Processor for the new Mac Pro only scores 17.63 on cinebench

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Display Modes
  09 September 2013
In other news... I've just read a review of the ATI FirePro W8000, and they got a rather impressive 132 in the OpenGL test of Cinebench.

Assuming that's the card Apple are using, and if the Mac drivers are able to get even close to that - big 'if' - then this would be a significant upgrade.
__________________
Behance | Twitter
- - - - - -
 
  09 September 2013
Nice! That's nearly double last years W9000, so might be a fake. Will be impressive if it's up at there to start.

cheers
brasc
__________________
brasco on vimeo
 
  09 September 2013
Originally Posted by brasco: Nice! That's nearly double last years W9000, so might be a fake. Will be impressive if it's up at there to start.


Yes, I'm suspicious - the source is 3DArtist magazine, it's printed in black and white, in a head-to-head with the new Quadro, so if it's wrong they'll get some flack!

The general conclusion was the nVidia was the better card, and cheaper of course, but certain applications shone with the FirePro, and Cinebench was one.
__________________
Behance | Twitter
- - - - - -
 
  09 September 2013
Originally Posted by ChrisCousins: certain applications shone with the FirePro, and Cinebench was one.

I wonder if they disabled the cheat AMD put in to improve performance. With that in place the performance is better, but you run into issues. Disabeling the special profile will reduce CB performance, but improve the actual work in CINEMA 4D.
__________________
- www.bonkers.de -
The views expressed on this post are my personal opinions and do not represent the views of my employer.
 
  09 September 2013
Actually might be legit, the latest W8000 scores 90+ here.

Although, because the results are skewed by the core speed, I'm not really a fan of the Cinebench OpenGL benchmark.
I wish Nvidia would offer better OpenGL and OpenCL performance for consumer cards, I can't go AMD because of my dependence on CUDA, if I wasn't, I'd seriously consider a FirePro.

cheers
brasc
__________________
brasco on vimeo
 
  09 September 2013
Originally Posted by Srek: I wonder if they disabled the cheat AMD put in to improve performance. With that in place the performance is better, but you run into issues. Disabeling the special profile will reduce CB performance, but improve the actual work in CINEMA 4D.


Ah, bet that's it. I'd heard of special 'optimised settings' ie cheats for standard game benchmarks, but I didn't realise Cinebench had been elevated to that level ;-)
__________________
Behance | Twitter
- - - - - -
 
  09 September 2013
Over the last couple days I've been in contact with some guys at Puget Systems, and I've been able to get some Cinebench numbers for the 20 and 24 core systems, and they're rather alarming:

Dual Xeon E5-2690 V2 (10-core, 3.0GHz): 25.57
Dual Xeon E5 2697 V2 (12-core, 2.7GHz): 22.66

By comparison, the older Dual E5-2687W (8 Core, 3.1Ghz) were scoring between 23.97 and 25.07.

What's even more interesting is that the Geekbench scores returned exactly what you would expect in terms of a speed boost going from the 16-core system to the 20 core to the 24 core.

I e-mailed Maxon about this and received over the course of a few replies the following response; bold text is my questions, italics is the response (this is before I got the scores back for the 24 core system):

EDIT: For some reason, the bold & italics aren't working correctly...but you should be able to follow along.




"Could you shed some light onto the issue as to why the scores came back as they did?"


"This is because some processes in Cinema 4D are still single core processes and cannot take advantage of multiple cores. As the developers have been improving the underlying structure of Cinema 4D they have been making more and more process multi-core processes. It has been a slow transition but things have been changing and will continue to change. The latest process that has been altered to work with multi-core is the deformers within Cinema 4D. I hope this helps explain why the scores were so similar. We do not have a maximum core recommendation (as this would be scene dependant) or a list of all the processes that are still single core unfortunately."




"I understand the fact that many processes are single core only; however - your render engine is completely multithreaded. I was under the impression that Cinebench directly tested the speed at which a given system renders a final product. Is this not true?"

"The render engine is multithreaded but if a process that is being calculated is not, a single core is assigned to the task and this can sometimes slow things down completely as other processes cannot continue until the single core process has finished. Cinebench is testing how the machine renders the two scenes within it. The car and the abstract scene. These scenes are very outdated and still use processes that have not yet been made multi-core."




"If so, why are the Cinebench scores identical for a 16 core system and a 20 core system? Will this new 20 core system render out projects at exactly the same speed as the 16 core unit?"

"This is dependant on the scene file and what is used to build it. If you build a scene in R14 and did not use anything that was upgraded or changed for R15 then you will probably see the same render times. At the same time, if you are using any kind of advanced lighting there are different systems in R15 then R14 and this will change the output time."




"Does Cinebench & Cinema 4D have a render speed ceiling? I,e - once your computer reaches a certain level of speed, it won't create a higher score?"

"I do not believe so but to be honest this is something I would need to ask the developers about. I have nevers seen anything from the developers stating that this is the max score and nothing can go beyond this. I will look into it for you though and see what I can find. I hope this helps."





Well.
 
  09 September 2013
interesting but i dont know what this has to do with the new mac pro unless these exact chips are to be used in that machine for certain ? if so they would not be dual chipsets. This info, while really interesting, is more about overall cinebench chip scores and the interpretation of these scores right ?
 
  09 September 2013
Originally Posted by studiolocal: This info, while really interesting, is more about overall cinebench chip scores and the interpretation of these scores right ?


True; considering 3-4 pages of this thread have been devoted to processor speed in general, I thought it worth sharing.

Last edited by LukeLetellier : 09 September 2013 at 12:29 AM.
 
  09 September 2013
definitely. just checking ! thanks for the info.
 
  09 September 2013
It was worth sharing Luke, I'm interested. This is a good thread for those wondering what cpu's the c4d community is using, even if has turned into a hardware discussion. I take the blame for a lot of it, with my obsession with multiple cores.

I'd be interested to see those benchmarks if Cinebench was as current as R15.
But more on point, the 1st page lists a single 12 core E5 with a score of 17.63

I would wager that doubling that amount of cores would score you around 30. That 24 core score of 22.6 seems very low, like something was not configured properly.

I'll let you know first hand in a couple of weeks. I'll be benching a pair myself.
__________________
Portfolio:https://www.behance.net/Project1Media
 
  09 September 2013
Here are the complete numbers they sent over, including Geekbench:

Xeon E5 2687W (2 CPUs, 8-core each, 3.1GHz):
2 x 8 x 3.1 = 49.6 - just a pure mathematical look at CPUs x cores x clock speed
Cinebench multi-CPU score = 25.09 (one specific sample system; other similar builds scored around this number as well)
GeekBench FP Blur Image - Multi Thread Scalar = 237.3 Mpixels/sec
GeenBench FP Dot Product - Multi Thread Vector = 152.9 Gflops

Xeon E5 2690 V2 (2 CPUs, 10-core each, 3.0GHz):
2 x 10 x 3.0 = 60
Cinebench = 25.57
GeekBench FP Blur Image - Multi Thread Scalar = 341.5 Mpixels/sec
GeenBench FP Dot Product - Multi Thread Vector = 245.4 Gflops

Xeon E5 2697 V2 (2 CPUs, 12-core each, 2.7GHz):
2 x 12 x 2.7 = 64.8
Cinebench = 22.66
GeekBench FP Blur Image - Multi Thread Scalar = 359.4 Mpixels/sec
GeenBench FP Dot Product - Multi Thread Vector = 265.5 Gflops
 
  09 September 2013
Well, it looks bad on paper and it won't do for bragging rights. But I'm sure we won't see these performance dips in real world scenes on versions of C4D that are later than R12.
__________________
Portfolio:https://www.behance.net/Project1Media

Last edited by ThePriest : 09 September 2013 at 03:39 AM.
 
  09 September 2013
Originally Posted by ThePriest: Well, it looks bad on paper and it won't do for bragging rights. But I'm sure we won't see these performance dips in real world scenes on versions of C4D that re llater than R12.


Let's certainly hope so.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright ©2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 06:39 PM.


Powered by vBulletin
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.