Computer Organization and Assembly Language: Lecture-8-Machine Level Programming-Array Allocation and Access

COMPUTER
ORGANIZATION AND
ASSEMBLY LANGUAGE
LECTURE-8-MACHINE LEVEL
PROGRAMMING-ARRAY ALLOCATION
AND ACCESS
MUHAMMAD HAFEEZ
DEPARTMENT OF COMPUTER SCIENCE
GC UNIVERSITY LAHORE
Arrays
 Arrays in C are one means of aggregating scalar
data into larger data types. C uses a particularly
simple implementation of arrays, and hence the
translation into machine code is fairly
straightforward. One unusual feature of C is that
we can generate pointers to elements within
arrays and perform arithmetic with these
 pointers. These are translated into address
computations in machine code.
 Optimizing compilers are particularly good at
simplifying the address computations used by
array indexing. This can make the correspondence
between the C code and its translation into
machine code somewhat difficult to decipher.
Arrays
 For data type T and integer constant N, the
declaration
T A[N];
has two effects.
 First, it allocates a contiguous region of L.N bytes
in memory, where L is the size (in bytes) of data
type T. Let us denote the starting location as xA
 Second, it introduces an identifier A that can be
used as a pointer to the beginning of the array.
The value of this pointer will be xA
Arrays
 The array elements can be accessed using an
integer index ranging between 0 and N−1.
 Array element i will be stored at address
xA+L.i
Arrays
As examples, consider the following declarations:
char A[12];
char *B[8];
double C[6];
double *D[5];
Array
 Array A consists of 12 single-byte (char)
elements.
 Array C consists of six double-precision floating-
point values, each requiring 8 bytes.
 B and D are both arrays of pointers, and
hence the array elements are 4 bytes each.
Array
 The memory referencing instructions of IA32 are
designed to simplify array access.
 For example, suppose E is an array of int’s, and
we wish to evaluate E[i], where the address of E
is stored in register %edx and i is stored in
register %ecx.
 Then the instruction movl (%edx,%ecx,4),%eax
will perform the address computation xE+4i, read
that memory location, and copy the result to
register %eax.
 The allowed scaling factors of 1, 2, 4, and 8 cover
the sizes of the common primitive data types.
Pointer Arithmetic
 C allows arithmetic on pointers, where the
computed value is scaled according to the size of
the data type referenced by the pointer.
 That is, if p is a pointer to data of type T, and the
value of p is xp, then the expression p+i has
value xp+L.i, where L is the size of data type T.
Pointer Arithmetic
 The unary operators & and * allow the generation
and dereferencing of pointers.
 That is, for an expression Expr denoting some
object, &Expr is a pointer giving the address of
the object.
 For an expression Aexpr denoting an
address,*Aexpr gives the value at that address.
 The expressions Expr and *&Expr are therefore
equivalent.
 The array subscripting operation can be applied to
both arrays and pointers. The array reference
A[i] is identical to the expression*(A+i). It
computes the address of the ith array element
and then accesses this memory location.
Pointer Arithmetic
 The unary operators & and * allow the generation
and dereferencing of pointers.
 That is, for an expression Expr denoting some
object, &Expr is a pointer giving the address of
the object.
 For an expression Aexpr denoting an
address,*Aexpr gives the value at that address.
 The expressions Expr and *&Exprare therefore
equivalent.
 The array subscripting operation can be applied to
both arrays and pointers. The array reference A[i]
is identical to the expression*(A+i). It computes
the address of the ith array element and then
accesses this memory location.
Pointer Arithmetic
A(28/12/20)
 Suppose the starting address of integer array E
and integer index i are stored in registers %edx
and %ecx, respectively.
Pointer Arithmetic
 In these examples, the leal instruction is used to
generate an address, while movl is used to
reference memory (except in the first and last
cases, where the former copies an address and
the latter copies the index).
 The final example shows that one can compute
the difference of two pointers within the same
data structure, with the result divided by the size
of the data type.
Nested Arrays-Section C-
Continue (24/12/20)
 The general principles of array allocation and
referencing hold even when we create arrays of
arrays.
 For example, the declaration
int A[5][3];
is equivalent to the declaration
typedef int row3_t[3];
row3_t A[5];
 Data type row3_t is defined to be an array of
three integers. Array A contains five such
elements, each requiring 12 bytes to store the
three integers. The total array size is then
 4.5.3=60 bytes.
Nested Arrays
 Array A can also be viewed as a two-dimensional
array with five rows and three columns,
referenced as A[0][0] through A[4][2].
 The array elements are ordered in memory in
“row major” order, meaning all elements of row 0,
which can be written A[0], followed by all
elements of row 1 (A[1]), and so on.
Nested Arrays
 This ordering is a consequence of our nested
declaration. Viewing A as an array of five
elements, each of which is an array of three int’s,
we first haveA[0], followed byA[1], and so on.
Nested Arrays
 To access elements of multidimensional arrays,
the compiler generates code to compute the offset
of the desired element and then uses one of the
mov instructions with the start of the array as the
base address and the (possibly scaled) offset as
an index. In general, for an array declared as
T D[R][C];
Nested Arrays
 Array element D[i][j] is at memory address
&D[i][j]=xD+L (C.i+j)
Where L is the size of data type T in bytes.

Nested Arrays
 As an example, consider the 5×3 integer array A
defined earlier. Suppose xA, i, and j are at offsets
8, 12, and 16 relative to %ebp, respectively.
 Then array element A[i][j] can be copied to
register %eax by the following code:
Nested Arrays E1-24/12/20
 As can be seen, this code computes the element’s
address as xA+4j+12i=xA+4(3i+j) using a
combination of shifting, adding, and scaling to
avoid more costly multiplication instructions
Fixed-Sized Arrays
 The C compiler is able to make many
optimizations for code operating on
multidimensional arrays of fixed size. For
example, suppose we declare data type fix_matrix
to be 16×16 arrays of integers as follows:
#define N 16
typedef int fix_matrix[N][N];
Always use #define as best programming practice

Fixed-Sized Arrays
 The following code computes element i, k of the
product of arrays A and B, according to the
formula ∑0≤j<N ai,j.bj,k
 .
Fixed-Sized Arrays
 The following code computes element i, k of the
product of arrays A and B, according to the
formula ∑0≤j<N ai,j.bj,k
 .
Fixed-Sized Arrays
 The recoded code in C
Fixed-Sized Arrays
 This code contains a number of clever
optimizations. It recognizes that the loop will
access just the elements of row i of array A, and
so it creates a local pointer variable, which we
have named Arow, to provide direct access to
row i of the array. Arow is initialized to &A[i][0],
and so array element A[i][j] can be accessed as
Arow[j].
Fixed-Sized Arrays
 It also recognizes that the loop will access the
elements of array B as B[0][k], B[1][k],...,B[15]
[k] in sequence.
 These elements occupy positions in memory
starting with the address of array element B[0][k]
and spaced 64 bytes apart.
 The program can therefore use a pointer variable
Bptr to access these successive locations.
 In C, this pointer is shown as being incremented
by N(16), although in fact the actual address is
incremented by 4.16=64.
Fixed-Sized Arrays
 The following is the actual assembly code for the
loop. We see that four variables are maintained in
registers within the loop: Arow , Bptr, j, and
result.
Fixed-Sized Arrays
 As can be seen, register %ecx is incremented by
64 within the loop (line 6).Machine code considers
every pointer to be a byte address, and so in
compiling pointer arithmetic, it must scale every
increment by the size of the underlying data type.
Variable-Sized Arrays
 Historically, C only supported multidimensional
arrays where the sizes (with the possible
exception of the first dimension) could be
determined at compile time.
 Programmers requiring variable-sized arrays had
to allocate storage for these arrays using
functions such as malloc or calloc, and had to
explicitly encode the mapping of multidimensional
arrays into single-dimension ones via row-major
indexing
 ISO C99 introduced the capability to have array
dimensions be expressions that are computed as
the array is being allocated, and recent versions
of gcc support most of the conventions for
variable-sized arrays in ISO C99.
 In the C version of variable-size arrays, we can
declare an array int A[expr1][expr2], either as a
local variable or as an argument to a function,
and then the dimensions of the array are
determined by evaluating the expressions expr1
and expr2 at the time the declaration is
encountered. So, for example, we can write a
function to access element i, j of an n×n array as
follows:
 The parameter n must precede the parameter
A[n][n], so that the function can compute the
array dimensions as the parameter is
encountered.
 As the annotations show, this code computes the
address of element i, j as xA+4(n.i+j).
 The address computation is similar to that of the
fixed-size array except that
 (1) the positions of the arguments on the stack
are shifted due to the addition of parameter n,
 (2) a multiply instruction is used (line 4) to
compute n.i, rather than an leal instruction to
compute 3i.
 When variable-sized arrays are referenced within
a loop, the compiler can often optimize the index
computations by exploiting the regularity of the
access patterns.
 C code to compute element i, k of the product of
two n×n arrays A and B. The compiler generates
code similar to what we saw for fixed-size arrays.
 Except that it scales Bptr, the pointer to element
B[j][k], by the variable value n rather than the
fixed value N on each iteration.
 We see that the program makes use of both a

scaled value 4n (register %edi) for incrementing
Bptr and the actual value of n stored at offset 8
from %ebp to check the loop bounds.
 The code retrieves the value of n from memory on
each iteration to check for loop termination (line
7). This is an example of register spilling: there
are not enough registers to hold all of the needed
temporary data, and hence the compiler must
keep some local variables in memory. In this case
the compiler chose to spill n, because it is a
“read-only” value—it does not change value within
the loop.
 IA32 must often spill loop values to memory,
since the processor has so few registers. In
general, reading from memory can be done more
readily than writing to memory, and so spilling
read-only variables is preferable.
QUESTIONS
 ??????????????????????????

Computer Organization and Assembly Language: Lecture-8-Machine Level Programming-Array Allocation and Access

Uploaded by

Copyright:

Available Formats

Computer Organization and Assembly Language: Lecture-8-Machine Level Programming-Array Allocation and Access

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Computer Organization and Assembly Language: Lecture-8-Machine Level Programming-Array Allocation and Access

Uploaded by

Copyright:

Available Formats

COMPUTER

As examples, consider the following declarations:

Where L is the size of data type T in bytes.

Always use #define as best programming practice

 We see that the program makes use of both a

You might also like