« previous next »
Pages: 1 [2]

Author Topic: Performance differences for MOVAPS, MOVDQA and MOVAPD  (Read 23952 times)

phaap

  • Guest
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #15 on: January 18, 2013, 03:36:34 AM »
...
« Last Edit: January 18, 2013, 08:48:17 AM by phaap »
Logged

phaap

  • Guest
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #16 on: January 18, 2013, 03:37:47 AM »
...
« Last Edit: January 18, 2013, 08:48:07 AM by phaap »
Logged

phaap

  • Guest
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #17 on: January 18, 2013, 03:41:11 AM »
...
« Last Edit: January 18, 2013, 08:47:54 AM by phaap »
Logged

qWord

  • Member
  • *****
  • Posts: 1475
  • The base type of a type is the type itself
    • SmplMath macros
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #18 on: January 18, 2013, 04:24:21 AM »
can you please explain how I should combine those blocks (your current code is an infinite loop). Also, how can I test the algorithm for correctness?
Logged
MREAL macros - when you need floating point arithmetic while assembling!

phaap

  • Guest
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #19 on: January 18, 2013, 05:26:05 AM »
...
« Last Edit: January 18, 2013, 08:48:32 AM by phaap »
Logged

qWord

  • Member
  • *****
  • Posts: 1475
  • The base type of a type is the type itself
    • SmplMath macros
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #20 on: January 18, 2013, 05:45:24 AM »
zip the source file and attach it  ;)
Logged
MREAL macros - when you need floating point arithmetic while assembling!

Gunther

  • Member
  • *****
  • Posts: 4198
  • Forgive your enemies, but never forget their names
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #21 on: January 18, 2013, 07:31:04 AM »
Quote from: qWord on January 18, 2013, 05:45:24 AM
zip the source file and attach it  ;)

very good proposal.  :t

Gunther
Logged
You have to know the facts before you can distort them.

qWord

  • Member
  • *****
  • Posts: 1475
  • The base type of a type is the type itself
    • SmplMath macros
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #22 on: January 18, 2013, 12:26:24 PM »
phaap,
is there any special reason why you "cleared" your post? It is maybe that the code doesn't work correct?
At least I was not able to get any useful result (in compare to the test values from your link) - see attachment.

(for assembling you need jwasm, polink and Japhet's WinInc. For the used folder structure, take a look in WinIncRT.inc)
Logged
MREAL macros - when you need floating point arithmetic while assembling!

qWord

  • Member
  • *****
  • Posts: 1475
  • The base type of a type is the type itself
    • SmplMath macros
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #23 on: January 18, 2013, 11:06:21 PM »
OK, at least for the case of an empty string I get the correct result in the low order DWORD of xmm2-5.
It would much be easier if attach your whole code.

regards, qWord
Logged
MREAL macros - when you need floating point arithmetic while assembling!

frktons

  • Member
  • ****
  • Posts: 512
Re: Performance differences for MOVAPS, MOVDQA and MOVAPD
« Reply #24 on: January 19, 2013, 02:16:25 AM »
Quote from: qWord on January 18, 2013, 11:06:21 PM
OK, at least for the case of an empty string I get the correct result in the low order DWORD of xmm2-5.
It would much be easier if attach your whole code.

regards, qWord
Maybe you've to wait until phaap'll pop-up again. His misterious
pop-out has to mean something. Let's wait and see... :lol:
Logged
There are only two days a year when you can't do anything: one is called yesterday, the other is called tomorrow, so today is the right day to love, believe, do and, above all, live.

Dalai Lama
Pages: 1 [2]
« previous next »