CodeXL Quick Start Guide

AMD Developer Tools Team

Version 2.0
Revision 1

Table of Contents
INTRODUCTION ................................................................................................................... 3
LATEST VERSION OF THIS DOCUMENT ....................................................................... 3
PREREQUISITES ................................................................................................................... 3
DOWNLOAD AND INSTALL CODEXL ............................................................................. 4
CODEXL HELP ....................................................................................................................... 5
SYSTEM INFORMATION .................................................................................................... 6
TEAPOT SAMPLE PROJECT .............................................................................................. 8
Debug the Teapot Sample Application ........................................................................................ 9
Basic Debugging ........................................................................................................................ 10
Source Code View ..................................................................................................................... 11
Breakpoint View ....................................................................................................................... 12
Watch and Locals Views ............................................................................................................ 13
Explorer View ............................................................................................................................ 15
Call Stack View .......................................................................................................................... 18
Function Calls History View ....................................................................................................... 19
Debugged Process Events View ................................................................................................. 19
Memory View ........................................................................................................................... 19
Statistics View ........................................................................................................................... 19
MATRIX MULTIPLICATION PROJECT ........................................................................ 21
Perform CPU Profile for the Matrix Multiply Sample Application .............................................. 22
CPU Time Based Profile Navigation ........................................................................................... 22
Source Code View ..................................................................................................................... 23
Run the classic textbook sample ............................................................................................... 24
Analyzing the classic implementation ....................................................................................... 26
Analyzing the improved implementation .................................................................................. 28
D3DMULTITHREADING PROJECT............................................................................... 29
Open CodeXL D3DMultiThreading sample ................................................................................ 30
Start a Frame Analysis session .................................................................................................. 32
Capture frame for analysis ........................................................................................................ 33
Analyze a captured frame ......................................................................................................... 34
The frame timeline .................................................................................................................... 35
Navigating the frame timeline................................................................................................... 36
PROFILE MODE ................................................................................................................. 37
CPU Profiling ............................................................................................................................. 38
Overview Tab ........................................................................................................................... 39
Modules Tab ............................................................................................................................. 39
Call Graph Tab .......................................................................................................................... 40
Functions Tab ........................................................................................................................... 41
GPU Profiling ............................................................................................................................. 42
Summary Tab ........................................................................................................................... 43
Performance Counters View .................................................................................................... 44
CodeXL Explorer Tree ............................................................................................................... 45
Power Profiling ......................................................................................................................... 47
Switching to Power Profiling mode .......................................................................................... 47
Starting a new Power Profiling session .................................................................................... 47
 Getting Started with CodeXL
Setting the Sampling Interval ................................................................................................... 49
Stopping a Power Profiling session .......................................................................................... 49
Power Profiling Real-Time Values ............................................................................................ 49
Power Profiling Timeline View ................................................................................................. 49
Power Profiling Summary View ................................................................................................ 51
Configuring Power Profiler Sessions ........................................................................................ 51
ANALYZE MODE ................................................................................................................ 52
Static Kernel and Shader Analysis ............................................................................................. 52
Switching to Analyze mode ...................................................................................................... 52
Creating a new project for Analysis ......................................................................................... 53
Working with the new CodeXL Analyzer Explorer Tree ........................................................... 53
Working with Programs ........................................................................................................... 56
Working with Folders ............................................................................................................... 58
Adding source files to an existing project ................................................................................ 60
Creating a new source file in an existing project ..................................................................... 61
Selecting target devices ........................................................................................................... 63
Build Options - defning kernel/shader compilation options .................................................... 64
Output Tab ............................................................................................................................... 68
Statistics Tab ............................................................................................................................ 69
Viewing compilation output: ISA and IL ................................................................................... 69
Navigating through ISA code with the Enhanced ISA View ...................................................... 71
KNOWN ISSUES ................................................................................................................. 72
SUPPORT ............................................................................................................................. 72

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Introduction
CodeXL™ is a tool suite with a unified user interface that lets you harness the
benefits of AMD CPUs, GPUs, and APUs. It has powerful capabilities for APU/GPU
debugging, CPU and GPU profiling, and static OpenCL™ kernel analysis. These
features let you find bugs, optimize application performance, and easily access
heterogeneous computing. CodeXL is available as a stand-alone application for
Windows® and Linux®, as well as a Microsoft® Visual Studio® extension for
Windows.
Getting the most out of the CodeXLtool suite requires a relatively recent AMD
APU, a recent version of Catalyst, and the OpenCL APP SDK.
This document describes how to
 get started using CodeXL
 find information about known CodeXL issues
 contact AMD for support

Latest Version of This Document

 For the latest and greatest version of the documentation, go to the CodeXL
Website.

Prerequisites
Operating Systems

• Microsoft Windows 7 64-bit

• Microsoft Windows 8.1 64-bit
• Microsoft Windows 10 64-bit
• Linux 64-bit (Red Hat, Ubuntu, SUSE)

For detailed system requirements see the CodeXL Release Notes in the CodeXL
installation folder or on the Documentation section of the CodeXL web page.

CodeXL Visual Studio Extension

• [Optional] Microsoft Visual Studio 2010 (Standard/Professional/Team

System Edition)
• [Optional] Microsoft Visual Studio 2012
(Professional/Premium/Ultimate Edition)
• [Optional] Microsoft Visual Studio 2013
(Professional/Premium/Ultimate Edition)
• [Optional] Microsoft Visual Studio 2015
(Community/Professional/EnterpriseEdition)
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
Profiling OpenCL™ Applications

• [GPU device] AMD Catalyst driver with OpenCL™ GPU support

• [GPU device] AMD Radeon™ HD 5000 series or newer
• AMD APP SDK (requirements)

For detailed system requirements see the CodeXL Release Notes in the CodeXL
installation folder or on the Documentation section of the CodeXL web page.

Download and Install CodeXL

Installation is system-specific (Windows or Linux); but once installed and
started, the CodeXL operation is system-independent.
Download the CodeXLinstallation package from http:// http://gpuopen.com/

For Windows
1. Download the .exe file CodeXL_Win*.exe.
2. When the download completes, double-click the .exe file to install
CodeXL.
The installer guides you through the installation process.
The CodeXL Visual Studio 2010, 2012, 2013 and 2015 extensions are part
of the installer package and are installed by default.
3. In the configuration dialog, de-select the Visual Studio extensions if you
do not want to install them.
4. Launch CodeXL from C:\Program Files (x86)\CodeXL\CodeXL.exe or from
the created Desktop shortcut.

For Linux
Either install the dedicated distribution package or use the tar archive.
The default installation folder
For Red Hat and other Fedora based Linux distributions
1. Download the 64-bit Linux .rpm package CodeXL_Linux*.rpm.
2. If installed version 1. at the machine, remove it first:
$ rpm -qa | grep -i codexl
$ sudo rpm –e <package name>
3. Install the RPM package:
$ sudo rpm -Uvh CodeXL_Linux*.rpm
4. Navigate to /opt/CodeXL_X.Y-ZZZZ/
5. Launch CodeXL using ./CodeXL.

For Ubuntu and other Debian based Linux distributions

1. Download the 64-bit Linux .deb package codexl_*.deb.
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
2. If installed version 1. at the machine, remove it first:
$ dpkg -l |grep -i codexl
$ sudo dpkg –r <package name>
3. Install the DEB package:
$ sudo dpkg -i codexl_x.x-x_amd64.deb
$ sudo apt-get -f install
4. Navigate to /opt/CodeXL_X.Y-ZZZZ/
5. Launch CodeXL using ./CodeXL.

Use the tar archive

1. Download the 64-bit Linux .tar package CodeXL_Linux*.tar.gz.
2. Extract the tar package:
$ tar –xvzf CodeXL_Linux*.tar.gz
3. Install the PowerProfile driver:
$ sudo <codexl-install-dir>/ CXLTPwrProfDriver.sh
install
6. Navigate to <codexl-install-dir>
7. Launch CodeXL using ./CodeXL.

CodeXL Help
To bring up a CodeXL Help window:
1. Click on the CodeXLUser Guide link on the CodeXL startup Home Page.
OR
2. Select Help >> View Help from the CodeXL toolbar.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

CodeXL Help provides some of the same information provided in this document,
but also includes additional details about CodeXL views and modules.
To bring up the CodeXL Help window for the Visual Studio extension:
1. Select CodeXL >> Help >> View Help from the VS menu.

System information
To display system information:
1. Select Tools >> System Information from the CodeXL toolbar.
The tabs let you select a category of information. The following screenshot
shows OpenCL device information for a GPU device and a CPU device on the
runtimes available locally - a 32-bit and 64-bit runtime.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

To display project settings, the project must be stopped.

To edit the settings of a project:
1. Select File >> Project Settings from the drop-down File menu.

See the CodeXL Help for more details about project settings.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Teapot Sample Project
The CodeXL distribution includes a sample project that displays a smoking
teapot. The project uses OpenCL kernels to solve Navier-Stokes equations. It
shares a 3D texture between OpenCL and OpenGL, copies a density field grid into
the 3D texture, and renders the smoke using OpenGL.
For the Visual Studio extension:
1. Select CodeXL >> Open Teapot Sample Project from the VS menu.
Visual Studio displays the teapot sample project.

Screenshots in the remainder of this document show the standalone version of

CodeXL. The Visual Studio version is similar, but contains a VS window rather
than a CodeXL window.
For Windows or Linux:
1. In the CodeXL welcome page (in the CodeXL menu bar, click on File-
>Welcome Page), Under the Samples header, click the CodeXL Teapot link.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

The CodeXL Explorer view now shows:

CodeXL TeaPot | Debug Mode - Not running
The CodeXL window also displays several other views, but since the program is
not running, those views do not display any information.

Debug the Teapot Sample Application

Note: Before debugging the Teapot sample application, you must load it (see the
previous section).
After the teapot sample is loaded, run the debug program:
1. Select Debug >> Start Debugging from the taskbar,

or
2. Click on the green right arrow taskbar.
The program begins execution, and soon displays a rotating smoking teapot in a
separate window.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

To stop the program:

1. Select Debug >> Stop Debugging from the taskbar,
or
2. Click the black square taskbar Stop button ,
or
3. Click the close button in the upper-right corner of the teapot window.

Basic Debugging
The CodeXL GPU Debugger lets you examine the runtime behavior of your
OpenCL/OpenGL application in detail. You can use the information it provides to
find bugs and to improve application performance. You can debug OpenCL
kernels, inspect variable values across different work items and work groups,
and inspect call stacks, among other things.
This quick start guide presumes you are familiar with the use of a GUI debugger;
so the guide provides only a quick introduction to the basic CodeXL debugging
features.
The following four buttons, at the far left of the CodeXL taskbar, let you select
Debug mode, Profile mode, Frame Analysis or Analyze mode.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Hovering over a taskbar button displays a pop-up help description.
The following taskbar buttons control program execution during debugging.

These controls are (left to right): start, pause and stop debugging, frame / api /
draw step (in drop-down menu), step in, step over, step out. You can also
perform these actions from the taskbar Debug pull-down menu, or by using
function keys.
Note: host code debugging (stepping and breaking in C/C++ code) is currently
only available in the Linux CodeXL standalone application, and in Windows
CodeXL Visual Studio extension (32-bit native C/C++ target applications only). In
all other configurations, the step commands will only be enabled in kernel
debugging mode.
The following taskbar buttons show, or hide, various views.

These buttons are (left to right): CodeXL Explorer , Properties, Function Calls
History, Debugged Process Events, Call Stack, Locals, Watch, OpenGL™ State
Variables, OpenCL Multi-Watch (1,2,3), Breakpoints, Memory, and Statistics.
You can resize views, drag, and drop views to rearrange them, or move them to a
separate window. The next sections of this guide describe individual CodeXL
views in more detail.

Source Code View

Source Code views display C, C++, or OpenCL code. To display the Source Code
view:
1. Start the teapot program, as described above.
2. Hit the Break button to interrupt it.
A Source Code view displays the source file where the break occurred, with a
yellow arrow indicating the current line number. In the following
screenshot, it is line 431 in the amdtteapottoglcanvas.cpp file.
3. In all configurations, you can now step to the next API call, OpenGL draw
function call, or Frame Terminator function call by choosing the appropriate
command from the API steps drop-down button (your last choice will be
remembered for quick access).
4. In configurations that allow Host debugging (see note above), you can also
step through the host code with the step in, step over and step out
commands.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Breakpoint View
The Breakpoint view shows active breakpoints. Initially, the Breakpoint view
shows no breakpoints:

To add a breakpoint:
1. Double-click “Double-click to add or remove breakpoints…”
A new Breakpoints window appears.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

2. Select the API Functions tab to set a breakpoint on an API function, or select the
Kernel Functions tab to set a breakpoint on a kernel function.
When program execution hits a breakpoint, the Source view displays the line
where the breakpoint occurs. A yellow arrow indicates the current location.
A red dot next to the line number indicates a set breakpoint.

Watch and Locals Views

In kernel debugging mode

The Watch view shows the values and types of program variables you specify.
The Locals view displays the values and types of local variables in a kernel.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

In the image above, the Watch view displays the value of variable normPos. The
Locals view displays the values of all local variables in the current kernel (in this
case, applySources in tpApplySources.cl). For a structured variable, click on
the triangle to the left of the variable name to see the name and value of each
member.

When CodeXL is in Kernel Debugging mode, move the mouse cursor to hover
over any variable name in the OpenCL kernel source code to display a tooltip
with the variable value. This is demonstrated in the screenshot below.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
In Host debugging mode

In configurations where host debugging is supported (see note above), the

Watch and Locals views will show the currently selected host thread call stack
frame's variables. You can even view host variables during kernel debugging, by
switching to a host thread in the Kernel Work Item toolbar's Threads and
Wavefronts combobox.

Multi-Watch view
The Multi-Watch view lets you compare the values of an OpenCL kernel variable
across work items and work groups.

Explorer View
The Explorer view displays OpenCL-allocated objects and OpenCL/OpenGL
shared contexts.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

1. Click on an object to bring up information about the object in the Properties

view.
For example, clicking on Texture 2 in the view above brings up its properties,
as shown in the next screenshot.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

2. Click on Vertex Buffer object VBO 1 to display its data, with a variety of
available drop-down menu display and format options in the right-
hand panel.

3. Double-click on an object to display an appropriate view.

For example, double-click on Vertex Shader 1 under Shaders to bring up a
Source Code view of its source file tpVertexSharder.glsl. Alternatively,
double-click on Depth buffer to bring up an Image view of the depth buffer.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

You can manipulate an Image view with the following image manipulation
buttons on the CodeXL toolbar:

These buttons let you select, zoom in, zoom out, pan, enable R/G/B/alpha
channels, enable grayscale mode, enable color invert mode, original size, best fit,
and rotate CCW/CW. Hovering over the image displays pixel-specific
information (position and color) in the Image Information panel.
Alternatively, select the Data view tab of the depth buffer to display the buffer as
raw spreadsheet data rather than as an image.

Call Stack View

The Call Stack view displays a combined C/C++/OpenCL call stack.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Function Calls History View
The Function Calls History view displays a log of OpenCL API calls.
1. Click on a function call to display call details in a Properties view.

Debugged Process Events View

The Debugged Process Events view displays process events.

Memory View
The Memory view summarizes memory use.

Statistics View
The Statistics view provides statistical information about the program. Select a
tab to choose among options, such as Function Types:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

or Function Calls:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Matrix Multiplication Project

The CodeXL distribution includes a CPU Profile sample. The sample includes 3
functions that implement matrix multiplication.
Use command line arguments to call each of the 3 implementation:
 Running the sample without any argument will invoke inefficient
implementation of matrix multiplication.
 Running the sample with –c will invoke the classic textbook implementation
of matrix multiplication.
 Running the sample with –i will invoke improved implementation of matrix
multiplication.
For the Visual Studio extension:
1. Select CodeXL >> Open Matrix Multiplication Sample Project from the VS menu.
Visual Studio displays the matrix multiply sample project.

Screenshots in the remainder of this document show the standalone version of

CodeXL. The Visual Studio version is similar, but contains a VS window rather
than a CodeXL window.
For Windows or Linux:
2. In the CodeXL welcome page (in the CodeXL menu bar, click on File-
>Welcome Page), Under the Samples header, click the CodeXL Matrix
Multiply link.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

The CodeXL Explorer view now shows:

CodeXL Matrix Multiply | Profile Mode (CPU:Time-Based Sampling) - Not running

Perform CPU Profile for the Matrix Multiply Sample Application

Note: Before profiling the matrix multiply sample application, you must load it
(see the previous section).
After the sample is loaded, run a Time-Based profile session for the sample:
1. Select CodeXL >> Profile >> Time-Based Sampling to switch to Time Based CPU
Profile mode.
2. Start Profile the inefficient implementation:

Click CodeXL -> Start Profiling or Click on the green right arrow
taskbar. (Local means local host profiling. Use the right black arrow to
configure remote host settings)
The program begins execution, and soon displays a command line window that
will run the matrix multiplication executable. Once the sample execution
completes, CodeXL will open a CPU Profile session overview that displays the
profile results.

CPU Time Based Profile Navigation

After the execution is complete, CodeXL display a profile session overview
window.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

This screenshot display the session overview window. See marked red
rectangles:
1. CodeXL Explorer - The current profile session selected in CodeXL
explorer. Double click on this node in the tree will open the session after it
is closed.
2. Functions view – display the 5 most sampled functions. See that the
function “inefficient_multiply_matrices” was sampled 490 times.
3. Modules view – display the 5 most sampled modules. In this example this
table is not useful. Use it to find inefficient modules in multiple modules
executables.
4. Profile Overview – displays general information of the session.
Executable path, working directory, etc’.

Source Code View

While looking at the overview of this session, we can see that the function
“inefficient_multiply_matrices” is consuming resources. Right click on this
function, and click “Open Source Code”.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Clicking this will open the source code view for the file containing
“inefficient_multiply_matrices”.

The source code view display a line-by-line performance table for the requested
function. In this sample, we can see, marked in red in the above screenshot, that
line 126 had 100% of the samples for this function. Looking at the marked
comment we can see that the function is called 3 times, which is redundant.
To call the “classic_multiply_matrices” only once, we will change the command
line arguments.

Run the classic textbook sample

In order to call “classic_multiply_matrices” in this sample, user should give “-c”
command line argument. In order to configure the command line argument sent
for the session, open project session window.
1. Click File->Project Settings
2. Go to the “General” page
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
3. Click “-c” in the command line argument text box
4. Click OK.
5. Run the profile session again (click Profile -> Start Profile)

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Analyzing the classic implementation

Looking at the session again, we can see that “classic_multiply_matrices” was

sampled 497 times (which is expected, since we call it once, instead of three
times). Right click on “classic_multiply_matrices”, and click “Open Source Code”
to open the source code again for further analysis.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
The source code shows the following line was sampled 475 times.
sum = sum + matrix_a[row][inCol] * matrix_b[inCol][outCol];

In order to look for ideas how to improve this line, we can run an Assess
Performance session, to see how our sample is consuming the current system
resources.
1. Click on Profile -> CPU: Assess Performance to select this profile type.
2. Click on Profile -> Start Profile to run an assess performance session.
3. Look at the displayed session.
4. Right click on “classic_multiply_matrices” to see a line-by-line display of
the system resources.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

In the source code view, look at the values for each of the counters for the
specified line. We can see that we have many data cache misses, which might
cause a performance bottle neck.

Analyzing the improved implementation

The third sample implementation is called “improved_matrix_multiplication”.
This implementation is only slightly different from the classic one. We change
this line:
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
sum = sum + matrix_a[row][inCol] * matrix_b[inCol][outCol];

to this line:
matrix_r[row][outCol] = matrix_r[row][outCol] + matrix_a[row][inCol] *
matrix_b[inCol][outCol];

This change is constructed of 2 changes:

1. We do not use “sum” as temporary variable for the summary
accumulation.
2. We change the order of the summary operation. Instead of multiplying
matric A row by a matrix B column, we go over each element of A, and
multiply by a whole row of B. Since the matrix is kept as a single line in
the memory, we perform much less skips in the memory, therefore expect
to have less Cache Misses.
To run the improved implementation:
1. Click File->Project Settings
2. Go to the “General” page
3. Click “-i” in the command line argument text box
4. Click OK.
5. Run the profile session again (click Profile -> Start Profile)
Now open the source code view again:

The improved function, for the same line, we only have 97 cache misses.

D3DMultiThreading Project

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Open CodeXL D3DMultiThreading sample

The CodeXL distribution includes a DX12 Frame Analysis sample. This sample is
a taken from Microsoft DX12 SDK and can be used to try the Frame Analysis
features in CodeXL.
Start by clicking on the D3DMultiThreading sample link in CodeXL welcome
page.

The sample project is loaded. You can see that the sample name in CodeXL
Explorer. CodeXL will also select the Frame Analysis mode for this sample.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Click on the File->Project Settings command to see the project settings for the
sample. The project Executable Path will point to CodeXL bundled
D3DMultiThreading sample.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Start a Frame Analysis session

Click the Start to execute a Frame Analysis session.

A window with the D3DMultiThreading sample should open

 CodeXL will open a session window. The session window monitors the
running application and will help you Capture the requested frames for later
Frame Analysis.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

The screenshot above display a session view, displaying the running

D3DMultiThreading sample. The image in the center of the window display a
snapshot of the currently running frame.
The text above the frame image will display the following details for the current
frame:
1. Frame index
2. Elapsed time since the application started
3. FPS
Below the frame image there are 2 buttons:
1. Capture – capture the current frame for later analysis
2. Stop – stop the session and start analyzing the captured frames.

Capture frame for analysis

After clicking the “Capture” button, the current frame thumbnail and details will
be added to the right panel. Each of the captured frame will be available after the
session will stop for analysis.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

The above screenshot contain 5 captured frames.

Notice: double clicking on a frame should open it for detailed analysis. This will
not be available while the session is running.

Analyze a captured frame

 Stop the Frame Analysis session (click the stop button).
 Double click on one of the frames on the right panel.

A frame timeline will open with the detailed trace of the D3DMultiThreading

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

The frame timeline

The frame timeline view contains the following elements (from bottom to top).
Each of the elements’ numbers is in the screenshot above.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
1. A navigation chart that allows you to zoom in and out, view the whole
frame timeline or focus on a fragment of it, and display API calls duration,
count and concurrency.
2. A collapsed detailed view of the API calls and their
durations/count/concurrency.
3. A timeline chart visualizing the API calls over the frame timeline.
4. An API calls table for each of the CPU threads.
5. A commands table for each of the GPU command queues.
6. Summary tables that summarize the time consumption for each of the API
types in the frame.
7. A Top 20 calls table for the currently selected API in the summary table.

Navigating the frame timeline

A graphics frame can be very busy and contains tens of thousands of API calls.
The following buttons and UI elements are useful in navigating the frame
timeline and highlighting API calls of interest.

Use the left and right handles to expand/reduce the focused timeline fragment in
and out:

Double click a CPU API table item, and the timeline chart will zoom to the
corresponding timeline item and highlight it. If there is a linked GPU API item, it
will also be highlighted in the GPU API table

Select “Duration” in the top left combo box, to view the API call duration in
details:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Select “Count” in the top left combo box, to view the API calls count for each time
fragment:

Select “Concurrency” in the top left combo box, to view the max / average busy
threads concurrency over the frame timeline:

Profile Mode
CodeXL profile mode is a powerful performance analysis tool that supports CPU
and GPU profiling to provide program performance data. CodeXL profiling does
not require modifications to your source code or project. Profiling does not
require recompilation, except for CPU profiling, which requires compilation with
debugging enabled. Profiling lets you find performance hotspots and issues,
determine the top data transfer and kernel execution operations, and identify
problems such as failed API calls and resource leaks. You can use profiling to
improve application performance through proper synchronization, bottleneck
elimination, and load balancing.
CodeXL provides several modes of profiling. These modes let you assess
program performance, use instruction-based sampling (IBS) or time-based
sampling (TBS), or investigate branching, data access, instruction access, or L2
cache access. GPU profiling provides application timeline trace and performance
counter modes.
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
The following is a quick introduction to CPU and GPU profiling. For further
details, see the CodeXL Help information.

CPU Profiling
To profile a program:
1. Click on the profiling mode taskbar button.
2. Use the Profile drop-down menu to select the profiling mode.
For example, for CPU performance profiling, select Profile >> CPU: Assess
Performance.
3. Click the start button to launch the application for profiling.
4. To stop it, use the stop button any time during profiling. The bottom of the
CodeXL window displays the elapsed clock time.
Profiling is available up to the time the application is closed. For the teapot
example: click on the ‘x’ in the upper right corner of the teapot window.
After profiling is complete and data translation is over, a node in the left session
tree is added for this session.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Overview Tab

The first page shown is the overview page. It shows the Modules and Functions
tables and a brief description of the execution environment and profile detail. If
multiple processes are profiled, then the Process table is shown. Each table
shows the top five hot items.

Modules Tab

To open the Modules view:

1. Double click or use the “Open Modules” command from the context menu of
the tree’s Modules node.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Call Graph Tab

To open the Call Graph view:

1. Double click the Call graph node in the Explorer tree, or use the “Open Call
Graph” command from the context menu of the Call Graph node (available
only if Call Stack Sampling was enabled).

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Functions Tab

To open the Functions view:

1. Double-click the Functions node in the Explorer tree, or use the “Open
Functions” command from the context menu of the Function node.
The Functions list can be filtered based on the module to which they belong.
To do this, invoke a dialog from the hyperlink at the top of function table that
lists the displayed and hidden modules. The Functions list also can be filtered
to display functions for a specific process using the Process drop list.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

GPU Profiling
For GPU application timeline trace profiling:
1. Click on the profiling mode taskbar button.
2. Select Profile >> GPU: Application Timeline Trace from the Profile drop-
down menu.
3. Run the program, then let it complete, or terminate it.
An Application Timeline Trace view appears with a timeline of the
program execution. This timeline shows the created OpenCL contexts
and command queues, as well as the relationships between them.
To select a subrange of the timeline, hold down <Ctrl>, and click and
drag on a section of the timeline.
To shift the timeline display left or right, simply click on it and drag.
To zoom in/out, use the mouse wheel or the +/- keys. Selecting a small
subrange lets you zoom in to see details about each event.
For additional information, hover over an event; this displays a pop-up.
The following screenshot is an example of a COPY_BUFFER_TO_IMAGE data
transfer event at 7752.980 ms on the timeline. The pop-up provides detailed
timing data.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Summary Tab

The Summary tab provides several options for viewing profiling data: API,
context, kernel, top 10 data transfer, top 10 kernel, warnings/errors.
The following screenshot shows an example of a Top 10 Kernel Summary.

The Warning(s)/Error(s) summary also includes a helpful list of best practice

recommendations to improve program performance. The following example
indicates issues with blocking write calls and small global work size.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Performance Counters View

The Performance Counters view in a GPU Performance Counters profile provides

kernel performance details, including global work size and time. This mode
collects performance counters from the GPU or APU for each kernel dispatched
to the device. It also displays statistics from the shader compiler for each kernel
dispatched. The performance counters and statistics can be used to discover
kernel bottlenecks.
To display a Code viewer with kernel code:
1. Click on a kernel name (Method) in the Performance Counters view.

A pull-down bar at the top of the window under the Code Viewer tab (see
following screenshot) lets you select OpenCL source (CL), intermediate language
(IL), or instruction set architecture (ISA) code.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

CodeXL Explorer Tree

The Explorer view lets you switch between profiling sessions. This view lists all
profiling sessions for the current project.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

To display a session’s data, double-click on it. To rename or delete it, right-click

on it.
To import profiling data, right-click or drag/drop a session data file to the
Explorer.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Power Profiling
Power Profiling mode allows monitoring the power consumption of supported
AMD APUs in real-time. Additional attributes can also be monitored: GPU and
CPU frequencies, APU thermal trend and CPU Cores P-State.
For a list of currently supported APUs see the System Requirements section of
the CodeXL Release Notes, available in the CodeXL installation folder and on the
CodeXL web page.
Switching to Power Profiling mode

In the Profile Mode drop-down list, choose Power Profiling.

Starting a new Power Profiling session

After switching to Power Profiling mode, double click New Power Session… in
the CodeXL Explorer tree.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Please note that if your hardware does not support Power Profiling, you will not
be able to start the session.
After double clicking on New Power Session… in the CodeXL Explorer tree, an
empty Power Profiling session window will be opened.

Clicking the Start button will start the session with the current configuration.
We will now see how to change a Power Profiling session’s configurations.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Setting the Sampling Interval

To configure the sampling interval (the time period which will pass between
every two consecutive samples of the active counters), click on Profile->Profile
Settings -> Power Profile:

Please note that the current minimum sampling interval is 100 milliseconds.
Click OK to apply your changes.
Stopping a Power Profiling session

To stop a running Power Profiling session, click on the Stop button .

Power Profiling Real-Time Values

Throughout the Power Profiling session, the collected data will be streaming in
real-time to the graphs. There are two tabs that present data: Timeline View and
Summary View.
Power Profiling Timeline View

The Power Profiler Timeline View displays the measured values of the activated
counters throughout the session. The horizontal axis of all charts represents the
time that has passed since the beginning of the session. During profiling sessions,
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
all of the charts in the Timeline View are being updated in real-time with the
measured values which are streaming in. The uppermost ribbon (titled “Total
APU Power”) displays the overall power consumption of the APU throughout the
session.
The top chart has an adjustable range slider that controls the display of all the
other timeline charts. By performing such actions as dragging the slider
sideways, extending or retracting it, you set the scope of attention and the focus
of the timeline charts. Each of the charts below displays only the data that was
collected in the time range corresponding to the slider’s position and length.
That is, the data in all timeline charts, except for the Total APU Power chart itself,
is dictated by the time range which is selected by the Total APU Power chart’s
range slider.
Below the Total APU Power ribbon, you will find additional ribbons containing
more graphs, according to the set of activated counters: A Power chart which
displays the power consumed by specific APU components (such as CPU cores or
integrated GPU), a Frequency chart which displays the frequency of the selected
components, a Temperature chart which displays the thermal trend of the
selected components, and a CPU State chart which displays the CPU core states.
The APU Power graph is always displayed, since the Total APU Power counter is
activated by default and cannot be deactivated. The other charts (frequency,
temperature and CPU core state) are optional, and will only be displayed if the
relevant counters were activated.
To the right side of each chart you will find a legend that displays the measured
values at a specific point in time. To change the point in time for which the values
are displayed, reposition the mouse cursor horizontally on one of the graphs. The
list of counters in the legends is customizable, and specific counters can be
removed/added between profile sessions.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Power Profiling Summary View

The Power Profiler Summary View displays an analysis of the values measured
throughout the session. Similarly to the Timeline View, this view is updated in
real-time when power profiling sessions are running.
At the upper-left side of the summary view, you can see the session duration
which is the amount of time that the profiling session was in progress.
The following Power histograms graphs are plotted in the Summary View:
 Total Energy Consumption
This histogram graph displays the cumulative energy consumed by the APU
and discrete GPU components, measured in Joules.
 Average Power Consumption
This graph displays the average power consumption of the APU and the
discrete GPU components, measured in Watts.
If CPU or GPU frequency counters were activated for the session, you will find
additional histogram graphs below the Power graph in the Summary View:
 CPU Frequency Graph
This stacked histograms graph displays for each CPU core how much time it
spent at each frequencies range.
 GPU Frequency Graph
This graph displays how much time the GPU spent at each frequencies range.
Configuring Power Profiler Sessions

For further configuration options of the Power Profiler such as selecting which
counters will be sampled and launching an application when the session begins,
see the CodeXL User Guide by selecting from the menu bar Help -> View Help.
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
Analyze Mode
Static Kernel and Shader Analysis
In Analyze mode, you can compile and generate performance statistics for
OpenCL kernels, DirectX Shaders, OpenGL Programs and Vulkan Programs. The
compilation and statistics generation process can be targeted at a variety of AMD
GPUs and APUs, independent from the actual type of GPU/APU that is installed
on your system. Using the Analyzer, you can generate and inspect performance
statistics, ISA code, IL code and D3D ASM code.
Switching to Analyze mode

Option 1:

Click on the Analyze Mode button in the CodeXL Mode toolbar.

Option 2:

Click Analyze in the main menu.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Once you switch to Analyze mode, you can create a new project, open a
previously saved project or load the Teapot sample.
Creating a new project for Analysis

Click on the “File->Create Project”, or use the Ctrl+N shortcut. The following
CodeXL Project Settings dialog will appear:

Rename the project, and click on the OK.

After the new project has been created, the CodeXL Analyzer Explorer Tree
should appear in the left pane:
Working with the new CodeXL Analyzer Explorer Tree

If you are familiar with the former versions of the Analyzer, you probably noticed
that the tree has a different structure than the one used in previous versions.
Let’s examine the structure of the new CodeXL Analyzer Explorer:
1. The Source Files pool: this is a logical container that holds all of the source files that
have been added or created in the project since its creation. Note that source files
cannot be built directly from the pool. In order to build a source file, we first need to
attach it to a suitable Program or a Folder. This can be easily done by dragging the
source file from under the Source Files pool to the relevant program or folder. The
next item in this section discusses the concept of Programs and Folders in more
detail.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

2. Programs and Folders: before describing how to technically create Programs and
Folders, let’s first discuss what those objects are, and why they can be useful.
a. Programs (OpenGL, Vulkan):
As of version 2.0, CodeXL can compile and link together multiple
source files for OpenGL and Vulkan. This is especially important when
different shaders have mutual impact on one another’s ISA and
performance statistics. To provide that type of support, CodeXL
Analyzer introduced the concept of a Program. There are two types of
Programs in CodeXL 2.0:
- Rendering Programs
- Compute Programs

A Rendering Program represents a graphics pipeline, and can have a

single shader attached to each of its stages:
- Vertex
- Tessellation Control
- Tessellation Evaluation
- Geometry
- Fragment

A Compute Program represents a compute pipeline, and can have a

single compute shader attached to its single stage.
When you build a program that has multiple shaders attached to it, all shaders
are being compiled and linked together. This way, you get more accurate ISA
and performance statistics than those generated using previous versions of
CodeXL.
b. Folders (OpenCL, DirectX):
Folders are logical containers of source files. When you build a folder that has
multiple source files attached to it, the source files are simply being built one
after the other. Unlike programs, there is no kind of interdependency between
the source files in a given folder: when a folder is being built, each source file is
being compiled independently. Folders can be used to organize the project, by
serving as a logical separator. They can also be used to ease the process of
comparing build results, since now the build results are being maintained per-
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
folder: you can create two different Folders, each containing the same source
files, but have a different configuration (for example, create two DirectX Folders,
each with a different shader model). After building the two Folders, you can
toggle between the performance statistics of the two Folders to see the
differences.

You may ask yourself why CodeXL does not support the concept of DirectX
Programs, just like it does for OpenGL and Vulkan. This is a good point.
Supporting DirectX Programs is at a high priority in the Analyzer’s roadmap, and
we will do our best to add that feature in the upcoming versions of the product.
Creating a new Program or Folder
To create a new Program or a Folder, double-click on the “Create new
program/folder” item in CodeXL Analyzer Explorer Tree:

Then, the following dialog would pop-up:

Select the Program/Folder type of choice, and click OK.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Then, the empty Program/Folder would appear in the Explorer Tree.
For Example, if you choose an OpenGL Rendering Program, you will
see an empty OpenGL Rendering Program created:

Working with Programs

After creating a new program, you will see that it contains an empty placeholder
for every pipeline stage. Right-click on any stage to add an existing shader or
create a new one:

Note: You can also double-click on a stage to create a new shader and
automatically attach it to that Program’s stage.
Whether you chose to create a new shader or to attach an existing one, after the
shader is added to the program, it will also being listed under the Source Files
pool for future use. This will enable you to attach that specific shader to other,

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
future or existing, programs as well. There is no limit for the number of
programs that can reference the same shader.

As you can see in the above screenshot, we attached SimpleVertexShader.vs as

the vertex shader to our OpenGL Rendering Program, and it was also
automatically added to the Source Files pool. We can now drag
SimpleVertexShader.vs from the Source Files pool and drop it on the stage node
of any Program that we may add to the project, to reuse SimpleVertexShader.vs
(there is no dependency in the build process between different Programs).
To build the program, right-click on it and select the Build option, or use the F7
shortcut:

You can also select the Program and manually click on the Build button in the
Analyzer toolbar:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Working with Folders

After creating a new OpenCL or DirectX Folder, an empty Folder would be listed
in the Explorer Tree:

To create a new source file, and automatically add it to the Folder, double-click
on the “Create new source file item…” item of the folder:

To add an existing source file, and automatically add it to the Folder, double-click
on the “Add existing source file item…” item of the folder:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

After adding source files to the Folder, they will be listed both under the active
Folder and under the Source Files pool (for future use):

To configure the build properties of a source file under a specific Folder, click on
that source file and use the Analyzer toolbar’s Type and Entry point drop-down
lists. The first sets the type of the shader and the latter specifies the specific
target shader (among the shaders in the source file). This configuration is Folder-
specific. That is, the same source file can be set with different properties under
different Folders. CodeXL will remember those configurations for you.
To configure the build properties of the Folder, click on the Folder and adjust the
enabled items in the Analyzer toolbar. For CodeXL 2.0, this is only relevant to the
DX Shader Model property of DX Folders:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Once set, the DX Shader Model value will hold for all the shaders in the selected
Folder. For example, if you choose 5_0 as the DX Shader Model, any D3D vertex
shader in that Folder will be compiled using shader model vs_5_0.
To build the whole Folder, right-click on it and select the Build item:

Unlike the case with Programs, Folders are more flexible as they allow you to
build selected source files, without being required to build the whole Folder. To
build selected source files, click on the selected source files under the program,
while holding the Ctrl key. Then, right-click on one of the selected files and select
the build option:

Adding source files to an existing project

The previous sections described how to add existing source files to specific
Programs or Folders. You can also add existing source files to the Source Files
pool of a project, without attaching the files to any Program or Folder. To do so,

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
double-click on the “Add an existing source file…” node under the Source Files
pool:

Creating a new source file in an existing project

The previous sections described how to create a new source files and attach it to
a specific Program or Folder. You can also create new source files which will only
be added to the Source Files pool of a project, without attaching the files to any
Program or Folder. To do so, double click on the “Create new source file…” node
under the Source Files pool:

This will open up the “Create a New Source File” dialog:

First, you need to select the platform of choice, in order to specify the type of
source file that you would like to create.
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
After selecting the platform, you need to select the type of the source file (this
step is relevant to all platforms, except for OpenCL, where there are only .cl files):

As a last step, you need to select what to do with that file using the Associate
with program drop-down list:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

There are three options:

a. Create a New Program (or: Folder): CodeXL will automatically create a new program
of the relevant type, and attach the source file to that program/folder.
b. <Program/Folder name>: the new file would be attached to the selected
Program/Folder.
c. None: the new file will be listed under the Source Files pool, without being attached
to any Program/Folder.

Selecting target devices

CodeXL Analyzer can target a variety of devices, independent of the device that is
physically installed on your system. To select the target devices, for which the
build would be performed, first click on the Select Devices button in the Analyzer
toolbar:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Then, the CodeXL Options dialog would pop-pup with its Analyze tab activated.
The devices are grouped by generations. You can use the check boxes to select
and remove devices:

Build Options - defning kernel/shader compilation options

In the Static Analyze toolbar, there is a text box where you can manually define
specific OpenCL/HLSL build options (there is no support for GLSL build options):

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
DirectX (Direct3D) shaders written on high-level shader language (HLSL) are
supported on MS Windows only.

Note: The display of this toolbar is dynamic; you can set it from the right-click
menu in the main CodeXL frame:

The Build Options box is a place to set compiler build flags such as –x clc++ or
–o3. Any compiler build flag can be placed in this box.

OpenCL Build Options Dialog

This dialog will help you choose the correct OpenCL build options for you and
hopefully will prevent you from making spelling mistakes while typing the
options manually.

To open the OpenCL Build Options dialog, press the Button. You can
browse between the ‘General & Optimization’ tab and the ‘Other’ tab to view all
the available options. Once you choose an option, the option text will appear in
the text box below marked as ‘OpenCL Build Command Line’. This string will also
appear in the menu bar after you click the OK button.
© Copyright 2016 (c), Advanced Micro Devices, Inc.
 Getting Started with CodeXL
Typing the command line in the text box will also mark the corresponding check
boxes in the dialog.

Examples of using build options

For building the tpAdvectFieldScalar.cl kernel from CodeXL TeaPot sample

project, enter the following options:
-D GRID_NUM_CELLS_X=64 -D GRID_NUM_CELLS_Y=64 -D GRID_NUM_CELLS_Z=64
-D GRID_INV_SPACING=1.000000f -D GRID_SPACING=1.000000f -D
GRID_SHIFT_X=6 -D GRID_SHIFT_Y=6 -D GRID_SHIFT_Z=6 -D
GRID_STRIDE_Y=64 -D GRID_STRIDE_SHIFT_Y=6 -D GRID_STRIDE_Z=4096 -D
GRID_STRIDE_SHIFT_Z=12 -I path_to_example_src

On windows, path_to_example_src should be:

C:\Users\Public\Public Documents\CodeXL\Examples\Teapot\res

On Linux, path_to_example_src should be:

/opt/ CodeXL_X.X/examples/Teapot/TeaPotLib/
Adding the option ‘-h’ will dump the list of OpenCL compiler available options in
the output tab.

HLSL Build Options Dialog

This dialog will help you choose the correct DirectX build options for you and
hopefully will prevent making spelling mistakes while typing the options
manually.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

To open the dialog, press The Button. The dialog will be opened. Click
the”HLSL Build Options” node to view the available options.
Once you choose an option, the option text is displayed in the”HLSL Build
Command Line” text box that appears below.
This build option string will also appear in the toolbar’s build options box after
you click the OK button.
As an alternative to selecting options through the radio buttons, it is possible to
type a command in the “HLSL Build Command Line” text box. Build options types
in the text box will automatically be translated to update of the relevant controls
accordingly. For example, typing “D3DCOMPILE_DEBUG” in the lower text box
automatically updates the “Debug” check box to be checked.

HLSL Compiler Tools

First select the build tool of choice for compiling the shader – D3d compiler / FXC
compiler.
The CodeXL installation includes a copy of the Microsoft DirectX compiler DLL:
d3dcompiler_47.dll. You may specify a different path if you want CodeXL to use a
different d3dcompiler module. If you select the FXC compiler tool you must
specify a path to the location of this tool.
To select the path of the compiler module, click the ‘Browse…” option from the
combo-box. When selecting Browse, a dialog box will open for selecting the
compiler file.
 For D3D compiler – any file called d3compiler_*.dll can be selected
 For FXC compiler – only FXC.exe file can be selected.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Note: for D3D compiler the bundled file is selected by default.

Output Tab

The compiler output appears in the Output tab. The example below shows
successful builds (no warnings or errors) for 4 devices.
If errors occur, the output will display the error and the line in which the error
occurred.

If there were errors, the output pane will display the error and the line where the
error occurred:

Double clicking on an error navigates the user to the Source Code view,
displaying the kernel/shader source code:

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
Statistics Tab

The Statistics tab gives detailed statistics for the selected kernel/shader for each
target device. To open the Statistics tab, expand the desired kernel in the project
tree, and double-click the Statistics node:

Note: the statistics tab for pre-GCN devices (v4 & v5 generations) is a bit
different. For more information, see the complete help manual document.
Viewing compilation output: ISA and IL

The performance statistics tab will be opened automatically when the build
process is over. To view the compilation output, double click the node of the
desired ASIC in the explorer tree, under the Program/Folder and configuration
(32-bit or 64-bit):

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

This will open a tab containing the source code, the IL and the ISA. The program
source code and the AMD IL code will be presented as standard text documents.
The ISA will be presented in the “Enhanced ISA View” for GCN devices, and as a
standard text document for pre-GCN devices.

The context menu enables to display/hide line numbers for each source
code/IL/ISA tab.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Navigating through ISA code with the Enhanced ISA View

Using this view, you can inspect the ISA code of GCN devices and see the
estimation for instruction cost in clock cycle. The view contains 5 columns:
- Address: the instruction’s offset within the program (in bytes)
- Opcode: the operation to be performed
- Operands: the data for the operation
- Cycles: the number of clock cycles which are required by a Compute Unit
in order to process the instruction for a 64-thread Wavefront, while
neglecting the system load and any other runtime-related factor.
- Instruction Type: the category of instructions to which the instruction
belongs
- Hex: binary representation of the instruction, in hexadecimal format

Notes:
1. Note that code labels which appear in the Operands column are clickable. By
clicking on a label link, you can navigate to the label’s spot in the code.
2. Note that this view is only available for GCN devices. For pre-GCN devices, the
plain textual ISA view will be displayed.

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL

Known Issues
For a list of known CodeXL issues, review the release notes on the CodeXL web
page and the AMD Developer Tools CodeXL forum:
https://github.com/GPUOpen-Tools/CodeXL/issues

Support
AMD general developer support page:
http://developer.amd.com/support/
Tools & SDKs section in AMD Developer Tools website:
http://developer.amd.com/tools-and-sdks/
OpenCL Zone in AMD Developer Tools website:
http://developer.amd.com/tools-and-sdks/opencl-zone/

© Copyright 2016 (c), Advanced Micro Devices, Inc.

 Getting Started with CodeXL
AMD Accelerated Parallel Processing OpenCL Programming Guide:
http://developer.amd.com/wordpress/media/2013/07/AMD_Accelerated_Paral
lel_Processing_OpenCL_Programming_Guide-rev-2.7.pdf
For GPU development issues relating to other AMD tools, see the AMD GPU
Developer Tools Forum:
community.amd.com/community/devgurus/gpu_developer_tools
To report a specific problem or request help with CodeXL, visit the CodeXL
Forum at:
https://github.com/GPUOpen-Tools/CodeXL/issues
DISCLAIMER
The information contained herein is for informational purposes only, and is
subject to change without notice. While every precaution has been taken in the
preparation of this document, it may contain technical inaccuracies, omissions
and typographical errors, and AMD is under no obligation to update or otherwise
correct this information. Advanced Micro Devices, Inc. makes no representations
or warranties with respect to the accuracy or completeness of the contents of
this document, and assumes no liability of any kind, including the implied
warranties of non-infringement, merchantability or fitness for particular
purposes, with respect to the operation or use of AMD hardware, software or
other products described herein. No license, including implied or arising by
estoppel, to any intellectual property rights is granted by this document. Terms
and limitations applicable to the purchase or use of AMD’s products are as set
forth in a signed agreement between the
parties or in AMD's Standard Terms and Conditions of Sale.
AMD, the AMD Arrow logo, AMD Radeon, AMD FirePro, gDEBugger, and
combinations thereof are trademarks of Advanced Micro Devices, Inc. Windows
and DirectX are registered trademarks of Microsoft Corporation. PCIe and PCI
Express are registered trademarks of PCI-SIG Corporation. OpenCL is a
trademark of Apple Inc. Linux is a registered trademark of Linus Torvalds. Other
product names used in this publication are for identification purposes only and
may be trademarks of their respective companies.
© 2015 Advanced Micro Devices, Inc. All rights reserved.

© Copyright 2016 (c), Advanced Micro Devices, Inc.