ASM Sample request

[Scott Turchin]

I'd like to compare to a friends Python assembler, and use PB to just run a time test...

Basically want it in ASM however...and I don't do ASM well (Or at all for that matter)...

Must decrement from one billion to zero and then spit out how long it took...

I'll probably actually write the code in PB/CC vs PBDLL, don't know that it matters either way...

A chance for a ASM programmer to see what PB is capable of, don't let me down!! LOL


Scott


------------------
Scott
mailto:[email protected][email protected]</A>

[Peter Lameijn]

A PB Asm example (don't have an asm32 compiler myself)

#Compile Exe
#include "win32api.inc"
Function PbMain () As Long
  Local lTimer As Double
  SetPriorityClass GetCurrentProcess, %HIGH_PRIORITY_CLASS  'To speed things up a bit
  lTimer = Timer
  ! mov eax, 1000000000
deceax:
  ! dec eax
  ! jne deceax
  MsgBox "Elapsed time: " + Str$(Timer - lTimer) + " seconds"
End Function

------------------
Peter.
mailto[email protected][email protected]</A>



[This message has been edited by Peter Lameijn (edited September 26, 2000).]

[Borje Hagsten]

I think it'll run a bit faster if you use cmp to check for zero, like:

  ! mov eax, 1000000000

  deceax:           ' label - start of loop
    ! dec eax       ; decrease value
    ! cmp eax, 0    ; compare with zero
    ! jne deceax    ; if not equal to zero, jump to label

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

[Scott Turchin]

Actually using Peter's way was 4.5 seconds to the 6.7 on the last example.

But, point is, we're there, he was bragging 4 seconds or so, so tomorrow we will find out for sure

I thank you very much !, who knows, maybe we'll have another PB user here soon LOL


Scott


------------------
Scott
mailto:[email protected][email protected]</A>

[Borje Hagsten]

This was interesting. I use an old Intel P120 computer here as
Internet machine and when I did a quick test, using cmp was about
twice as fast. Took the code over to a newer 550 MHz machine and,
like you say, Peter's original code without cmp was a bit faster.
Both machines with the same version of Win98.

What's up? How is it possible to optimize code when it behaves
differently in different machines, like this? Maybe one must
detect what processor it has and use different approaches
depending upon the results, to get best optimization for all
kinds of processors?

BTW, just tested a simple DO/LOOP variant and it was just as fast..


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