Welcome to the MacNN Forums.

Eug · Jul 29, 2003, 04:46 PM

See here:

Code:
Renderman Finding Nemo benchmark
2.8 Ghz Xeon  icc  4:06 minutes
2.0 Ghz G5    gcc  3:36 minutes

Sorry I don't have any more details. Don't shoot the messenger.

MindFad · Jul 29, 2003, 05:24 PM

Yummy, and it's only going to get a lot better within 12 months.

Socially Awkward Solo · Jul 29, 2003, 05:43 PM

Another one of these? Can't we add it to the other benchmark ones that you started?

MindFad · Jul 29, 2003, 06:22 PM

Good to see some more real world apps than SPEC stuff. I can't wait to give OS X a whirl on a G5 (or two).

Commodus · Jul 29, 2003, 08:07 PM

My question is, what about a 3.06 GHz Xeon? I know that the reason Pixar tried a 2.8 was because that's what they have (I remember them buying a large batch of Linux systems for their render farm), but it might make a bit of a difference here. I suspect that the G5 will still pull ahead - but not by as much.

Either way, I think I know what systems Pixar is going to install for its artists very soon, given who their CEO is.

Superchicken · Jul 29, 2003, 08:13 PM

Proves why I failed math hahaha I pulled out my calculator and was like ha I'll find it.. then I remembered I hated working with minutes and couldn't even remember a good way to get around it... hence why I took the lowest possible level math! GO ME!

cowerd · Jul 29, 2003, 08:41 PM

Siggraph today:

On the shake apple meeting. there was the expected announcement by Pixar of their support of the OSX enviroment.

No other details. From a cgtalk thread.

Eug Wanker · Jul 29, 2003, 09:53 PM

Originally posted by Commodus:
My question is, what about a 3.06 GHz Xeon? I know that the reason Pixar tried a 2.8 was because that's what they have (I remember them buying a large batch of Linux systems for their render farm), but it might make a bit of a difference here. I suspect that the G5 will still pull ahead - but not by as much.

If it was perfect scaling (unlikely) then the Xeon would score about 3:43.

CubeBoy · Jul 29, 2003, 10:59 PM

Originally posted by Eug Wanker:
If it was perfect scaling (unlikely) then the Xeon would score about 3:43.

The new 3.06 GHz Xeon with 1 meg on die cache actually performs slightly better than a Pentium 4 3.2 GHz with a 800 MHz FSB in SPECint and slightly worse in SPECfp if it is any indication as to how it would perform in renderman.

Xeon DP 3.06, 1 MB On-Die L3 Cache
SPECint Base:1242
SPECfp Base: 1173

Xeon DP 2.80
SPECint Base: 1022
SPECfp Base: 1010

All in all a 21% improvement in SPECint base and a 16% improvement in SPECfp base. I don't know if either of these is indicative of how a 3.06 GHz Xeon would perform in Renderman. Does anyone know any details about Renderman? Is it heavy in floating point code/integer code, memory footprint, branch heavy, etc etc.

The testing configurations of both the 3.06 GHz Xeon and 2.8 GHz Xeon are below:

3.06 Xeon SPECint
http://www.spec.org/cpu2000/results/...0630-02332.asc

3.06 Xeon SPECfp
http://www.spec.org/cpu2000/results/...0630-02330.asc

2.80 Xeon SPECint
http://www.spec.org/cpu2000/results/...0407-02046.asc

2.80 Xeon SPECfp
http://www.spec.org/cpu2000/results/...0407-02048.asc

geekwagon · Jul 30, 2003, 02:35 AM

Kinda off-topic but it is kinda strange to me how the 3.06 Xeons are only available in the lower-end servers (at least from HP.) All the better gear tops out at 2.8GHz procs. I wonder if they have problems keeping them cool, or if it is just due to slower product cycles on the higher end machines.

I can't speak for what Dell does on the server end since after being burned there once I wouldn't even consider buying servers from them again...

CubeBoy · Jul 30, 2003, 07:37 AM

Originally posted by geekwagon:
Kinda off-topic but it is kinda strange to me how the 3.06 Xeons are only available in the lower-end servers (at least from HP.) All the better gear tops out at 2.8GHz procs. I wonder if they have problems keeping them cool, or if it is just due to slower product cycles on the higher end machines.

I can't speak for what Dell does on the server end since after being burned there once I wouldn't even consider buying servers from them again...

The 3.06 GHz Xeon DP w/1 meg cache can only be used in dual processor configurations and are specifically geared towards workstations. The fastest Xeon MP or server Xeon tops out at 2.8 GHz but has somewhere around a 2 meg L3 cache.

Eug · Jul 30, 2003, 08:46 AM

Originally posted by CubeBoy:
The new 3.06 GHz Xeon with 1 meg on die cache actually performs slightly better than a Pentium 4 3.2 GHz with a 800 MHz FSB in SPECint and slightly worse in SPECfp if it is any indication as to how it would perform in renderman.

Xeon DP 3.06, 1 MB On-Die L3 Cache
SPECint Base:1242
SPECfp Base: 1173

Xeon DP 2.80
SPECint Base: 1022
SPECfp Base: 1010

All in all a 21% improvement in SPECint base and a 16% improvement in SPECfp base. I don't know if either of these is indicative of how a 3.06 GHz Xeon would perform in Renderman. Does anyone know any details about Renderman? Is it heavy in floating point code/integer code, memory footprint, branch heavy, etc etc.

Well, I'm sure that SPEC results won't be representative here. Plus part of the increase from 2.8 to 3.06 may in fact just be due to improvements in the compiler, etc., and anyway, we have no details on what options Pixar used.

Anyways, I'm not a programmer, but the Ars people say Renderman is FPU heavy, not Altivec'd, and would not likely max out the 533 MHz bus of the Xeon.

saru boy · Jul 30, 2003, 08:54 AM

Was it a dual 2.0Ghz G5 vs a dual 2.8Ghz Xeon, or dual G5 vs single P4, or a single G5 vs single P4?

osxisfun · Jul 30, 2003, 01:52 PM

when apple comes out with a g5 rendering only xserve like they have one for the g4 now (blast) things are going to be peachy keen price wise.

nice....

clarkgoble · Jul 31, 2003, 04:58 AM

Hell, skip the G5 XServe. There will be G5 Blade systems. The XBlade with 16 G5's will be killer!

Eug · Jul 31, 2003, 10:07 AM

Pixar says it's moving to Mac OS X.

osxisfun · Jul 31, 2003, 10:14 AM

anyone got a confirm?