Re: Re: SSE2 benchmarks

[tprinceusa at mindspring dot com]

Paolo Carlini <pcarlini@unitus.it> wrote:
> Hi,

Tim Prince wrote:

> gcc -msse doesn't work with floating point on P-III, as it assumes the
> mmx and x87 registers are distinct.  If you are doing benchmarks
> involving doubles on Windows, you should at least be certain that you
> have built binutils to allow doubles to be aligned correctly in memory,
> and that any use of malloc() in VC is also forced to aligned memory.
> You will probably not be able to get the gcc loop alignments for
> P-II/P-III to work correctly in Windows, but they will be more often
> correct if you increase the binutils alignments from the standard
> 4-bytes.
>
> gcc-3.1 -msse2 has been working correctly for me; this demonstrates that
> it is possible to get 8-byte alignments on Windows with the newer
> binutils, but I think it is not depending on 16-byte alignments.

Thank you very much for your interesting discussion of those fine technical
points.

However, I would like to point out, that, I was talking about a different
kind of comparison, perhaps not completely legitimate, you may object: gcc
(2.95.x, 3.0) running under Linux vs VC6 running under Windoze, both of them
compiling the flops.c benchmark (perhaps the comparison is not that
illegitimate, since the test does not interact with OS services, apart from
the calls to the timer). Also, no SSE or SSE2, since I have only a poor PII.

I agree that I'm a little bit off topic with (my ;) subject line...

But I'm nonetheless curious to know which baseline (no SSE or SSE2) float
(double) performances are expected from gcc versus those others commercial
compilers...

Perhaps you (Prince) have at hand gcc also on Windoze so you are able to
carry out even more meaningful comparisons??? And perhaps you have at hand
also the Intel plug-in compiler for VC!?!

Thanks,
Paolo.

--
Paolo Carlini
Dipartimento di Scienze Ambientali
Università degli Studi della Tuscia
Via De Lellis, snc
I-01100 Viterbo - ITALY
Tel. +39 0761 357027 - Fax. +39 0761 357179


OK, gcc should show its best without so much tinkering when run on linux, but few people find the considerations of choosing among various Windows compilers interesting enough for it to be on topic here.  My experience agrees with you that proper configuration enables the compilers to perform nearly the same in Windows and linux, if you are avoiding the higher overheads of invoking programs from scripts and dealing with files etc. in Windows.  I doubt that the benchmarks which you cite will have much interest for this newsgroup.