 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nicolas Calimet wrote:
>> Official : 45 minutes 08 seconds ( 2708 seconds )
>> gcc pentium4 optimized : 36 minutes 43 seconds ( 2203 seconds )
>> ICC pentium4 optimized : 33 minutes 01 seconds ( 1981 seconds )
>
>
> As was implied by Thorsten's answer, to make a reliable comparison
> you cannot simply use the two-years-old official binary, which was probably
> compiled at best with gcc-3.1.x. You should take the source and recompile
> them with the gcc you are using to produce your pentium4-optimized version.
Pentium 1 (i586) is about 8 years old, so I don't think it still is in
developpement, but I'll recompile it to get better comparison data.
> Also please report the gcc and icc versions as well as the details
> of processor and OS on which you run the benchmarks.
gcc is :
gcc (GCC) 3.2.2 (Mandrake Linux 9.1 3.2.2-3mdk)
Copyright (C) 2002 Free Software Foundation, Inc.
and ICC is v8.0
> Also be sure to
> render
> the benchmark.ini file (and its accompagnying benchmark.pov scene)
> available
> on povray.org at http://www.povray.org/download/benchmark.php
Of course it is ! that's the first thing I have done !
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
>> Also be sure to render
>> the benchmark.ini file (and its accompagnying benchmark.pov scene)
>> available
>> on povray.org at http://www.povray.org/download/benchmark.php
>
> Of course it is ! that's the first thing I have done !
It was not so obvious from my point since benchmark.ini is not
distributed with the povray-3.50c source code (only benchmark.pov).
Thanks for doing the benchs.
- NC
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
"Thorsten Froehlich" <tho### [at] trf de> wrote in message >
> And the gcc and Intel C++ compiler versions happen to be?
>
> I suppose they were more recent versions of the compiler used for the
> official version two years ago? If so, your statement about optimisations
> happens to have no validity at all...
>
> Thorsten
>
They might have no validity from a true benchmarking perspective. However
they illustrate that compiling the source with a recent compiler can improve
performance when compared to using the official binary distribution. While
it is helpfull to understand where exactly the performance gain is comming
from, to me that is secondary to the fact that there is a performance gain.
I can understand how you want to make sure benchmarks are really comparing
apples to apples.
-ross
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nicolas Calimet wrote:
>>> Also be sure to render
>>> the benchmark.ini file (and its accompagnying benchmark.pov scene)
>>> available
>>> on povray.org at http://www.povray.org/download/benchmark.php
>>
>>
>> Of course it is ! that's the first thing I have done !
>
>
> It was not so obvious from my point since benchmark.ini is not
> distributed with the povray-3.50c source code (only benchmark.pov).
>
> Thanks for doing the benchs.
>
> - NC
>
In fact, I didn't know how to run the benchmark and that's why I have
look for this immediately
The test with povray compiled without optimisations on my pc has just
ended : 45 minutes 13.0 seconds (2713 seconds)
3 seconds longer : this is not signifiant.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> The test with povray compiled without optimisations on my pc has just
> ended : 45 minutes 13.0 seconds (2713 seconds)
>
> 3 seconds longer : this is not signifiant.
So if I understand you correctly, you do obtain:
gcc 3.2.2 -march=i586 : 45 minutes 13 seconds ( 2713 seconds )
gcc 3.2.2 -march=pentium4 : 36 minutes 43 seconds ( 2203 seconds )
icc 8.0 -tpp7 : 33 minutes 01 seconds ( 1981 seconds )
Is this right ?
I'd be interested if you could post the full gcc/icc command-line
that you actually used to compile the binaries :-)
- NC
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <40b250b5$1@news.povray.org> , "Ross" <rli### [at] everestkcnet>
wrote:
> They might have no validity from a true benchmarking perspective. However
> they illustrate that compiling the source with a recent compiler can improve
> performance when compared to using the official binary distribution. While
> it is helpfull to understand where exactly the performance gain is comming
> from, to me that is secondary to the fact that there is a performance gain.
This is not the point. The point is that you cannot take some old version
of a compiler that was set to optimise for one processor and compare it with
a newer version of the same compiler set to optimise for another processor
and conclude anything valid about the effect of the optimisations. It is
like taking a 1949 Porsche 356 with a top speed of about 140 km/h and
compare it to a 2004 Golf with a top speed of about 160 km/h (smallest
engine) and concluding that a Golf is faster than a Porsche.
As such, only using the same compiler version can possibly be used to
compare optimisation efficiency. So, looking at the example, and taking any
2004 Porsche, one will quickly find all models have a top speed of over 200
km/h. So, now only one variable is left (the car type) and as such the
comparison is valid.
Thorsten
____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde
Visit POV-Ray on the web: http://mac.povray.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nicolas Calimet wrote:
>> The test with povray compiled without optimisations on my pc has just
>> ended : 45 minutes 13.0 seconds (2713 seconds)
>>
>> 3 seconds longer : this is not signifiant.
>
>
> So if I understand you correctly, you do obtain:
>
> gcc 3.2.2 -march=i586 : 45 minutes 13 seconds ( 2713 seconds )
> gcc 3.2.2 -march=pentium4 : 36 minutes 43 seconds ( 2203 seconds )
> icc 8.0 -tpp7 : 33 minutes 01 seconds ( 1981 seconds )
>
> Is this right ?
>
> I'd be interested if you could post the full gcc/icc command-line
> that you actually used to compile the binaries :-)
>
> - NC
>
The GCC CFLAGS variable was the one used in the original one with only
'i586' replaced by 'pentium4'
the icc CFLAGS variable was :
CFLAGS = -O3 -mcpu=pentium4 -march=pentium4 -ip
(-tpp7 is like -mcpu=pentium4, -march=pentium4 could be replaced by -xP)
-ip adds Interprocedural Optimizations (IPO)
more optimisations could be realized with Profile-guided Optimizations
(PGO) and static linking
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nicolas Calimet <pov### [at] freefr> wrote:
> > #define DISTRIBUTION_MESSAGE_1 "This is an unofficial version compiled by:"
> > Jong
> > #define DISTRIBUTION_MESSAGE_2 "FILL IN NAME HERE...."
> > #define DISTRIBUTION_MESSAGE_3 "The POV-Ray Team(tm) is not responsible for
> > supporting this version."
>
> Given what you did above, I'd suggest you do not try compiling povray,
> but rather use the official Linux binary that should work on your RedHat distro.
>
> - NC
Yes. I had use the official Linux binary and it works pretty good on RedHat.
But as I try to install parallel patch as describe by
http://news.povray.org/povray.binaries.programming/thread/%3C3E40469B.3080402@web.de%3E/
The configuration needs Makefile.am in all POVRAY folder which can only be
found in POVRAY source code.
So, that's why I try to compiling povray source now.
Maybe I have to ask for the binary of parapov...
Is there have another way to configure parapov without compiling the source
code?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
gRRosminet wrote:
>
> I have downloaded the official binary for linux and ran the benchmark.
>
> Official : 45 minutes 08 seconds ( 2708 seconds )
> gcc pentium4 optimized : 36 minutes 43 seconds ( 2203 seconds )
> ICC pentium4 optimized : 33 minutes 01 seconds ( 1981 seconds )
>
> ratios (official / optimized) :
> Official : 1
> gcc pentium4 optimized : 1.229
> icc pentium4 optimized : 1.367
My test results look quite different:
official 3.5: 2394s
3.6 RC1 (gcc 3.4): 2292s
3.6, optimized* (gcc 3.4): 2274s
*) -march=athlon-xp -mfpmath=sse -mmmx -msse -m3dnow
And before someone asks: profiling based optimization also does not
change much about this (i tested this some time ago).
The most likely explanation for your results to me seems the pentium 4
being particularly bad at running code not specifically optimized for
its design.
Christoph
--
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 01 May. 2004 _____./\/^>_*_<^\/\.______
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
In article <c8uv2j$rvd$1@chho.imagico.de> , Christoph Hormann
<chr### [at] gmxde> wrote:
> official 3.5: 2394s
> 3.6 RC1 (gcc 3.4): 2292s
> 3.6, optimized* (gcc 3.4): 2274s
>
> *) -march=athlon-xp -mfpmath=sse -mmmx -msse -m3dnow
>
> And before someone asks: profiling based optimization also does not
> change much about this (i tested this some time ago).
>
> The most likely explanation for your results to me seems the pentium 4
> being particularly bad at running code not specifically optimized for
> its design.
Note: You cannot compare 3.5 and 3.6 benchmark results as anti-aliasing
changed a bit and a photons change makes those much faster. As such, only
the two 3.6 results you posted can be compared.
Thorsten
____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde
Visit POV-Ray on the web: http://mac.povray.org
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
