call instead of invoke!

[Ratch]

pbrennick,

'What difference does it make whether you get the answer in picoseconds or nanoseconds as long as the answer is correct, because for all intents and purposes; we cannot tell the difference.'

Those pico's and nano's add up when the load is heavy.  My Windows boots up correctly, but it sure takes a long time to do so.  I sure can tell the difference using MATHCAD between my old slow and new fast machine.  Ratch

[MichaelW]

Running on my P3, this shows only a few cycles of variation from run to run, and the "horror" code is no faster than the sensible code.

Code Select Expand


; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    .586                       ; create 32 bit code
    .model flat, stdcall       ; 32 bit memory model
    option casemap :none       ; case sensitive

    include \masm32\include\windows.inc
    include \masm32\include\masm32.inc
    include \masm32\include\user32.inc
    include \masm32\include\kernel32.inc
    includelib \masm32\lib\masm32.lib
    includelib \masm32\lib\user32.lib
    includelib \masm32\lib\kernel32.lib
    include \masm32\macros\macros.asm
    include timers.asm
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    .data
    .code
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
start:
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    LOOPCOUNT EQU 5000000

    counter_begin LOOPCOUNT, HIGH_PRIORITY_CLASS
      invoke PostMessage,0,EM_LIMITTEXT,256,0
      invoke PostMessage,0,EM_LIMITTEXT,256,0
      invoke PostMessage,0,EM_LIMITTEXT,256,0
    counter_end
    print ustr$(eax)
    print chr$(" cycles",13,10)

    counter_begin LOOPCOUNT, HIGH_PRIORITY_CLASS
        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET Label_1

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage
        jmp  PostMessage

      Label_1:

    counter_end
    print ustr$(eax)
    print chr$(" cycles",13,10)

    mov   eax,input(13,10,"Press enter to exit...")
    exit
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
end start


Code Select Expand


2151 cycles
2152 cycles


[Vortex]

Hi MichaelW,

After three consecutive tests on my PIV, I got these results:

Code Select Expand


3244 cycles
3301 cycles

3284 cycles
3382 cycles

3225 cycles
3291 cycles


[Grincheux]

Ok, You all are right.

I thought that I could win more cycles. I knew that this code would be hard to modify... later.
It was a test.

And if the push were replaced by the "mov ESP+XXX,data". Would this code be quick ?

I have to search an other method.

Thanks for your help. Sincerely

Kenavo

[MichaelW]

In this case, optimizing the instructions that call the API code will make no significant difference because almost all of the execution time (~99.3%) is spent in the API code.

Code Select Expand


; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    .586                       ; create 32 bit code
    .model flat, stdcall       ; 32 bit memory model
    option casemap :none       ; case sensitive

    include \masm32\include\windows.inc
    include \masm32\include\masm32.inc
    include \masm32\include\user32.inc
    include \masm32\include\kernel32.inc
    includelib \masm32\lib\masm32.lib
    includelib \masm32\lib\user32.lib
    includelib \masm32\lib\kernel32.lib
    include \masm32\macros\macros.asm
    include timers.asm
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    .data
    .code
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
start:
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
    LOOPCOUNT EQU 5000000

    counter_begin LOOPCOUNT, HIGH_PRIORITY_CLASS
      invoke PostMessage,0,EM_LIMITTEXT,256,0
      invoke PostMessage,0,EM_LIMITTEXT,256,0
      invoke PostMessage,0,EM_LIMITTEXT,256,0
    counter_end
    print ustr$(eax)
    print chr$(" cycles",13,10)

    counter_begin LOOPCOUNT, HIGH_PRIORITY_CLASS
        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET Label_1

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage
        jmp  PostMessage
      Label_1:
    counter_end
    print ustr$(eax)
    print chr$(" cycles",13,10)

    counter_begin LOOPCOUNT, HIGH_PRIORITY_CLASS
        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET Label_1

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage

        push 0
        push 256
        push EM_LIMITTEXT
        push 0
        push OFFSET PostMessage
        ;jmp  PostMessage
        REPEAT 15
          pop   eax
        ENDM
    counter_end
    print ustr$(eax)
    print chr$(" cycles",13,10)

    mov   eax,input(13,10,"Press enter to exit...")
    exit
; ««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
end start


Code Select Expand


2141 cycles
2144 cycles
30 cycles


[rea]

But in the case that the majority is my own code, would matter if I use movs instead of pushes???, altought that will require take care of the movement of memory to memory that push can do and mov cant.