1.2 and Higher

The global memory access by all threads of a half-warp is coalesced into a single
memory transaction as soon as the words accessed by all threads lie in the same
segment of size equal to:
‰ 32 bytes if all threads access 1-byte words,
‰ 64 bytes if all threads access 2-byte words,
‰ 128 bytes if all threads access 4-byte or 8-byte words.
Coalescing is achieved for any pattern of addresses requested by the half-warp,
including patterns where multiple threads access the same address. This is in
contrast with devices of lower compute capabilities where threads need to access
words in sequence.
If a half-warp addresses words in n different segments, n memory transactions are
issued (one for each segment), whereas devices with lower compute capabilities
would issue 16 transactions as soon as n is greater than 1. In particular, if threads
access 16-byte words, at least two memory transactions are issued.
Unused words in a memory transaction are still read, so they waste bandwidth. To
reduce waste, hardware will automatically issue the smallest memory transaction that
contains the requested words. For example, if all the requested words lie in one half
of a 128-byte segment, a 64-byte transaction will be issued.
More precisely, the following protocol is used to issue a memory transaction for a
half-warp:
‰ Find the memory segment that contains the address requested by the lowest
numbered active thread. Segment size is 32 bytes for 1-byte data, 64 bytes for
2-byte data, 128 bytes for 4-, 8- and 16-bit data.
‰ Find all other active threads whose requested address lies in the same segment.
‰ Reduce the transaction size, if possible:
‰ If the transaction size is 128 bytes and only the lower or upper half is used,
reduce the transaction size to 64 bytes;
‰ If the transaction size is 64 bytes and only the lower or upper half is used,
reduce the transaction sizez to 32 bytes.
‰ Carry out the transaction and mark the serviced threads as inactive.
‰ Repeat until all threads in the half-warp are serviced.

Figure 5-4 shows some examples of global memory accesses for devices of compute
capability 1.2 and higher.

CUDA Programming Guide Version 2.3 83

Chapter 5. Performance Guidelines

Thread 0 Address 128 Thread 0 Address 128

Thread 1 Address 132 Thread 1 Address 132

Thread 2 Address 136 Thread 2 Address 136

Thread 3 Address 140 Thread 3 Address 140

Thread 4 Address 144 Thread 4 Address 144

Thread 5 Address 148 Thread 5 Address 148

Thread 6 Address 152 Thread 6 Address 152

Thread 7 Address 156 Thread 7 Address 156

Thread 8 Address 160 Thread 8 Address 160

Thread 9 Address 164 Thread 9 Address 164

Thread 10 Address 168 Thread 10 Address 168

Thread 11 Address 172 Thread 11 Address 172

Thread 12 Address 176 Thread 12 Address 176

Thread 13 Address 180 Thread 13 Address 180

Thread 14 Address 184 Thread 14 Address 184

Thread 15 Address 188 Thread 15 Address 188

Left: coalesced float memory access, resulting in a single memory transaction.

Right: coalesced float memory access (divergent warp), resulting in a single memory transaction.

Figure 5-1.Examples of Coalesced Global Memory Access

Patterns

84 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Thread 0 Address 128 Thread 0 Address 128

Thread 1 Address 132 Thread 1 Address 132

Thread 2 Address 136 Thread 2 Address 136

Thread 3 Address 140 Thread 3 Address 140

Thread 4 Address 144 Thread 4 Address 144

Thread 5 Address 148 Thread 5 Address 148

Thread 6 Address 152 Thread 6 Address 152

Thread 7 Address 156 Thread 7 Address 156

Thread 8 Address 160 Thread 8 Address 160

Thread 9 Address 164 Thread 9 Address 164

Thread 10 Address 168 Thread 10 Address 168

Thread 11 Address 172 Thread 11 Address 172

Thread 12 Address 176 Thread 12 Address 176

Thread 13 Address 180 Thread 13 Address 180

Thread 14 Address 184 Thread 14 Address 184

Thread 15 Address 188 Thread 15 Address 188

Left: non-sequential float memory access, resulting in 16 memory transactions.

Right: access with a misaligned starting address, resulting in 16 memory transactions.

Figure 5-2.Examples of Global Memory Access Patterns That

Are Non-Coalesced for Devices of Compute
Capability 1.0 or 1.1

CUDA Programming Guide Version 2.3 85

Chapter 5. Performance Guidelines

Thread 0 Address 128 Thread 0 Address 128

Thread 1 Address 132

Thread 2 Address 136

Thread 3 Address 140 Thread 1 Address 140

Thread 4 Address 144

Thread 5 Address 148

Thread 6 Address 152 Thread 2 Address 152

Thread 7 Address 156

Thread 8 Address 160

Thread 9 Address 164 Thread 3 Address 164

Thread 10 Address 168

Thread 11 Address 172

Thread 12 Address 176 Thread 4 Address 176

Thread 13 Address 180

Thread 14 Address 184

Thread 15 Address 188 Thread 5 Address 188

Left: non-contiguous float memory access, resulting in 16 memory transactions.

Right: non-coalesced float3 memory access, resulting in 16 memory transactions.

Figure 5-3.Examples of Global Memory Access Patterns That

Are Non-Coalesced for Devices of Compute
Capability 1.0 or 1.1

86 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Address Address Address

120 120 96
Address Address Address
124 124 100
Address Address Address
Thread Thread
0 128 0 128 104

32B segment
Address Address Address
Thread Thread
1 132 1 132 108
Address Address Address
Thread Thread
2 136 2 136 112
Address Address Address
Thread Thread
3 140 3 140 116
Address Address Address
Thread Thread
4 144 4 144 120
Address Address Address
Thread Thread
5 148 5 148 124
Address Address Address
Thread Thread Thread
6 152 6 152 0 128
64B segment

Address Address Address

Thread Thread Thread
7 156 7 156 1 132
Address Address Address
Thread Thread Thread
8 160 8 160 2 136
Address Address Address

128B segment
Thread Thread Thread
9 164 9 164 3 140
Address Address Address
Thread Thread Thread
10 168 10 168 4 144
Address Address Address
Thread Thread Thread
11 172 11 172 5 148
Address Address Address
Thread Thread Thread
12 176 12 176 6 152

64B segment
Thread Address Address Address
Thread Thread
13 180 13 180 7 156

Thread Address Thread Address Address

Thread
14 184 14 184 8 160

Thread Address Thread Address Address

Thread
15 188 15 188 9 164
Address Address Address
Thread
192 192 10 168
Address Address Address
Thread
196 196 11 172
Address Address
…
Thread
200 12 176
Address Address Address
Thread
204 204 13 180
Address Address Thread Address
208 252 14 184
Address Address Thread Address
212 256 15 188

Left: random float memory access within a 64B segment, resulting in one memory transaction.
Center: misaligned float memory access, resulting in one transaction.
Right: misaligned float memory access, resulting in two transactions.

Figure 5-4.Examples of Global Memory Access by Devices

with Compute Capability 1.2 and Higher

CUDA Programming Guide Version 2.3 87

Chapter 5. Performance Guidelines

Common Access Patterns

Array of Structures
A common global memory access pattern is when each thread of thread ID tid
accesses one element of an array located at address BaseAddress of type type*
using the following address:
BaseAddress + tid
To get memory coalescing, type must meet the size and alignment requirements
discussed above. In particular, this means that if type is a structure larger than 16
bytes, it should be split into several structures that meet these requirements and the
data should be laid out in memory as a list of several arrays of these structures
instead of a single array of type type*.
Two-Dimensional Array
Another common global memory access pattern is when each thread of index
(tx,ty) accesses one element of a 2D array located at address BaseAddress of
type type* and of width width using the following address:
BaseAddress + width * ty + tx
In such a case, one gets memory coalescing for all half-warps of the thread block
only if:
‰ The width of the thread block is a multiple of half the warp size;
‰ width is a multiple of 16.
In particular, this means that an array whose width is not a multiple of 16 will be
accessed much more efficiently if it is actually allocated with a width rounded up to
the closest multiple of 16 and its rows padded accordingly. The
cudaMallocPitch() and cuMemAllocPitch() functions and associated
memory copy functions described in the reference manual enable programmers to
write non-hardware-dependent code to allocate arrays that conform to these
constraints.

5.1.2.2 Local Memory

Like the global memory space, the local memory space is not cached, so accesses to
local memory are as expensive as accesses to global memory. Local memory
accesses are always coalesced though since they are per-thread by definition.
Local memory accesses only occur for some automatic variables as mentioned in
Section B.2.5. Inspection of the PTX assembly code (obtained by compiling with
the –ptx or -keep option) will tell if a variable has been placed in local memory
during the first compilation phases as it will be declared using the .local
mnemonic and accessed using the ld.local and st.local mnemonics. If it has
not, subsequent compilation phases might still decide otherwise though if they find
it consumes too much register space for the targeted architecture. There is no way
to check this for a particular variable, but the compiler reports total local memory
usage per kernel (lmem) when compiling with the --ptxas-options=-v option.
Automatic variables that are likely to be placed in local memory are large structures
or arrays that would consume too much register space, and arrays for which the
compiler cannot determine that they are indexed with constant quantities.

88 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

5.1.2.3 Constant Memory

The constant memory space is cached so a read from constant memory costs one
memory read from device memory only on a cache miss, otherwise it just costs one
read from the constant cache.
For all threads of a half-warp, reading from the constant cache is as fast as reading
from a register as long as all threads read the same address. The cost scales linearly
with the number of different addresses read by all threads. We recommend having
all threads of the entire warp read the same address as opposed to all threads within
each of its halves only, as future devices will require it for full speed read.

5.1.2.4 Texture Memory

The texture memory space is cached so a texture fetch costs one memory read from
device memory only on a cache miss, otherwise it just costs one read from the
texture cache. The texture cache is optimized for 2D spatial locality, so threads of
the same warp that read texture addresses that are close together will achieve best
performance. Also, it is designed for streaming fetches with a constant latency, i.e. a
cache hit reduces DRAM bandwidth demand, but not fetch latency.
Reading device memory through texture fetching present some benefits that can
make it an advantageous alternative to reading device memory from global or
constant memory:
‰ If the memory reads do not follow the access patterns that global or constant
memory reads must respect to get good performance (see Sections 5.1.2.1 and
5.1.2.3), higher bandwidth can be achieved providing that there is locality in the
texture fetches;
‰ The latency of addressing calculations is hidden better, possibly improving
performance for applications that perform random accesses to the data;
‰ Packed data may be broadcast to separate variables in a single operation;
‰ 8-bit and 16-bit integer input data may be optionally converted to 32-bit floating-
point values in the range [0.0, 1.0] or [-1.0, 1.0] (see Section 3.2.4.1).
However, within the same kernel call, the texture cache is not kept coherent with
respect to global memory writes, so that any texture fetch to an address that has
been written to via a global write in the same kernel call returns undefined data. In
other words, a thread can safely read via texture some memory location only if this
memory location has been updated by a previous kernel call or memory copy, but
not if it has been previously updated by the same thread or another thread from the
same kernel call. This is only relevant when fetching from linear memory as a kernel
cannot write to CUDA arrays anyway.

5.1.2.5 Shared Memory

Because it is on-chip, the shared memory space is much faster than the local and
global memory spaces. In fact, for all threads of a warp, accessing the shared
memory is as fast as accessing a register as long as there are no bank conflicts
between the threads, as detailed below.
To achieve high memory bandwidth, shared memory is divided into equally-sized
memory modules, called banks, which can be accessed simultaneously. So, any
memory read or write request made of n addresses that fall in n distinct memory
banks can be serviced simultaneously, yielding an effective bandwidth that is n times
as high as the bandwidth of a single module.

CUDA Programming Guide Version 2.3 89

Chapter 5. Performance Guidelines

However, if two addresses of a memory request fall in the same memory bank, there
is a bank conflict and the access has to be serialized. The hardware splits a memory
request with bank conflicts into as many separate conflict-free requests as necessary,
decreasing the effective bandwidth by a factor equal to the number of separate
memory requests. If the number of separate memory requests is n, the initial
memory request is said to cause n-way bank conflicts.
To get maximum performance, it is therefore important to understand how memory
addresses map to memory banks in order to schedule the memory requests so as to
minimize bank conflicts.
In the case of the shared memory space, the banks are organized such that
successive 32-bit words are assigned to successive banks and each bank has a
bandwidth of 32 bits per two clock cycles.
For devices of compute capability 1.x, the warp size is 32 and the number of banks
is 16 (see Section 5.1); a shared memory request for a warp is split into one request
for the first half of the warp and one request for the second half of the warp. As a
consequence, there can be no bank conflict between a thread belonging to the first
half of a warp and a thread belonging to the second half of the same warp.
A common case is for each thread to access a 32-bit word from an array indexed by
the thread ID tid and with some stride s:
__shared__ float shared[32];
float data = shared[BaseIndex + s * tid];
In this case, the threads tid and tid+n access the same bank whenever s*n is a
multiple of the number of banks m or equivalently, whenever n is a multiple of m/d
where d is the greatest common divisor of m and s. As a consequence, there will be
no bank conflict only if half the warp size is less than or equal to m/d. For devices
of compute capability 1.x, this translates to no bank conflict only if d is equal to 1,
or in other words, only if s is odd since m is a power of two.
Figure 5-5 and Figure 5-6 show some examples of conflict-free memory accesses
while Figure 5-7 shows some examples of memory accesses that cause bank
conflicts.
Other cases worth mentioning are when each thread accesses an element that is
smaller or larger than 32 bits in size. For example, there are bank conflicts if an array
of char is accessed the following way:
__shared__ char shared[32];
char data = shared[BaseIndex + tid];
because shared[0], shared[1], shared[2], and shared[3], for example,
belong to the same bank. There are no bank conflicts however, if the same array is
accessed the following way:
char data = shared[BaseIndex + 4 * tid];
There are also 2-way bank conflicts for arrays of double:
__shared__ double shared[32];
double data = shared[BaseIndex + tid];
since the memory request is compiled into two separate 32-bit requests. One way to
avoid bank conflicts in this case is two split the double operands like in the
following sample code:
__shared__ int shared_lo[32];

90 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

__shared__ int shared_hi[32];

double dataIn;
shared_lo[BaseIndex + tid] = __double2loint(dataIn);
shared_hi[BaseIndex + tid] = __double2hiint(dataIn);

double dataOut =
__hiloint2double(shared_hi[BaseIndex + tid],
shared_lo[BaseIndex + tid]);
It might not always improve performance though and will perform worse on future
architectures.
A structure assignment is compiled into as many memory requests as necessary for
each member in the structure, so the following code, for example:
__shared__ struct type shared[32];
struct type data = shared[BaseIndex + tid];
results in:
‰ Three separate memory reads without bank conflicts if type is defined as
struct type {
float x, y, z;
};
since each member is accessed with a stride of three 32-bit words;
‰ Two separate memory reads with bank conflicts if type is defined as
struct type {
float x, y;
};
since each member is accessed with a stride of two 32-bit words;
‰ Two separate memory reads with bank conflicts if type is defined as
struct type {
float f;
char c;
};
since each member is accessed with a stride of five bytes.
Finally, shared memory also features a broadcast mechanism whereby a 32-bit word
can be read and broadcast to several threads simultaneously when servicing one
memory read request. This reduces the number of bank conflicts when several
threads of a half-warp read from an address within the same 32-bit word. More
precisely, a memory read request made of several addresses is serviced in several
steps over time – one step every two clock cycles – by servicing one conflict-free
subset of these addresses per step until all addresses have been serviced; at each
step, the subset is built from the remaining addresses that have yet to be serviced
using the following procedure:
‰ Select one of the words pointed to by the remaining addresses as the broadcast
word,
‰ Include in the subset:
‰ All addresses that are within the broadcast word,
‰ One address for each bank pointed to by the remaining addresses.

CUDA Programming Guide Version 2.3 91

Chapter 5. Performance Guidelines

Which word is selected as the broadcast word and which address is picked up for
each bank at each cycle are unspecified.
A common conflict-free case is when all threads of a half-warp read from an address
within the same 32-bit word.
Figure 5-8 shows some examples of memory read accesses that involve the
broadcast mechanism.

92 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Thread 0 Bank 0 Thread 0 Bank 0

Thread 1 Bank 1 Thread 1 Bank 1

Thread 2 Bank 2 Thread 2 Bank 2

Thread 3 Bank 3 Thread 3 Bank 3

Thread 4 Bank 4 Thread 4 Bank 4

Thread 5 Bank 5 Thread 5 Bank 5

Thread 6 Bank 6 Thread 6 Bank 6

Thread 7 Bank 7 Thread 7 Bank 7

Thread 8 Bank 8 Thread 8 Bank 8

Thread 9 Bank 9 Thread 9 Bank 9

Thread 10 Bank 10 Thread 10 Bank 10

Thread 11 Bank 11 Thread 11 Bank 11

Thread 12 Bank 12 Thread 12 Bank 12

Thread 13 Bank 13 Thread 13 Bank 13

Thread 14 Bank 14 Thread 14 Bank 14

Thread 15 Bank 15 Thread 15 Bank 15

Left: linear addressing with a stride of one 32-bit word.

Right: random permutation.

Figure 5-5.Examples of Shared Memory Access Patterns

without Bank Conflicts

CUDA Programming Guide Version 2.3 93

Chapter 5. Performance Guidelines

Thread 0 Bank 0

Thread 1 Bank 1

Thread 2 Bank 2

Thread 3 Bank 3

Thread 4 Bank 4

Thread 5 Bank 5

Thread 6 Bank 6

Thread 7 Bank 7

Thread 8 Bank 8

Thread 9 Bank 9

Thread 10 Bank 10

Thread 11 Bank 11

Thread 12 Bank 12

Thread 13 Bank 13

Thread 14 Bank 14

Thread 15 Bank 15

Linear addressing with a stride of three 32-bit words.

Figure 5-6.Example of a Shared Memory Access Pattern

without Bank Conflicts

94 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

Thread 0 Bank 0 Thread 0 Bank 0

Thread 1 Bank 1 Thread 1 Bank 1

Thread 2 Bank 2 Thread 2 Bank 2

Thread 3 Bank 3 Thread 3 Bank 3

Thread 4 Bank 4 Thread 4 Bank 4

Thread 5 Bank 5 Thread 5 Bank 5

Thread 6 Bank 6 Thread 6 Bank 6

Thread 7 Bank 7 Thread 7 Bank 7

Thread 8 Bank 8 Thread 8 Bank 8

Thread 9 Bank 9 Thread 9 Bank 9

Thread 10 Bank 10 Thread 10 Bank 10

Thread 11 Bank 11 Thread 11 Bank 11

Thread 12 Bank 12 Thread 12 Bank 12

Thread 13 Bank 13 Thread 13 Bank 13

Thread 14 Bank 14 Thread 14 Bank 14

Thread 15 Bank 15 Thread 15 Bank 15

Left: Linear addressing with a stride of two 32-bit words causes 2-way bank conflicts.
Right: Linear addressing with a stride of eight 32-bit words causes 8-way bank conflicts.

Figure 5-7.Examples of Shared Memory Access Patterns with

Bank Conflicts

CUDA Programming Guide Version 2.3 95

Chapter 5. Performance Guidelines

Thread 0 Bank 0 Thread 0 Bank 0

Thread 1 Bank 1 Thread 1 Bank 1

Thread 2 Bank 2 Thread 2 Bank 2

Thread 3 Bank 3 Thread 3 Bank 3

Thread 4 Bank 4 Thread 4 Bank 4

Thread 5 Bank 5 Thread 5 Bank 5

Thread 6 Bank 6 Thread 6 Bank 6

Thread 7 Bank 7 Thread 7 Bank 7

Thread 8 Bank 8 Thread 8 Bank 8

Thread 9 Bank 9 Thread 9 Bank 9

Thread 10 Bank 10 Thread 10 Bank 10

Thread 11 Bank 11 Thread 11 Bank 11

Thread 12 Bank 12 Thread 12 Bank 12

Thread 13 Bank 13 Thread 13 Bank 13

Thread 14 Bank 14 Thread 14 Bank 14

Thread 15 Bank 15 Thread 15 Bank 15

Left: This access pattern is conflict-free since all threads read from an address within the same 32-bit
word.
Right: This access pattern causes either no bank conflicts if the word from bank 5 is chosen as the
broadcast word during the first step or 2-way bank conflicts, otherwise.

Figure 5-8.Example of Shared Memory Read Access Patterns

with Broadcast

96 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

5.1.2.6 Registers
Generally, accessing a register is zero extra clock cycles per instruction, but delays
may occur due to register read-after-write dependencies and register memory bank
conflicts.
The delays introduced by read-after-write dependencies can be ignored as soon as
there are at least 192 active threads per multiprocessor to hide them.
The compiler and thread scheduler schedule the instructions as optimally as possible
to avoid register memory bank conflicts. They achieve best results when the number
of threads per block is a multiple of 64. Other than following this rule, an
application has no direct control over these bank conflicts. In particular, there is no
need to pack data into float4 or int4 types.

5.2 Execution Configuration

How the execution configuration affects the execution time of a kernel launch
generally depends on the kernel code. Experimentation is therefore recommended.
There are however general guidelines, described in this section.
For a start, the kernel will simply fail to launch if the number of threads per block
either is above the maximum number of threads per block as specified in
Appendix A, or requires too many registers or shared memory than available per
multiprocessor as mentioned in Section 4.1. The total number of registers required
for a block is equal to
R max
ceil ( R × ceil (T ,32), )
32
where R is the number of registers required for the kernel, R max is the number of
registers per multiprocessor given in Appendix A, T is the number of threads per
block, and ceil ( x, y ) is equal to x rounded up to the nearest multiple of y . The total
amount of shared memory required for a block is equal to the sum of the amount of
statically allocated shared memory, the amount of dynamically allocated shared
memory, and the amount of shared memory used to pass the kernel’s arguments.
The number of registers a kernel compiles to and the local, shared, and constant
memory usages are reported by the compiler when compiling with the
--ptxas-options=-v option. Note that each double or long long variable
uses two registers. However, devices of compute capability 1.2 and higher have
twice as many registers per multiprocessor as devices with lower compute capability.
Then, given a total number of threads per grid, the number of threads per block
might be dictated by the need to have enough blocks in the grid to maximize the
utilization of the available computing resources. First, there should be at least as
many blocks as there are multiprocessors in the device. Then, running only one
block per multiprocessor will force the multiprocessor to idle during thread
synchronization and also during device memory reads if there are not enough
threads per block to cover the load latency. It is therefore usually better to allow for
two or more blocks to be active on each multiprocessor to allow overlap between
blocks that wait and blocks that can run. For this to happen, not only should there
be at least twice as many blocks as there are multiprocessors in the device, but also
the amount of registers and shared memory required per block must be low enough

CUDA Programming Guide Version 2.3 97

Chapter 5. Performance Guidelines

to allow for more than one active block (see Section 4.1). More thread blocks stream
in pipeline fashion through the device and amortize overhead even more. The
number of blocks per grid should be at least 100 if one wants it to scale to future
devices; 1000 blocks will scale across several generations.
With a high enough number of blocks, the number of threads per block should be
chosen as a multiple of the warp size to avoid wasting computing resources with
under-populated warps, or better, a multiple of 64 for the reason invoked in
Section 5.1.2.6. Allocating more threads per block is better for efficient time slicing,
but the more threads per block, the fewer registers are available per thread, which
might prevent the kernel invocation from succeeding.
Usually, 64 threads per block is minimal and makes sense only if there are multiple
active blocks per multiprocessor; 192 or 256 threads per block is better and usually
allows for enough registers to compile.
The ratio of the number of active warps per multiprocessor to the maximum
number of active warps (given in Appendix A) is called the multiprocessor occupancy.
In order to maximize occupancy, the compiler attempts to minimize register usage
while keeping the number of instructions and local memory usage to a minimum.
This can be controlled using the - maxrregcount compiler option. The CUDA
Software Development Kit provides a spreadsheet to assist programmers in
choosing thread block size based on shared memory and register requirements.

5.3 Data Transfer between Host and Device

The bandwidth between device memory and the device is much higher than the
bandwidth between device memory and host memory. Therefore, one should strive
to minimize data transfer between the host and the device, for example, by moving
more code from the host to the device, even if that means running kernels with low
parallelism computations. Intermediate data structures may be created in device
memory, operated on by the device, and destroyed without ever being mapped by
the host or copied to host memory.
Also, because of the overhead associated with each transfer, batching many small
transfers into a big one always performs much better than making each transfer
separately.
Finally, higher performance for data transfers between host and device is achieved
by using page-locked host memory as described in Section 3.2.5.
In addition, when using mapped page-locked memory (Section 3.2.5.3), there is no
need to allocate any device memory and to explicitly copy data between device and
host memory. Data transfers are implicitly performed each time the kernel accesses
the mapped memory. For maximum performance, these memory accesses must be
coalesced like if they were accesses to global memory (see Section 5.1.2.1).
Assuming that they are and that the mapped memory is read or written only once,
using mapped page-locked memory instead of explicit copies between device and
host memory can be a win performance-wise.
On integrated systems where device memory and host memory are physically the
same, any copy between host and device memory is superfluous and mapped page-
locked memory should be used instead. Applications may query whether a device is

98 CUDA Programming Guide Version 2.3

 Chapter 5. Performance Guidelines

integrated or not by calling cudaGetDeviceProperties() and checking the

integrated property or checking the CU_DEVICE_ATTRIBUTE_INTEGRATED
attribute using cuDeviceGetAttribute().

5.4 Warp-Level Synchronization

Because a warp executes one common instruction at a time, threads within a warp
are implicitly synchronized and this can be used to omit __syncthreads() for
better performance.
In the following code sample, for example, both calls to __syncthreads() are
required to get the expected result (i.e. result[i] = 2 * myArray[i] for
i > 0). Without synchronization, any of the two references to myArray[tid]
could return either 2 or the value initially stored in myArray, depending on whether
the memory read occurs before or after the memory write from
myArray[tid + 1] = 2.
// myArray is an array of integers located in global or shared
// memory
__global__ void myKernel(int* result) {
int tid = threadIdx.x;
...
int ref1 = myArray[tid] * 1;
__syncthreads();
myArray[tid + 1] = 2;
__syncthreads();
int ref2 = myArray[tid] * 1;
result[tid] = ref1 * ref2;
...
}
However, in the following slightly modified code sample, threads are guaranteed to
belong to the same warp, so that there is no need for any __syncthreads().
// myArray is an array of integers located in global or shared
// memory
__global__ void myKernel(int* result) {
int tid = threadIdx.x;
...
if (tid < warpSize) {
int ref1 = myArray[tid] * 1;
myArray[tid + 1] = 2;
int ref2 = myArray[tid] * 1;
result[tid] = ref1 * ref2;
}
...
}
Simply removing the __syncthreads() is not enough however; myArray also
needs to be declared as volatile as described in Section B.2.4.

5.5 Overall Performance Optimization Strategies

Performance optimization revolves around three basic strategies:

CUDA Programming Guide Version 2.3 99

Chapter 5. Performance Guidelines

‰ Maximizing parallel execution;

‰ Optimizing memory usage to achieve maximum memory bandwidth;
‰ Optimizing instruction usage to achieve maximum instruction throughput.
Maximizing parallel execution starts with structuring the algorithm in a way that
exposes as much data parallelism as possible. At points in the algorithm where
parallelism is broken because some threads need to synchronize in order to share
data between each other, there are two cases: Either these threads belong to the
same block, in which case they should use __syncthreads() and share data
through shared memory within the same kernel call, or they belong to different
blocks, in which case they must share data through global memory using two
separate kernel invocations, one for writing to and one for reading from global
memory.
Once the parallelism of the algorithm has been exposed it needs to be mapped to
the hardware as efficiently as possible. This is done by carefully choosing the
execution configuration of each kernel invocation as detailed in Section 5.2.
The application should also maximize parallel execution at a higher level by
explicitly exposing concurrent execution on the device through streams, as
described in Section 3.2.6.1, as well as maximizing concurrent execution between
host and device.
Optimizing memory usage starts with minimizing data transfers with low-
bandwidth. That means minimizing data transfers between the host and the device,
as detailed in Section 5.3, since these have much lower bandwidth than data
transfers between the device and global memory. That also means minimizing data
transfers between the device and global memory by maximizing use of shared
memory on the device, as mentioned in Section 5.1.2. Sometimes, the best
optimization might even be to avoid any data transfer in the first place by simply
recomputing the data instead whenever it is needed.
As detailed in Sections 5.1.2.1, 5.1.2.3, 5.1.2.4, and 5.1.2.5, the effective bandwidth
can vary by an order of magnitude depending on access pattern for each type of
memory. The next step in optimizing memory usage is therefore to organize
memory accesses as optimally as possible based on the optimal memory access
patterns. This optimization is especially important for global memory accesses as
global memory bandwidth is low and its latency is hundreds of clock cycles (see
Section 5.1.1.3). Shared memory accesses, on the other hand, are usually worth
optimizing only in case they have a high degree of bank conflicts.
As for optimizing instruction usage, the use of arithmetic instructions with low
throughput (see Section 5.1.1.1) should be minimized. This includes trading
precision for speed when it does not affect the end result, such as using intrinsic
instead of regular functions (intrinsic functions are listed in Section C.2) or single-
precision instead of double-precision. Particular attention must be paid to control
flow instructions due to the SIMT nature of the device as detailed in Section 5.1.1.2.

100 CUDA Programming Guide Version 2.3

 Appendix A.
Technical Specifications

A.1 General Specifications

The general specifications and features of a compute device depend on its compute
capability (see Section 2.5).
The following sections describe the technical specifications and features associated
to each compute capability. The specifications for a given compute capability are the
same as for the compute capability just below unless otherwise mentioned. Similarly,
any feature supported for a given compute capability is supported for any higher
compute capability.
The compute capability and number of multiprocessors of all CUDA-enabled
devices are given in the following table:

Number of Compute
Multiprocessors Capability
(1 Multiprocessor
= 8 Processors)
GeForce GTX 295 2x30 1.3
GeForce GTX 285, GTX 280 30 1.3
GeForce GTX 260 24 1.3
GeForce 9800 GX2 2x16 1.1
GeForce GTS 250, GTS 150, 9800 GTX, 16 1.1
9800 GTX+, 8800 GTS 512
GeForce 8800 Ultra, 8800 GTX 16 1.0
GeForce 9800 GT, 8800 GT, GTX 280M, 14 1.1
9800M GTX
GeForce GT 130, 9600 GSO, 8800 GS, 12 1.1
8800M GTX, GTX 260M, 9800M GT
GeForce 8800 GTS 12 1.0
GeForce 9600 GT, 8800M GTS, 9800M GTS 8 1.1
GeForce 9700M GT 6 1.1
GeForce GT 120, 9500 GT, 8600 GTS, 8600 GT, 4 1.1
9700M GT, 9650M GS, 9600M GT, 9600M GS,
9500M GS, 8700M GT, 8600M GT, 8600M GS
GeForce G100, 8500 GT, 8400 GS, 8400M GT, 2 1.1

CUDA Programming Guide Version 2.3 101

Appendix A. Technical Specifications

9500M G, 9300M G, 8400M GS, 9400 mGPU,

9300 mGPU, 8300 mGPU, 8200 mGPU,
8100 mGPU
GeForce 9300M GS, 9200M GS, 9100M G, 1 1.1
8400M G
Tesla S1070 4x30 1.3
Tesla C1060 30 1.3
Tesla S870 4x16 1.0
Tesla D870 2x16 1.0
Tesla C870 16 1.0
Quadro Plex 2200 D2 2x30 1.3
Quadro Plex 2100 D4 4x14 1.1
Quadro Plex 2100 Model S4 4x16 1.0
Quadro Plex 1000 Model IV 2x16 1.0
Quadro FX 5800 30 1.3
Quadro FX 4800 24 1.3
Quadro FX 4700 X2 2x14 1.1
Quadro FX 3700M 16 1.1
Quadro FX 5600 16 1.0
Quadro FX 3700 14 1.1
Quadro FX 3600M 12 1.1
Quadro FX 4600 12 1.0
Quadro FX 2700M 6 1.1
Quadro FX 1700, FX 570, NVS 320M, FX 1700M, 4 1.1
FX 1600M, FX 770M, FX 570M
Quadro FX 370, NVS 290, NVS 140M, NVS 135M, 2 1.1
FX 360M
Quadro FX 370M, NVS 130M 1 1.1

The number of multiprocessors, the clock frequency and the total amount of device
memory can be queried using the runtime (see reference manual).

A.1.1 Specifications for Compute Capability 1.0

‰ The maximum number of threads per block is 512;
‰ The maximum sizes of the x-, y-, and z-dimension of a thread block are 512, 512,
and 64, respectively;
‰ The maximum size of each dimension of a grid of thread blocks is 65535;
‰ The warp size is 32 threads;
‰ The number of registers per multiprocessor is 8192;
‰ The amount of shared memory available per multiprocessor is 16 KB organized
into 16 banks (see Section 5.1.2.5);
‰ The total amount of constant memory is 64 KB;
‰ The total amount of local memory per thread is 16 KB;
‰ The cache working set for constant memory is 8 KB per multiprocessor;

102 CUDA Programming Guide Version 2.3

 Appendix A. Technical Specifications

‰ The cache working set for texture memory varies between 6 and 8 KB per
multiprocessor;
‰ The maximum number of active blocks per multiprocessor is 8;
‰ The maximum number of active warps per multiprocessor is 24;
‰ The maximum number of active threads per multiprocessor is 768;
‰ For a one-dimensional texture reference bound to a CUDA array, the maximum
width is 213;
‰ For a one-dimensional texture reference bound to linear memory, the maximum
width is 227;
‰ For a two-dimensional texture reference bound to linear memory or a CUDA
array, the maximum width is 216 and the maximum height is 215;
‰ For a three-dimensional texture reference bound to a CUDA array, the
maximum width is 211, the maximum height is 211, and the maximum depth is
211;
‰ The limit on kernel size is 2 million PTX instructions;

A.1.2 Specifications for Compute Capability 1.1

‰ Support for atomic functions operating on 32-bit words in global memory (see
Section B.10).

A.1.3 Specifications for Compute Capability 1.2

‰ Support for atomic functions operating in shared memory and atomic functions
operating on 64-bit words in global memory (see Section B.10);
‰ Support for warp vote functions (see Section B.11);
‰ The number of registers per multiprocessor is 16384;
‰ The maximum number of active warps per multiprocessor is 32;
‰ The maximum number of active threads per multiprocessor is 1024.

A.1.4 Specifications for Compute Capability 1.3

‰ Support for double-precision floating-point numbers.

A.2 Floating-Point Standard

All compute devices follow the IEEE-754 standard for binary floating-point
arithmetic with the following deviations:
‰ There is no dynamically configurable rounding mode; however, most of the
operations support IEEE rounding modes, exposed via device functions;
‰ There is no mechanism for detecting that a floating-point exception has occurred
and all operations behave as if the IEEE-754 exceptions are always masked, and
deliver the masked response as defined by IEEE-754 if there is an exceptional
event; for the same reason, while SNaN encodings are supported, they are not
signaling;

CUDA Programming Guide Version 2.3 103

Appendix A. Technical Specifications

‰ Absolute value and negation are not compliant with IEEE-754 with respect to
NaNs; these are passed through unchanged;
‰ For single-precision floating-point numbers only:
‰ Denormalized numbers are not supported; floating-point arithmetic and
comparison instructions convert denormalized operands to zero prior to the
floating-point operation;
‰ Underflowed results are flushed to zero;
‰ The result of an operation involving one or more input NaNs is the quiet
NaN of bit pattern 0x7fffffff; note that;
‰ Some instructions are not IEEE-compliant:
‰ Addition and multiplication are often combined into a single multiply-add
instruction (FMAD), which truncates the intermediate result of the
multiplication;
‰ Division is implemented via the reciprocal in a non-standard-compliant
way;
‰ Square root is implemented via the reciprocal square root in a non-
standard-compliant way;
‰ For addition and multiplication, only round-to-nearest-even and
round-towards-zero are supported via static rounding modes; directed
rounding towards +/- infinity is not supported;
But, IEEE-compliant software (and therefore slower) implementations are
provided through the following intrinsics from Section C.2.1:
‰ __fmaf_r{n,z,u,d}(float, float, float): single-precision
fused multiply-add with IEEE rounding modes,
‰ __frcp_r[n,z,u,d](float): single-precision reciprocal with IEEE
rounding modes,
‰ __fdiv_r[n,z,u,d](float, float): single-precision division with
IEEE rounding modes,
‰ __fsqrt_r[n,z,u,d](float): single-precision square root with
IEEE rounding modes,
‰ __fadd_r[u,d](float, float): single-precision addition with
IEEE directed rounding,
‰ __fmul_r[u,d](float, float): single-precision multiplication with
IEEE directed rounding;
‰ For double-precision floating-point numbers only:
‰ Round-to-nearest-even is the only supported IEEE rounding mode for
reciprocal, division, and square root.
In accordance to the IEEE-754R standard, if one of the input parameters to
fminf(), fmin(), fmaxf(), or fmax() is NaN, but not the other, the result is
the non-NaN parameter.
The conversion of a floating-point value to an integer value in the case where the
floating-point value falls outside the range of the integer format is left undefined by
IEEE-754. For compute devices, the behavior is to clamp to the end of the
supported range. This is unlike the x86 architecture behaves.

104 CUDA Programming Guide Version 2.3

 Appendix B.
C Extensions

B.1 Function Type Qualifiers

Function type qualifiers specify whether a function executes on the host or on the
device and whether it is callable from the host or from the device.

B.1.1 __device__
The __device__ qualifier declares a function that is:
‰ Executed on the device
‰ Callable from the device only.

B.1.2 __global__
The __global__ qualifier declares a function as being a kernel. Such a function is:
‰ Executed on the device,
‰ Callable from the host only.

B.1.3 __host__
The __host__ qualifier declares a function that is:
‰ Executed on the host,
‰ Callable from the host only.
It is equivalent to declare a function with only the __host__ qualifier or to declare
it without any of the __host__, __device__, or __global__ qualifier; in either
case the function is compiled for the host only.
However, the __host__ qualifier can also be used in combination with the
__device__ qualifier, in which case the function is compiled for both the host and
the device.

CUDA Programming Guide Version 2.3 105

Appendix B. C Extensions

B.1.4 Restrictions
__device__ and __global__ functions do not support recursion.
__device__ and __global__ functions cannot declare static variables inside
their body.
__device__ and __global__ functions cannot have a variable number of
arguments.
__device__ functions cannot have their address taken; function pointers to
__global__ functions, on the other hand, are supported.
The __global__ and __host__ qualifiers cannot be used together.
__global__ functions must have void return type.
Any call to a __global__ function must specify its execution configuration as
described in Section B.12.
A call to a __global__ function is asynchronous, meaning it returns before the
device has completed its execution.
__global__ function parameters are currently passed via shared memory to the
device and limited to 256 bytes.

B.2 Variable Type Qualifiers

Variable type qualifiers specify the memory location on the device of a variable.

B.2.1 __device__
The __device__ qualifier declares a variable that resides on the device.
At most one of the other type qualifiers defined in the next three sections may be
used together with __device__ to further specify which memory space the
variable belongs to. If none of them is present, the variable:
‰ Resides in global memory space,
‰ Has the lifetime of an application,
‰ Is accessible from all the threads within the grid and from the host through the
runtime library.

B.2.2 __constant__
The __constant__ qualifier, optionally used together with __device__,
declares a variable that:
‰ Resides in constant memory space,
‰ Has the lifetime of an application,
‰ Is accessible from all the threads within the grid and from the host through the
runtime library.

106 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

B.2.3 __shared__
The __shared__ qualifier, optionally used together with __device__, declares a
variable that:
‰ Resides in the shared memory space of a thread block,
‰ Has the lifetime of the block,
‰ Is only accessible from all the threads within the block.
When declaring a variable in shared memory as an external array such as
extern __shared__ float shared[];
the size of the array is determined at launch time (see Section B.12). All variables
declared in this fashion, start at the same address in memory, so that the layout of
the variables in the array must be explicitly managed through offsets. For example, if
one wants the equivalent of
short array0[128];
float array1[64];
int array2[256];
in dynamically allocated shared memory, one could declare and initialize the arrays
the following way:
extern __shared__ char array[];
__device__ void func() // __device__ or __global__ function
{
short* array0 = (short*)array;
float* array1 = (float*)&array0[128];
int* array2 = (int*)&array1[64];
}

B.2.4 Volatile
Only after the execution of a __threadfence_block(), __threadfence(),
or __syncthreads() (Sections B.5 and B.6) are prior writes to global or shared
memory guaranteed to be visible by other threads. As long as this requirement is
met, the compiler is free to optimize reads and writes to global or shared memory.
For example, in the code sample below, the first reference to myArray[tid]
compiles into a global or shared memory read instruction, but the second reference
does not as the compiler simply reuses the result of the first read.
// myArray is an array of non-zero integers
// located in global or shared memory
__global__ void myKernel(int* result) {
int tid = threadIdx.x;
int ref1 = myArray[tid] * 1;
myArray[tid + 1] = 2;
int ref2 = myArray[tid] * 1;
result[tid] = ref1 * ref2;
}
Therefore, ref2 cannot possibly be equal to 2 in thread tid as a result of thread
tid-1 overwriting myArray[tid] by 2.
This behavior can be changed using the volatile keyword: If a variable located in
global or shared memory is declared as volatile, the compiler assumes that its value

CUDA Programming Guide Version 2.3 107

Appendix B. C Extensions

can be changed at any time by another thread and therefore any reference to this
variable compiles to an actual memory read instruction.
Note that even if myArray is declared as volatile in the code sample above, there is
no guarantee, in general, that ref2 will be equal to 2 in thread tid since thread
tid might read myArray[tid] into ref2 before thread tid-1 overwrites its
value by 2. Synchronization is required as mentioned in Section 5.4.

B.2.5 Restrictions
These qualifiers are not allowed on struct and union members, on formal
parameters and on local variables within a function that executes on the host.
__shared__ and __constant__ variables have implied static storage.
__device__, __shared__ and __constant__ variables cannot be defined as
external using the extern keyword. The only exception is for dynamically allocated
__shared__ variables as described in Section B.2.3.
__device__ and __constant__ variables are only allowed at file scope.
__constant__ variables cannot be assigned to from the device, only from the
host through host runtime functions (Sections 3.2.1 and 3.3.4).
__shared__ variables cannot have an initialization as part of their declaration.
An automatic variable declared in device code without any of these qualifiers
generally resides in a register. However in some cases the compiler might choose to
place it in local memory, which can have adverse performance consequences as
detailed in Section 5.1.2.2.
Pointers in code that is executed on the device are supported as long as the compiler
is able to resolve whether they point to either the shared memory space or the
global memory space, otherwise they are restricted to only point to memory
allocated or declared in the global memory space.
Dereferencing a pointer either to global or shared memory in code that is executed
on the host or to host memory in code that is executed on the device results in an
undefined behavior, most often in a segmentation fault and application termination.
The address obtained by taking the address of a __device__, __shared__ or
__constant__ variable can only be used in device code. The address of a
__device__ or __constant__ variable obtained through
cudaGetSymbolAddress() as described in Section 3.3.4 can only be used in
host code.

B.3 Built-in Vector Types

B.3.1 char1, uchar1, char2, uchar2, char3, uchar3,

char4, uchar4, short1, ushort1, short2, ushort2,
short3, ushort3, short4, ushort4, int1, uint1, int2,
uint2, int3, uint3, int4, uint4, long1, ulong1, long2,

108 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

ulong2, long3, ulong3, long4, ulong4, longlong1,

longlong2, float1, float2, float3, float4, double1,
double2
These are vector types derived from the basic integer and floating-point types. They
are structures and the 1st, 2nd, 3rd, and 4th components are accessible through the
fields x, y, z, and w, respectively. They all come with a constructor function of the
form make_<type name>; for example,
int2 make_int2(int x, int y);
which creates a vector of type int2 with value (x, y).
In host code, the alignment requirement of a vector type is equal to the alignment
requirement of its base type. This is not always the case in device code as detailed in
Table B-1.

Table B-1. Alignment Requirements in Device Code

Type Alignment
char1, uchar1 1
char2, uchar2 2
char3, uchar3 1
char4, uchar4 4
short1, ushort1 2
short2, ushort2 4
short3, ushort3 2
short4, ushort4 8
int1, uint1 4
int2, uint2 8
int3, uint3 4
int4, uint4 16
long1, ulong1 Same as int1 or longlong1 depending on platform
long2, ulong2 Same as int2 or longlong2 depending on platform
long3, ulong3 Same as int3 depending on platform
long4, ulong4 Same as int4 depending on platform
longlong1 8
longlong2 16
float1 4
float2 8
float3 4
float4 16
double1 8
double2 16

CUDA Programming Guide Version 2.3 109

Appendix B. C Extensions

B.3.2 dim3
This type is an integer vector type based on uint3 that is used to specify
dimensions. When defining a variable of type dim3, any component left unspecified
is initialized to 1.

B.4 Built-in Variables

Built-in variables specify the grid and block dimensions and the block and thread
indices. They are only valid within functions that are executed on the device.

B.4.1 gridDim
This variable is of type dim3 (see Section B.3.2) and contains the dimensions of the
grid.

B.4.2 blockIdx
This variable is of type uint3 (see Section B.3.1) and contains the block index
within the grid.

B.4.3 blockDim
This variable is of type dim3 (see Section B.3.2) and contains the dimensions of the
block.

B.4.4 threadIdx
This variable is of type uint3 (see Section B.3.1) and contains the thread index
within the block.

B.4.5 warpSize
This variable is of type int and contains the warp size in threads (see Section 4.1
for the definition of a warp).

B.4.6 Restrictions
‰ It is not allowed to take the address of any of the built-in variables.
‰ It is not allowed to assign values to any of the built-in variables.

B.5 Memory Fence Functions

void __threadfence();

110 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

waits until all global and shared memory accesses made by the calling thread prior to
__threadfence() are visible to all threads in the device for global memory
accesses and all threads in the thread block for shared memory accesses.
void __threadfence_block();
waits until all global and shared memory accesses made by the calling thread prior to
__threadfence_block() are visible to all threads in the thread block.
In general, when a thread issues a series of writes to memory in a particular order,
other threads may see the effects of these memory writes in a different order.
__threadfence() and __threadfence_block() can be used to enforce
some ordering.
One use case is when threads consume some data produced by other threads as
illustrated by the following code sample of a kernel that computes the sum of an
array of N numbers in one call. Each block first sums a subset of the array and
stores the result in global memory. When all blocks are done, the last block done
reads each of these partial sums from global memory and sums them to obtain the
final result. In order to determine which block is finished last, each block atomically
increments a counter to signal that it is done with computing and storing its partial
sum (see Section B.10 about atomic functions). The last block is the one that
receives the counter value equal to gridDim.x-1. If no fence is placed between
storing the partial sum and incrementing the counter, the counter might increment
before the partial sum is stored and therefore, might reach gridDim.x-1 and let
the last block start reading partial sums before they have been actually updated in
memory.
__device__ unsigned int count = 0;
__shared__ bool isLastBlockDone;
__global__ void sum(const float* array, unsigned int N,
float* result)
{
// Each block sums a subset of the input array
float partialSum = calculatePartialSum(array, N);

if (threadIdx.x == 0) {

// Thread 0 of each block stores the partial sum

// to global memory
result[blockIdx.x] = partialSum;

// Thread 0 makes sure its result is visible to

// all other threads
__threadfence();

// Thread 0 of each block signals that it is done

unsigned int value = atomicInc(&count, gridDim.x);

// Thread 0 of each block determines if its block is

// the last block to be done
isLastBlockDone = (value == (gridDim.x - 1));
}

// Synchronize to make sure that each thread reads

// the correct value of isLastBlockDone
__syncthreads();

CUDA Programming Guide Version 2.3 111

Appendix B. C Extensions

if (isLastBlockDone) {

// The last block sums the partial sums

// stored in result[0 .. gridDim.x-1]
float totalSum = calculateTotalSum(result);

if (threadIdx.x == 0) {

// Thread 0 of last block stores total sum

// to global memory and resets count so that
// next kernel call works properly
result[0] = totalSum;
count = 0;
}
}
}

B.6 Synchronization Function

void __syncthreads();
waits until all threads in the thread block have reached this point and all global and
shared memory accesses made by these threads prior to __syncthreads() are
visible to all threads in the block.
__syncthreads() is used to coordinate communication between the threads of
the same block. When some threads within a block access the same addresses in
shared or global memory, there are potential read-after-write, write-after-read, or
write-after-write hazards for some of these memory accesses. These data hazards
can be avoided by synchronizing threads in-between these accesses.
__syncthreads() is allowed in conditional code but only if the conditional
evaluates identically across the entire thread block, otherwise the code execution is
likely to hang or produce unintended side effects.

B.7 Mathematical Functions

Section C.1 contains a comprehensive list of the C/C++ standard library
mathematical functions that are currently supported in device code, along with their
respective error bounds. When executed in host code, a given function uses the C
runtime implementation if available.
For some of the functions of Section C.1, a less accurate, but faster version exists in
the device runtime component; it has the same name prefixed with __ (such as
__sinf(x)). These intrinsic functions are listed in Section C.2, along with their
respective error bounds.
The compiler has an option (-use_fast_math) that forces each function in Table
B-2 to compile to its intrinsic counterpart. In addition to reduce accuracy of the
affected functions, it may also cause some differences in special case handling. A
more robust approach is to selectively replace mathematical function calls by calls to
intrinsic functions only where it is merited by the performance gains and where

112 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

changed properties such as reduced accuracy and different special case handling can
be tolerated.

Table B-2. Functions Affected by –use_fast_math

Type Alignment
x/y __fdividef(x,y)
sinf(x) __sinf(x)
cosf(x) __cosf(x)
tanf(x) __tanf(x)
sincosf(x,sptr,cptr) __sincosf(x,sptr,cptr)
logf(x) __logf(x)
log2f(x) __log2f(x)
log10f(x) __log10f(x)
expf(x) __expf(x)
exp10f(x) __exp10f(x)
powf(x,y) __powf(x,y)

B.8 Texture Functions

For texture functions, a combination of the texture reference’s immutable (compile-
time) and mutable (runtime) attributes determine how the texture coordinates are
interpreted, what processing occurs during the texture fetch, and the return value
delivered by the texture fetch (see Sections 3.2.4.1 and 3.2.4.2).

B.8.1 tex1Dfetch()
template<class Type>
Type tex1Dfetch(
texture<Type, 1, cudaReadModeElementType> texRef,
int x);

float tex1Dfetch(
texture<unsigned char, 1, cudaReadModeNormalizedFloat> texRef,
int x);

float tex1Dfetch(
texture<signed char, 1, cudaReadModeNormalizedFloat> texRef,
int x);

float tex1Dfetch(
texture<unsigned short, 1, cudaReadModeNormalizedFloat> texRef,
int x);

float tex1Dfetch(
texture<signed short, 1, cudaReadModeNormalizedFloat> texRef,
int x);
fetch the region of linear memory bound to texture reference texRef using integer
texture coordinate x. No texture filtering and addressing modes are supported. For

CUDA Programming Guide Version 2.3 113

Appendix B. C Extensions

integer types, these functions may optionally promote the integer to single-precision
floating point.
Besides the functions shown above, 2-, and 4-tuples are supported; for example:
float4 tex1Dfetch(
texture<uchar4, 1, cudaReadModeNormalizedFloat> texRef,
int x);
fetches the region of linear memory bound to texture reference texRef using
texture coordinate x.

B.8.2 tex1D()
template<class Type, enum cudaTextureReadMode readMode>
Type tex1D(texture<Type, 1, readMode> texRef,
float x);
fetches the CUDA array bound to texture reference texRef using floating-point
texture coordinates x.

B.8.3 tex2D()
template<class Type, enum cudaTextureReadMode readMode>
Type tex2D(texture<Type, 2, readMode> texRef,
float x, float y);
fetches the CUDA array or the region of linear memory bound to texture reference
texRef using texture coordinates x and y.

B.8.4 tex3D()
template<class Type, enum cudaTextureReadMode readMode>
Type tex3D(texture<Type, 3, readMode> texRef,
float x, float y, float z);
fetches the CUDA array bound to texture reference texRef using texture
coordinates x, y, and z.

B.9 Time Function

clock_t clock();
when executed in device code, returns the value of a per-multiprocessor counter
that is incremented every clock cycle. Sampling this counter at the beginning and at
the end of a kernel, taking the difference of the two samples, and recording the
result per thread provides a measure for each thread of the number of clock cycles
taken by the device to completely execute the thread, but not of the number of
clock cycles the device actually spent executing thread instructions. The former
number is greater that the latter since threads are time sliced.

114 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

B.10 Atomic Functions

An atomic function performs a read-modify-write atomic operation on one 32-bit or
64-bit word residing in global or shared memory. For example, atomicAdd()
reads a 32-bit word at some address in global or shared memory, adds an integer to
it, and writes the result back to the same address. The operation is atomic in the
sense that it is guaranteed to be performed without interference from other threads.
In other words, no other thread can access this address until the operation is
complete.
Atomic operations only work with signed and unsigned integers (with the exception
of atomicExch(), which is also supported for single-precision floating-point
numbers).
Atomic functions can only be used in device functions and are only available for
devices of compute capability 1.1 and above.
Atomic functions operating on shared memory and atomic functions operating on
64-bit words are only available for devices of compute capability 1.2 and above.

B.10.1 Arithmetic Functions

B.10.1.1 atomicAdd()
int atomicAdd(int* address, int val);
unsigned int atomicAdd(unsigned int* address,
unsigned int val);
unsigned long long int atomicAdd(unsigned long long int* address,
unsigned long long int val);
reads the 32-bit or 64-bit word old located at the address address in global or
shared memory, computes (old + val), and stores the result back to memory at
the same address. These three operations are performed in one atomic transaction.
The function returns old.
64-bit words are only supported for global memory.

B.10.1.2 atomicSub()
int atomicSub(int* address, int val);
unsigned int atomicSub(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes (old - val), and stores the result back to memory at the
same address. These three operations are performed in one atomic transaction. The
function returns old.

B.10.1.3 atomicExch()
int atomicExch(int* address, int val);
unsigned int atomicExch(unsigned int* address,
unsigned int val);
unsigned long long int atomicExch(unsigned long long int* address,
unsigned long long int val);
float atomicExch(float* address, float val);

CUDA Programming Guide Version 2.3 115

Appendix B. C Extensions

reads the 32-bit or 64-bit word old located at the address address in global or
shared memory and stores val back to memory at the same address. These two
operations are performed in one atomic transaction. The function returns old.
64-bit words are only supported for global memory.

B.10.1.4 atomicMin()
int atomicMin(int* address, int val);
unsigned int atomicMin(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes the minimum of old and val, and stores the result back to
memory at the same address. These three operations are performed in one atomic
transaction. The function returns old.

B.10.1.5 atomicMax()
int atomicMax(int* address, int val);
unsigned int atomicMax(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes the maximum of old and val, and stores the result back to
memory at the same address. These three operations are performed in one atomic
transaction. The function returns old.

B.10.1.6 atomicInc()
unsigned int atomicInc(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes ((old >= val) ? 0 : (old+1)), and stores the result
back to memory at the same address. These three operations are performed in one
atomic transaction. The function returns old.

B.10.1.7 atomicDec()
unsigned int atomicDec(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes (((old == 0) | (old > val)) ? val : (old-1)),
and stores the result back to memory at the same address. These three operations
are performed in one atomic transaction. The function returns old.

B.10.1.8 atomicCAS()
int atomicCAS(int* address, int compare, int val);
unsigned int atomicCAS(unsigned int* address,
unsigned int compare,
unsigned int val);
unsigned long long int atomicCAS(unsigned long long int* address,
unsigned long long int compare,
unsigned long long int val);
reads the 32-bit or 64-bit word old located at the address address in global or
shared memory, computes (old == compare ? val : old), and stores the
result back to memory at the same address. These three operations are performed in
one atomic transaction. The function returns old (Compare And Swap).

116 CUDA Programming Guide Version 2.3

 Appendix B. C Extensions

64-bit words are only supported for global memory.

B.10.2 Bitwise Functions

B.10.2.1 atomicAnd()
int atomicAnd(int* address, int val);
unsigned int atomicAnd(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes (old & val), and stores the result back to memory at the
same address. These three operations are performed in one atomic transaction. The
function returns old.

B.10.2.2 atomicOr()
int atomicOr(int* address, int val);
unsigned int atomicOr(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes (old | val), and stores the result back to memory at the
same address. These three operations are performed in one atomic transaction. The
function returns old.

B.10.2.3 atomicXor()
int atomicXor(int* address, int val);
unsigned int atomicXor(unsigned int* address,
unsigned int val);
reads the 32-bit word old located at the address address in global or shared
memory, computes (old ^ val), and stores the result back to memory at the
same address. These three operations are performed in one atomic transaction. The
function returns old.

B.11 Warp Vote Functions

Warp vote functions are only supported by devices of compute capability 1.2 and
higher (see Section 4.1 for the definition of a warp).
int __all(int predicate);
evaluates predicate for all threads of the warp and returns non-zero if and only if
predicate evaluates to non-zero for all of them.
int __any(int predicate);
evaluates predicate for all threads of the warp and returns non-zero if and only if
predicate evaluates to non-zero for any of them.

CUDA Programming Guide Version 2.3 117

Appendix B. C Extensions

B.12 Execution Configuration

Any call to a __global__ function must specify the execution configuration for that
call. The execution configuration defines the dimension of the grid and blocks that
will be used to execute the function on the device, as well as the associated stream
(see Section 3.3.9.1 for a description of streams).
When using the driver API, the execution configuration is specified through a series
of driver function calls as detailed in Section 3.3.3.
When using the runtime API (Section 3.2), the execution configuration is specified
by inserting an expression of the form <<< Dg, Db, Ns, S >>> between the
function name and the parenthesized argument list, where:
‰ Dg is of type dim3 (see Section B.3.2) and specifies the dimension and size of
the grid, such that Dg.x * Dg.y equals the number of blocks being launched;
Dg.z must be equal to 1;
‰ Db is of type dim3 (see Section B.3.2) and specifies the dimension and size of
each block, such that Db.x * Db.y * Db.z equals the number of threads per
block;
‰ Ns is of type size_t and specifies the number of bytes in shared memory that
is dynamically allocated per block for this call in addition to the statically
allocated memory; this dynamically allocated memory is used by any of the
variables declared as an external array as mentioned in Section B.2.3; Ns is an
optional argument which defaults to 0;
‰ S is of type cudaStream_t and specifies the associated stream; S is an optional
argument which defaults to 0.
As an example, a function declared as
__global__ void Func(float* parameter);
must be called like this:
Func<<< Dg, Db, Ns >>>(parameter);
The arguments to the execution configuration are evaluated before the actual
function arguments and like the function arguments, are currently passed via shared
memory to the device.
The function call will fail if Dg or Db are greater than the maximum sizes allowed
for the device as specified in Appendix A.1.1, or if Ns is greater than the maximum
amount of shared memory available on the device, minus the amount of shared
memory required for static allocation, functions arguments, and execution
configuration.

118 CUDA Programming Guide Version 2.3

 Appendix C.
Mathematical Functions

Functions from Section C.1 can be used in both host and device code whereas
functions from Section C.2 can only be used in device code.
Note that floating-point functions are overloaded, so that in general, there are three
prototypes for a given function <func-name>:
(1) double <func-name>(double), e.g. double log(double)
(2) float <func-name>(float), e.g. float log(float)
(3) float <func-name>f(float), e.g. float logf(float)
This means, in particular, that passing a float argument always results in a float
result (variants (2) and (3) above).

C.1 Standard Functions

This section lists all the mathematical standard library functions supported in device
code. It also specifies the error bounds of each function when executed on the
device. These error bounds also apply when the function is executed on the host in
the case where the host does not supply the function. They are generated from
extensive but not exhaustive tests, so they are not guaranteed bounds.

C.1.1 Single-Precision Floating-Point Functions

Addition and multiplication are IEEE-compliant, so have a maximum error of
0.5 ulp. However, on the device, the compiler often combines them into a single
multiply-add instruction (FMAD), which truncates the intermediate result of the
multiplication. This combination can be avoided by using the __fadd_rn() and
__fmul_rn() intrinsic functions (see Section C.2).
The recommended way to round a single-precision floating-point operand to an
integer, with the result being a single-precision floating-point number is rintf(),
not roundf(). The reason is that roundf() maps to an 8-instruction sequence on
the device, whereas rintf() maps to a single instruction. truncf(), ceilf(),
and floorf() each map to a single instruction as well.

CUDA Programming Guide Version 2.3 119

Appendix C. Mathematical Functions

Table C-1. Mathematical Standard Library Functions with

Maximum ULP Error
The maximum error is stated as the absolute value of the difference
in ulps between a correctly rounded single-precision result and the
result returned by the CUDA library function.

Function Maximum ulp error

x+y 0 (IEEE-754 round-to-nearest-even)
(except when merged into an FMAD)
x*y 0 (IEEE-754 round-to-nearest-even)
(except when merged into an FMAD)
x/y 2 (full range)
1/x 1 (full range)
1/sqrtf(x) 2 (full range)
rsqrtf(x)
sqrtf(x) 3 (full range)
cbrtf(x) 1 (full range)
hypotf(x,y) 3 (full range)
expf(x) 2 (full range)
exp2f(x) 2 (full range)
exp10f(x) 2 (full range)
expm1f(x) 1 (full range)
logf(x) 1 (full range)
log2f(x) 3 (full range)
log10f(x) 3 (full range)
log1pf(x) 2 (full range)
sinf(x) 2 (full range)
cosf(x) 2 (full range)
tanf(x) 4 (full range)
sincosf(x,sptr,cptr) 2 (full range)
asinf(x) 4 (full range)
acosf(x) 3 (full range)
atanf(x) 2 (full range)
atan2f(y,x) 3 (full range)
sinhf(x) 3 (full range)
coshf(x) 2 (full range)
tanhf(x) 2 (full range)
asinhf(x) 3 (full range)
acoshf(x) 4 (full range)
atanhf(x) 3 (full range)
powf(x,y) 8 (full range)
erff(x) 3 (full range)
erfcf(x) 8 (full range)
erfinvf(x) 5 (full range)

120 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

Function Maximum ulp error

erfcinvf(x) 7 (full range)
lgammaf(x) 6 (outside interval -10.001 ... -2.264; larger inside)
tgammaf(x) 11 (full range)
fmaf(x,y,z) 0 (full range)
frexpf(x,exp) 0 (full range)
ldexpf(x,exp) 0 (full range)
scalbnf(x,n) 0 (full range)
scalblnf(x,l) 0 (full range)
logbf(x) 0 (full range)
ilogbf(x) 0 (full range)
fmodf(x,y) 0 (full range)
remainderf(x,y) 0 (full range)
remquof(x,y,iptr) 0 (full range)
modff(x,iptr) 0 (full range)
fdimf(x,y) 0 (full range)
truncf(x) 0 (full range)
roundf(x) 0 (full range)
rintf(x) 0 (full range)
nearbyintf(x) 0 (full range)
ceilf(x) 0 (full range)
floorf(x) 0 (full range)
lrintf(x) 0 (full range)
lroundf(x) 0 (full range)
llrintf(x) 0 (full range)
llroundf(x) 0 (full range)
signbit(x) N/A
isinf(x) N/A
isnan(x) N/A
isfinite(x) N/A
copysignf(x,y) N/A
fminf(x,y) N/A
fmaxf(x,y) N/A
fabsf(x) N/A
nanf(cptr) N/A
nextafterf(x,y) N/A

C.1.2 Double-Precision Floating-Point Functions

The errors listed below only apply when compiling for devices with native double-
precision support. When compiling for devices without such support, such as
devices of compute capability 1.2 and lower, the double type gets demoted to
float by default and the double-precision math functions are mapped to their
single-precision equivalents.

CUDA Programming Guide Version 2.3 121

Appendix C. Mathematical Functions

The recommended way to round a double-precision floating-point operand to an

integer, with the result being a double-precision floating-point number is rint(),
not round(). The reason is that round() maps to an 8-instruction sequence on
the device, whereas rint() maps to a single instruction. trunc(), ceil(), and
floor() each map to a single instruction as well.

Table C-2. Mathematical Standard Library Functions with

Maximum ULP Error
The maximum error is stated as the absolute value of the difference
in ulps between a correctly rounded double-precision result and the
result returned by the CUDA library function.

Function Maximum ulp error

x+y 0 (IEEE-754 round-to-nearest-even)
x*y 0 (IEEE-754 round-to-nearest-even)
x/y 0 (IEEE-754 round-to-nearest-even)
1/x 0 (IEEE-754 round-to-nearest-even)
sqrt(x) 0 (IEEE-754 round-to-nearest-even)
rsqrt(x) 1 (full range)
cbrt(x) 1 (full range)
hypot(x,y) 2 (full range)
exp(x) 1 (full range)
exp2(x) 1 (full range)
exp10(x) 1 (full range)
expm1(x) 1 (full range)
log(x) 1 (full range)
log2(x) 1 (full range)
log10(x) 1 (full range)
log1p(x) 1 (full range)
sin(x) 2 (full range)
cos(x) 2 (full range)
tan(x) 2 (full range)
sincos(x,sptr,cptr) 2 (full range)
asin(x) 2 (full range)
acos(x) 2 (full range)
atan(x) 2 (full range)
atan2(y,x) 2 (full range)
sinh(x) 1 (full range)
cosh(x) 1 (full range)
tanh(x) 1 (full range)
asinh(x) 2 (full range)
acosh(x) 2 (full range)
atanh(x) 2 (full range)
pow(x,y) 2 (full range)
erf(x) 2 (full range)

122 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

Function Maximum ulp error

erfc(x) 7 (full range)
erfinv(x) 8 (full range)
erfcinv(x) 8 (full range)
lgamma(x) 4 (outside interval -11.0001 ... -2.2637; larger inside)
tgamma(x) 8 (full range)
fma(x,y,z) 0 (IEEE-754 round-to-nearest-even)
frexp(x,exp) 0 (full range)
ldexp(x,exp) 0 (full range)
scalbn(x,n) 0 (full range)
scalbln(x,l) 0 (full range)
logb(x) 0 (full range)
ilogb(x) 0 (full range)
fmod(x,y) 0 (full range)
remainder(x,y) 0 (full range)
remquo(x,y,iptr) 0 (full range)
modf(x,iptr) 0 (full range)
fdim(x,y) 0 (full range)
trunc(x) 0 (full range)
round(x) 0 (full range)
rint(x) 0 (full range)
nearbyint(x) 0 (full range)
ceil(x) 0 (full range)
floor(x) 0 (full range)
lrint(x) 0 (full range)
lround(x) 0 (full range)
llrint(x) 0 (full range)
llround(x) 0 (full range)
signbit(x) N/A
isinf(x) N/A
isnan(x) N/A
isfinite(x) N/A
copysign(x,y) N/A
fmin(x,y) N/A
fmax(x,y) N/A
fabs(x) N/A
nan(cptr) N/A
nextafter(x,y) N/A

C.1.3 Integer Functions

Integer min(x,y) and max(x,y) are supported and map to a single instruction on
the device.

CUDA Programming Guide Version 2.3 123

Appendix C. Mathematical Functions

C.2 Intrinsic Functions

This section lists the intrinsic functions that are only supported in device code.
Among these functions are the less accurate, but faster versions of some of the
functions of Section C.1; they have the same name prefixed with __ (such as
__sinf(x)).
Functions suffixed with _rn operate using the round-to-nearest-even rounding
mode.
Functions suffixed with _rz operate using the round-towards-zero rounding mode.
Functions suffixed with _ru operate using the round-up (to positive infinity)
rounding mode.
Functions suffixed with _rd operate using the round-down (to negative infinity)
rounding mode.
Unlike type conversion functions (such as __int2float_rn) that convert from
one type to another, type casting functions simply perform a type cast on the
argument, leaving the value unchanged. For example,
__int_as_float(0xC0000000) is equal to -2, __float_as_int(1.0f) is
equal to 0x3f800000.

C.2.1 Single-Precision Floating-Point Functions

__fadd_rn() and __fmul_rn() map to addition and multiplication operations
that the compiler never merges into FMADs. By contrast, additions and
multiplications generated from the '*' and '+' operators will frequently be combined
into FMADs.
Both the regular floating-point division and __fdividef(x,y) have the same
accuracy, but for 2126 < y < 2128, __fdividef(x,y) delivers a result of zero,
whereas the regular division delivers the correct result to within the accuracy stated
in Table C-3. Also, for 2126 < y < 2128, if x is infinity, __fdividef(x,y) delivers
a NaN (as a result of multiplying infinity by zero), while the regular division returns
infinity.
__saturate(x) returns 0 if x is less than 0, 1 if x is more than 1, and x
otherwise.
__float2ll_[rn,rz,ru,rd](x) (respectively
__float2ull_[rn,rz,ru,rd](x)) converts single-precision floating-point
parameter x to 64-bit signed (respectively unsigned) integer with specified IEEE-
754 rounding modes.

Table C-3. Single-Precision Floating-Point Intrinsic Functions

Supported by the CUDA Runtime Library with
Respective Error Bounds
Function Error bounds
__fadd_[rn,rz,ru,rd](x,y) IEEE-compliant.

124 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

__fmul_[rn,rz,ru,rd](x,y) IEEE-compliant.
__fmaf_[rn,rz,ru,rd](x,y,z) IEEE-compliant.
__frcp_[rn,rz,ru,rd](x) IEEE-compliant.
__fsqrt_[rn,rz,ru,rd](x) IEEE-compliant.
__fdiv_[rn,rz,ru,rd](x,y) IEEE-compliant.
__fdividef(x,y) For y in [2-126, 2126], the maximum ulp error is
2.
__expf(x) The maximum ulp error is
2 + floor(abs(1.16 * x)).
__exp10f(x) The maximum ulp error is
2 + floor(abs(2.95 * x)).
__logf(x) For x in [0.5, 2], the maximum absolute error
is 2-21.41, otherwise, the maximum ulp error is
3.
__log2f(x) For x in [0.5, 2], the maximum absolute error
is 2-22, otherwise, the maximum ulp error is 2.
__log10f(x) For x in [0.5, 2], the maximum absolute error
is 2-24, otherwise, the maximum ulp error is 3.
__sinf(x) For x in [-π, π], the maximum absolute error
is 2-21.41, and larger otherwise.
__cosf(x) For x in [-π, π], the maximum absolute error
is 2-21.19, and larger otherwise.
__sincosf(x,sptr,cptr) Same as sinf(x) and cosf(x).
__tanf(x) Derived from its implementation as
__sinf(x) * (1 / __cosf(x)).
__powf(x, y) Derived from its implementation as
exp2f(y * __log2f(x)).
__int_as_float(x) N/A
__float_as_int(x) N/A
__saturate(x) N/A
__float2int_[rn,rz,ru,rd](x) N/A
__float2uint_[rn,rz,ru,rd](x) N/A
__int2float_[rn,rz,ru,rd](x) N/A
__uint2float_[rn,rz,ru,rd](x) N/A
__float2ll_[rn,rz,ru,rd](x) N/A
__float2ull_[rn,rz,ru,rd](x) N/A

C.2.2 Double-Precision Floating-Point Functions

__dadd_rn() and __dmul_rn() map to addition and multiplication operations
that the compiler never merges into FMADs. By contrast, additions and
multiplications generated from the '*' and '+' operators will frequently be combined
into FMADs.

Table C-4. Double-Precision Floating-Point Intrinsic Functions

Supported by the CUDA Runtime Library with
Respective Error Bounds

CUDA Programming Guide Version 2.3 125

Appendix C. Mathematical Functions

Function Error bounds

__dadd_[rn,rz,ru,rd](x,y) IEEE-compliant.
__dmul_[rn,rz,ru,rd](x,y) IEEE-compliant.
__fma_[rn,rz,ru,rd](x,y,z) IEEE-compliant.
__double2float_[rn,rz](x) N/A
__double2int_[rn,rz,ru,rd](x) N/A
__double2uint_[rn,rz,ru,rd](x) N/A
__double2ll_[rn,rz,ru,rd](x) N/A
__double2ull_[rn,rz,ru,rd](x) N/A
__int2double_rn(x) N/A
__uint2double_rn(x) N/A
__ll2double_[rn,rz,ru,rd](x) N/A
__ull2double_[rn,rz,ru,rd](x) N/A
__double_as_longlong(x) N/A
__longlong_as_double(x) N/A
__double2hiint(x) N/A
__double2loint(x) N/A
__hiloint2double(x, ys) N/A

C.2.3 Integer Functions

__[u]mul24(x,y) computes the product of the 24 least significant bits of the
integer parameters x and y and delivers the 32 least significant bits of the result. The
8 most significant bits of x or y are ignored.
__[u]mulhi(x,y) computes the product of the integer parameters x and y and
delivers the 32 most significant bits of the 64-bit result.
__[u]mul64hi(x,y) computes the product of the 64-bit integer parameters x
and y and delivers the 64 most significant bits of the 128-bit result.
__[u]sad(x,y,z) (Sum of Absolute Difference) returns the sum of integer
parameter z and the absolute value of the difference between integer parameters x
and y.
__clz(x) returns the number, between 0 and 32 inclusive, of consecutive zero bits
starting at the most significant bit (i.e. bit 31) of integer parameter x.
__clzll(x) returns the number, between 0 and 64 inclusive, of consecutive zero
bits starting at the most significant bit (i.e. bit 63) of 64-bit integer parameter x.
__ffs(x) returns the position of the first (least significant) bit set in integer
parameter x. The least significant bit is position 1. If x is 0, __ffs() returns 0.
Note that this is identical to the Linux function ffs.
__ffsll(x) returns the position of the first (least significant) bit set in 64-bit
integer parameter x. The least significant bit is position 1. If x is 0, __ffsll()
returns 0. Note that this is identical to the Linux function ffsll.
__popc(x) returns the number of bits that are set to 1 in the binary representation
of 32-bit integer parameter x.

126 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

__popcll(x) returns the number of bits that are set to 1 in the binary
representation of 64-bit integer parameter x.
__brev(x) reverses the bits of 32-bit unsigned integer parameter x, i.e. bit N of
the result corresponds to bit 31-N of x.
__brevll(x) reverses the bits of 64-bit unsigned long long parameter x, i.e. bit N
of the result corresponds to bit 63-N of x.

CUDA Programming Guide Version 2.3 127

 Appendix D.
C++ Language Constructs

CUDA supports the following C++ language constructs for device code:
‰ Polymorphism
‰ Default Parameters
‰ Operator Overloading
‰ Namespaces
‰ Function Templates
These C++ constructs are implemented as specified in “The C++ Programming
Langue” reference. It is valid to use any of these constructs in .cu CUDA files for
host, device, and kernel (__global__) functions. Any restrictions detailed in previous
parts of this programming guide, like the lack of support for recursion, still apply.
The following subsections provide examples of the various constructs.

CUDA Programming Guide Version 2.3 129

Appendix D. C++ Language Constructs

D.1 Polymorphism
Generally, polymorphism is the ability to define that functions or operators behave
differently in different contexts. This is also referred to as function (and operator,
see below) overloading.
In practical terms, this means that it is permissible to define two different functions
within the same scope (namespace) as long as they have a distinguishable function
signature. That means that the two functions either consume a different number of
parameters or parameters of different types. When either of the multiple functions
gets invoked the compiler resolves to the function’s implementation that matches
the function signature.
Because of implicit typecasting, a compiler may encounter multiple potential
matches for a function invocation and in that case the matching rules as described in
the C++ Language Standard apply. In practice this means that the compiler will pick
the closest match in case of multiple potential matches.
Example: The following is valid CUDA code:
__device__ void f(float x)
{
// do something with x
}

__device__ void f(int i)

{
// do something with i
}

__device__ void f(double x, double y)

{
// do something with x and y
}

D.2 Default Parameters

With support for polymorphism as described in the previous subsection and the
function signature matching rules in place it becomes possible to provide support
for default values for function parameters.
Example:
__device__ void f(float x = 0.0f)
{
// do something with x
}
Kernel or other device functions can now invoke this version of f in one of two
ways:
f();
// or

130 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

float x = /* some value */;

f(x);

Default parameters can only be given for the last n parameters of a function.

D.3 Operator Overloading

Operator overloading allows programmers to define operators for new data-types.
Examples of overloadable operators in C++ are: +, -, *, /, +=, &, [], etc.
Example: The following is valid CUDA code, implementing the + operation
between two uchar4 vectors:
__device__ uchar4 operator+ (const uchar4 & a, const uchar4 & b)
{
uchar4 r;
r.x = a.x + b.x;
…
return r;
}

This new operator can now be used like this:

uchar4 a, b, c;
a = b = /* some initial value */;
c = a + b;

D.4 Namespaces
Namespaces in C++ allow for the creation of a hierarchy of scopes of visibility. All
the symbols inside a namespace can be used within this namespaces without
additional syntax.
The use of namespaces can be used to solve the problem of name-clashes (two
different symbols using identical names), which commonly occurs when using
multiple function libraries from different sources.
Example: The following code defines two functions “f()” in two separate
namespaces (“nvidia” and “other”):
namespace nvidia {
__device__ void f(float x)
{ /* do something with x */ ;}
}

namespace other {
__device__ void f(float x)
{ /* do something with x */ ;}
}

The functions can now be used anywhere via fully qualified names:
nvidia::f(0.5f);

CUDA Programming Guide Version 2.3 131

Appendix D. C++ Language Constructs

All the symbols in a namespace can be imported into another namespace (scope)
like this:
using namespace nvidia;
f(0.5f);

D.5 Function Templates

Function templates are a form of meta-programming that allows writing a generic
function in a data-type independent fashion. CUDA supports function templates to
the full extent of the C++ standard, including the following concepts:
‰ Implicit template parameter decuction.
‰ Explicit instantiation.
‰ Template specialization.

Example:
template <T>
__device__ bool f(T x)
{ return /* some clever code that turns x into a bool here */ }
This function will convert x of any data-type to a bool as long as the code in the
function’s body can be compiled for the actually type (T) of the variable x.
f() can be invoked in two ways:
int x = 1;
bool result = f(x);
This first type of invocation relies on the compiler’s ability to implicitly deduce the
correct function type for T. In this case the compiler would deduce T to be int and
instantiate f<int>(x).
The second type of invoking the template function is via explicit instantiation like
this:
bool result = f<double>(0.5);

Function templates may be specialized:

template <T>
__device__ bool f(T x)
{ return false; }

template <>
__device__ bool
f<int>(T x)
{ return true; }

In this case the implementation for T representing the int type are specialized to
return true, all other types will be caught by the more general template and return
false.
The complete set of matching rules (for implicitly deducing template parameters)
and matching polymorphous functions apply as specified in the C++ standard.

132 CUDA Programming Guide Version 2.3

 Appendix C. Mathematical Functions

CUDA Programming Guide Version 2.3 133

 Appendix E.
Texture Fetching

This appendix gives the formula used to compute the value returned by the texture
functions of Section B.8 depending on the various attributes of the texture reference
(see Section 3.2.4).
The texture bound to the texture reference is represented as an array T of N texels
for a one-dimensional texture, N × M texels for a two-dimensional texture, or
N × M × L texels for a three-dimensional texture. It is fetched using texture
coordinates x , y , and z .
A texture coordinate must fall within T ’s valid addressing range before it can be
used to address T . The addressing mode specifies how an out-of-range texture
coordinate x is remapped to the valid range. If x is non-normalized, only the
clamp addressing mode is supported and x is replaced by 0 if x < 0 and N − 1 if
N ≤ x . If x is normalized:
‰ In clamp addressing mode, x is replaced by 0 if x < 0 and 1 − 1 N if 1 ≤ x ,
‰ In wrap addressing mode, x is replaced by frac (x) , where
frac ( x ) = x − floor ( x ) and floor (x ) is the largest integer not greater than x .
In the remaining of the appendix, x , y , and z are the non-normalized texture
coordinates remapped to T ’s valid addressing range. x , y , and z are derived
from the normalized texture coordinates x̂ , ŷ , and ẑ as such: x = Nxˆ , y = Myˆ , and
z = Lzˆ .

CUDA Programming Guide Version 2.3 135

Appendix E. Texture Fetching

E.1 Nearest-Point Sampling

In this filtering mode, the value returned by the texture fetch is
‰ tex ( x ) = T [i ] for a one-dimensional texture,
‰ tex ( x, y ) = T [i , j ] for a two-dimensional texture,
‰ tex ( x, y , z ) = T [i, j , k ] for a three-dimensional texture,
where i = floor (x ) , j = floor ( y ) , and k = floor (z ) .
Figure D-1 illustrates nearest-point sampling for a one-dimensional texture with
N =4.
For integer textures, the value returned by the texture fetch can be optionally
remapped to [0.0, 1.0] (see Section 3.2.4.1).

tex(x)

T[3]

T[0]
T[2]

T[1]

x
0 1 2 3 4 Non-Normalized

0 0.25 0.5 0.75 1 Normalized

Figure D-1. Nearest-Point Sampling of a One-Dimensional

Texture of Four Texels

E.2 Linear Filtering

In this filtering mode, which is only available for floating-point textures, the value
returned by the texture fetch is
‰ tex ( x ) = (1 − α )T [i ] + αT [i + 1] for a one-dimensional texture,

136 CUDA Programming Guide Version 2.3

 Appendix D. Texture Fetching

‰ tex ( x, y ) = (1 − α )(1 − β )T [i, j ] + α (1 − β )T [i + 1, j ] + (1 − α ) β T [i, j + 1] + αβ T [i + 1, j + 1]

for a two-dimensional texture,
‰ tex ( x, y , z ) =
(1 − α )(1 − β )(1 − γ )T [i, j , k ] + α (1 − β )(1 − γ )T [i + 1, j , k ] +
(1 − α ) β (1 − γ )T [i, j + 1, k ] + αβ (1 − γ )T [i + 1, j + 1, k ] +
(1 − α )(1 − β )γT [i, j , k + 1] + α (1 − β )γT [i + 1, j , k + 1] +
(1 − α ) βγT [i, j + 1, k + 1] + αβγ T [i + 1, j + 1, k + 1]
for a three-dimensional texture,
where:
‰ i = floor ( x B ) , α = frac ( x B ) , x B = x − 0.5 ,
‰ j = floor ( y B ) , β = frac ( y B ) , y B = y − 0.5 ,
‰ k = floor ( z B ) , γ = frac ( z B ) , z B = z − 0.5 .
α , β , and γ are stored in 9-bit fixed point format with 8 bits of fractional value
(so 1.0 is exactly represented).
Figure F-2 illustrates nearest-point sampling for a one-dimensional texture with
N =4.

tex(x)

T[3]

T[0]
T[2]

T[1]

x
0 1 2 3 4 Non-Normalized

0 0.25 0.5 0.75 1 Normalized

Figure D-2. Linear Filtering of a One-Dimensional Texture of

Four Texels in Clamp Addressing Mode

CUDA Programming Guide Version 2.3 137

Appendix E. Texture Fetching

E.3 Table Lookup

A table lookup TL (x ) where x spans the interval [0, R] can be implemented as
N −1
TL( x) = tex( x + 0.5) in order to ensure that TL (0) = T [0] and TL ( R ) = T [ N − 1] .
R
Figure F-3 illustrates the use of texture filtering to implement a table lookup with
R = 4 or R = 1 from a one-dimensional texture with N = 4 .

TL(x)

T[3]

T[0]
T[2]

T[1]

x
0 4/3 8/3 4

0 1/3 2/3 1

Figure D-3. One-Dimensional Table Lookup Using Linear

Filtering

138 CUDA Programming Guide Version 2.3

Notice
ALL NVIDIA DESIGN SPECIFICATIONS, REFERENCE BOARDS, FILES, DRAWINGS, DIAGNOSTICS, LISTS, AND
OTHER DOCUMENTS (TOGETHER AND SEPARATELY, “MATERIALS”) ARE BEING PROVIDED “AS IS.” NVIDIA
MAKES NO WARRANTIES, EXPRESSED, IMPLIED, STATUTORY, OR OTHERWISE WITH RESPECT TO THE
MATERIALS, AND EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES OF NONINFRINGEMENT,
MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE.
Information furnished is believed to be accurate and reliable. However, NVIDIA Corporation assumes no
responsibility for the consequences of use of such information or for any infringement of patents or other
rights of third parties that may result from its use. No license is granted by implication or otherwise under any
patent or patent rights of NVIDIA Corporation. Specifications mentioned in this publication are subject to
change without notice. This publication supersedes and replaces all information previously supplied. NVIDIA
Corporation products are not authorized for use as critical components in life support devices or systems
without express written approval of NVIDIA Corporation.
Trademarks
NVIDIA, the NVIDIA logo, GeForce, Tesla, and Quadro are trademarks or registered trademarks of NVIDIA
Corporation. Other company and product names may be trademarks of the respective companies with which
they are associated.
OpenCL is trademark of Apple Inc. used under license to the Khronos Group Inc.
Copyright
© 2006-2009 NVIDIA Corporation. All rights reserved.
This work incorporates portions of on an earlier work: Scalable Parallel Programming with CUDA, in ACM
Queue, VOL 6, No. 2 (March/April 2008), © ACM, 2008. http://mags.acm.org/queue/20080304/?u1=texterity"

NVIDIA Corporation
2701 San Tomas Expressway
Santa Clara, CA 95050
www.nvidia.com