2x AMD Opteron 6282 SE benchmarks at SiSoftware

[jhu]

I think that will depend on application tested. Remember David Kanter's analysis? It might vary, whether its server or desktop type applications. In servers, specific compiler optimizations can used for tuning. However in desktop, applications are already pre-compiled (fixed). Thus would be interesting to see how this turns out. :hmm:

Unless you run open source software...

 

[BlueBlazer]

Unless you run open source software...

Like POVray and 7-Zip? Unfortunately majority of the other applications (used for benchmarking) are not open source. 😉

 

[jhu]

Like POVray and 7-Zip? Unfortunately majority of the other applications (used for benchmarking) are not open source. 😉

That is unfortunate. The official POV-ray binary, for example, is dreadfully slow because it's compiled for i386 (at least for the Linux version) and the Windows version is tuned for Intel processors. Fortunately it can be easily recompiled and the resulting binary runs much faster.

 

[BlueBlazer]

That is unfortunate. The official POV-ray binary, for example, is dreadfully slow because it's compiled for i386 (at least for the Linux version) and the Windows version is tuned for Intel processors. Fortunately it can be easily recompiled and the resulting binary runs much faster.

And which compiler did you use? ICC or GCC? :hmm:

Example of PovRay on Athlon II x4: GCC vs ICC

 

[jhu]

And which compiler did you use? GCC? :hmm:

gcc

Some results for Povray 3.6.1 (which is single thread only) on my Athlon II x4, 2.8 GHz

gcc 4.4.5, -march=barcelona, -ffast-math -unroll-loops
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 16 seconds (16 seconds)
Render Time: 0 hours 13 minutes 23 seconds (803 seconds)
Total Time: 0 hours 13 minutes 40 seconds (820 seconds)

Official binary
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 28 seconds (28 seconds)
Render Time: 0 hours 23 minutes 3 seconds (1383 seconds)
Total Time: 0 hours 23 minutes 32 seconds (1412 seconds)

And just to show it's instruction scheduling and not the x87's fault:

gcc 4.4.5, -march=barcelona -ffast-math -unroll-loops -mfpmath=387
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 21 seconds (21 seconds)
Render Time: 0 hours 15 minutes 32 seconds (932 seconds)
Total Time: 0 hours 15 minutes 54 seconds (954 seconds)

Here's comparison with icc 11.1:

icc 11.1, -march=core2
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 18 seconds (18 seconds)
Render Time: 0 hours 13 minutes 26 seconds (806 seconds)
Total Time: 0 hours 13 minutes 45 seconds (825 seconds)

 

[BlueBlazer]

gcc

Its about as fast as ICC at times, and sometimes slower. :hmm:

 

[Idontcare]

gcc

Some results for Povray 3.6.1 (which is single thread only) on my Athlon II x4, 2.8 GHz

gcc 4.4.5, -march=barcelona, -ffast-math -unroll-loops
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 16 seconds (16 seconds)
Render Time: 0 hours 13 minutes 23 seconds (803 seconds)
Total Time: 0 hours 13 minutes 40 seconds (820 seconds)

Official binary
Parse Time: 0 hours 0 minutes 1 seconds (1 seconds)
Photon Time: 0 hours 0 minutes 28 seconds (28 seconds)
Render Time: 0 hours 23 minutes 3 seconds (1383 seconds)
Total Time: 0 hours 23 minutes 32 seconds (1412 seconds)

For AMD architectures I highly recommend The Portland Group compilers :thumbsup:

That said, yeah its rather pathetic that an official binary compile of something as mundane as Povray (we aren't talking about a source pile that takes 3 days per compile run here) is so poorly optimized for today's garden variety architectures.

 

[BlueBlazer]

Some results for Povray 3.6.1 (which is single thread only) on my Athlon II x4, 2.8 GHz

Didn't you mention having a Phenom II X6 earlier? :hmm:

 

[BlueBlazer]

For AMD architectures I highly recommend The Portland Group compilers :thumbsup:

Still not in ICC's class, but getting there. 🙂

That said, yeah its rather pathetic that an official binary compile of something as mundane as Povray (we aren't talking about a source pile that takes 3 days per compile run here) is so poorly optimized for today's garden variety architectures.

Yups, the original binaries are slow. 🙁

 

[jhu]

Didn't you mention having a Phenom II X6 earlier? :hmm:

Haven't installed it yet. Also, I see you've found my posts on their mailing list...

 

[BlueBlazer]

Haven't installed it yet. Also, I see you've found my posts on their mailing list...

What are you waiting for? The X6 should cut down your rendering times by quite a margin. :thumbsup: Yups, I found your posts.. 😀

 

[jhu]

What are you waiting for? The X6 should cut down your rendering times by quite a margin. :thumbsup: Yups, I found your posts.. 😀

It should speed up by a bit more than 33%, but the computer it's going into is in the middle of a render. After that the x4 is going into the file server and will double as the second render node. 10 cores of rendering power! Mwahahahahahaaahahaa!!!111!11!!1!