In assembly langauge, need to indicate exactly how the increased efficiency is a

Question

In assembly langauge, need to indicate exactly how the increased efficiency is achieved

Printers use a color-encoding system known as CYMK (cyan-yellow-magenta-black).
38   GIF images remain popular on the web because their size is relatively small: each
39   red-green-blue combination is expressed in 8 bits, which means that GIF is limited to
40   256 colors. For instance, the RGB combination 188-143-143 is 'rosy red' in GIF; in
41   CYMK, the corresponding color is 0-45-45-67. Here is a function that converts from
42   GIF to CYMK; the CYMK values are returned as elements in an array:
43  
44   /* rgb2cymk.c is the file */
45   void rgb2cmyk(int r, int g, int b, int ret[]) {
46   int c = 255 - r;
47   int m = 255 - g;
48   int y = 255 - b;
49   int k = (c < m) ? (c < y ? c : y) : (m < y ? m : y);
50   c -= k; m -= k; y -= k;
51   ret[0] = c; ret[1] = m; ret[2] = y; ret[3] = k;
52   }
53  
54   Generate two assembly files for this:
55  
56   1. gcc -S rgb2cymk.c
57   2. gcc -S -O1 rgb2cymk.c
58  
59  
60  
61  
62   After step 1, be sure to rename the file rgb2cymk.s to something else,
63   as step 2 will overwrite the file otherwise.
64  
65   In step 2, the "O1" is capital letter 'O' ("optimize") and the numeral 1 is the
66   optimization level.
67  
68   The assembly-language code generated with optimization on should be more
69   efficient than the other assembly-language code. You need not document the
70   assembly-language lines, but do need to indicate exactly how the increased
71   efficiency is achieved. Be as precise as possible in your comparision of the
72   two assembly-language versions.

----------------------------------------------------------------------------------

No Opt:

   .section   __TEXT,__text,regular,pure_instructions
   .globl   _rgb2cmyk
   .align   4, 0x90
_rgb2cmyk: ## @rgb2cmyk
   .cfi_startproc
## BB#0:
   pushq   %rbp
Ltmp2:
   .cfi_def_cfa_offset 16
Ltmp3:
   .cfi_offset %rbp, -16
   movq   %rsp, %rbp
Ltmp4:
   .cfi_def_cfa_register %rbp
   ;;initializing variable c, m, y
   movl   $255, %eax
   movl   %edi, -4(%rbp)
   movl   %esi, -8(%rbp)
   movl   %edx, -12(%rbp)
   movq   %rcx, -24(%rbp)
   movl   %eax, %edx
   subl   -4(%rbp), %edx
   movl   %edx, -28(%rbp)
   movl   %eax, %edx
   subl   -8(%rbp), %edx
   movl   %edx, -32(%rbp)
   subl   -12(%rbp), %eax
   movl   %eax, -36(%rbp)

   movl   -28(%rbp), %eax
   ;;compare
   cmpl   -32(%rbp), %eax
   jge   LBB0_5
## BB#1:
   movl   -28(%rbp), %eax
   cmpl   -36(%rbp), %eax
   jge   LBB0_3
## BB#2:
   movl   -28(%rbp), %eax
   movl   %eax, -44(%rbp) ## 4-byte Spill
   jmp   LBB0_4
LBB0_3:
   movl   -36(%rbp), %eax
   movl   %eax, -44(%rbp) ## 4-byte Spill
LBB0_4:
   movl   -44(%rbp), %eax ## 4-byte Reload
   movl   %eax, -48(%rbp) ## 4-byte Spill
   jmp   LBB0_9
LBB0_5:
   movl   -32(%rbp), %eax
   cmpl   -36(%rbp), %eax
   jge   LBB0_7
## BB#6:
   movl   -32(%rbp), %eax
   movl   %eax, -52(%rbp) ## 4-byte Spill
   jmp   LBB0_8
LBB0_7:
   movl   -36(%rbp), %eax
   movl   %eax, -52(%rbp) ## 4-byte Spill
LBB0_8:
   movl   -52(%rbp), %eax ## 4-byte Reload
   movl   %eax, -48(%rbp) ## 4-byte Spill
LBB0_9:
   movl   -48(%rbp), %eax ## 4-byte Reload
   movl   %eax, -40(%rbp)
   movl   -40(%rbp), %eax
   movl   -28(%rbp), %ecx
   subl   %eax, %ecx
   movl   %ecx, -28(%rbp)
   movl   -40(%rbp), %eax
   movl   -32(%rbp), %ecx
   subl   %eax, %ecx
   movl   %ecx, -32(%rbp)
   movl   -40(%rbp), %eax
   movl   -36(%rbp), %ecx
   subl   %eax, %ecx
   movl   %ecx, -36(%rbp)
   movl   -28(%rbp), %eax
   movq   -24(%rbp), %rdx
   movl   %eax, (%rdx)
   movl   -32(%rbp), %eax
   movq   -24(%rbp), %rdx
   movl   %eax, 4(%rdx)
   movl   -36(%rbp), %eax
   movq   -24(%rbp), %rdx
   movl   %eax, 8(%rdx)
   movl   -40(%rbp), %eax
   movq   -24(%rbp), %rdx
   movl   %eax, 12(%rdx)
   popq   %rbp
   retq
   .cfi_endproc

.subsections_via_symbols

-----------------------------------------------------------------

Optimization:

   .section   __TEXT,__text,regular,pure_instructions
   .globl   _rgb2cmyk
   .align   4, 0x90
_rgb2cmyk: ## @rgb2cmyk
   .cfi_startproc
## BB#0:
   pushq   %rbp
Ltmp2:
   .cfi_def_cfa_offset 16
Ltmp3:
   .cfi_offset %rbp, -16
   movq   %rsp, %rbp
Ltmp4:
   .cfi_def_cfa_register %rbp
   movl   $255, %r8d
   movl   $255, %eax
   subl   %edi, %eax
   movl   $255, %edi
   subl   %esi, %edi
   subl   %edx, %r8d
   cmpl   %edi, %eax ##;; compare m and c
   jge   LBB0_2
## BB#1: ;; c < m
   cmpl   %r8d, %eax ## compare y and c
   movl   %r8d, %edx
   cmovlel   %eax, %edx
   jmp   LBB0_3
LBB0_2: ##;; c >= m
   cmpl   %r8d, %edi ## compare y and m
   movl   %r8d, %edx
   cmovlel   %edi, %edx
LBB0_3:
   subl   %edx, %eax
   subl   %edx, %edi
   subl   %edx, %r8d
   movl   %eax, (%rcx)
   movl   %edi, 4(%rcx)
   movl   %r8d, 8(%rcx)
   movl   %edx, 12(%rcx)
   popq   %rbp
   retq
   .cfi_endproc

.subsections_via_symbols

Explanation / Answer

What is your question. Looks like you have already ran the code.

Related Questions

Navigate

• Browse (All)
• Browse I
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.