Comments on: Review: AMD’s ATI FirePro professional GPUs

By: FREDDO71

FREDDO71 — Fri, 09 Mar 2012 07:29:40 +0000

Very nice post.
It’s very interesting to make a choice, and here is the essential for.
Big thanks.

By: gescom

gescom — Wed, 25 May 2011 16:09:40 +0000

What a great review! A real world performance instead of predefined synthetic benchmarks. Please more of those! Thanks!!

By: Panos Greg

Panos Greg — Fri, 11 Feb 2011 13:49:25 +0000

Your review is superb, i was looking for something like this for so much time, it’s the only one that does a proper benchmarking. The most important thing is that you do not rely on ready benchmarks like specviewperf and that is what it sets it apart (the viewport engine in specview for each program does not reflect the reality for the current program engines like those on the 2011 releases) . The whole essence of benchmarking such cards is to see it for yourself on real world scenarios like the ones you shown (various scenes with various polygon counts) and most importantly on latest software like the 2011 releases from autodesk and at the end of the day write down the fps count for each case in order to see how much apples that busket can fill.
There is only 2 things i wanted to ask/request if there is ever a possibility. If you could add the quadro fx 4000 in the benchmark (i think it’s the best value for money at this time, the 5000 is way too expensive and not that much better, while even though the 2000 is affordable its low memory capacity and low memory performance do not cut to do the job on the long run). The 2nd thing i would like is some more scenes/cases on maya 2011 and especially in viewport 2.0 and finally if you could add modo 501 in the mix, it would be great.
All in all the best thing about the whole review is the fact that each application has differerent programming implementations about the viewports and hence we see vast differences on the performance. So doing a hands-on real usage of each application is the only way to see the real performance when using each particular application, which is exactly what you did, and it is a vast help for anyone looking to buy a card. Way to go man!

By: Simon Gregory

Simon Gregory — Fri, 07 Jan 2011 10:21:21 +0000

id be interested in seeing a comparison that takes the quadro 4000 5000 and 6000 into account, they are signicantly better than the fx cards, also would be great to see what happens when you take SLI/crossfire into account, i hear max cant avail of multi cards but id be interested in seeing a study about this.
a friend is currently building a machine for me, 2 x quadcore xeon 4.0 ghz i think 16 cores total 8 of which are virtual , 24gb ddr 3 ram, 1 x gtx460 (temporary), 7200rpm hard drive, waiting to get new card either a quadro 4000 or a firepro v8800.
would love advice cause i really am not sure which one to get, use max, mudbox, zbrush mainly

By: Ozzy

Ozzy — Tue, 04 Jan 2011 11:43:37 +0000

I just noticed Softimage is way faster than maya and max.. 10.4M pol at 59FPS for Softimage, maya with 6.8M at only 9 fps, and max with 2.4 M at 49fps..

By: Jacob Abreu

Jacob Abreu — Wed, 22 Dec 2010 16:15:40 +0000

I did a test my self in mudbox I went up to 131 million polygon 57 FPS viewport and 46 FPS editing using i7 3.06 ghz 12gigs ram 7200rpm hard drive FirePro V7800 how can your test contribute that little amount of polygon at a low frame rate during editing when your using a more powerful machine than mine

By: David Jones

David Jones — Wed, 15 Dec 2010 10:36:24 +0000

You know jim, i understand you dont want your material floating. No problem there. Glad to see you take the comments seriously. I think the matrix model, untextured was free and it would help people test what they have vs the results shown and add to the “results” of what people can expect with different cards. So people here become part of the test

Then again I sure someone like one of the 3d model web companies might help you here too. Then you can see what a difference getting a card like this would make. Right now its just a number, and you have no way to compare your system to what jason has done. Because you don’t have the same testing models.

So while its a solid tst, its really in a vacumn, i’d like it to be more open.

By: Jim Thacker

Jim Thacker — Mon, 13 Dec 2010 22:16:07 +0000

Posted on behalf of Jason Lewis, who is having login problems (long response, broken into separate posts):

“And lastly, in response to David Jones, You make a very good point, and setting up a universal test scene that can be used across all cards and all applications is a great idea, and something that I will be looking into for future articles. As for providing links to the test scenes, all of my testing is done with mostly models & scenes from my own personal portfolio and a few donated by friends and colleagues, and while you do make a good point, I am not too comfortable with my work or the work of my donors floating out all over the internet.

“Perhaps when I set up an actual test scene as I mentioned above, I can make that one available for people to download and test themselves, and perhaps we can set up a sort of database to compare how individual users scores compare to mine. Also, if anyone has any scenes they would like to donate to the benchmarking cause, we can include those as well as a reference to the performance you are getting with it and how it compares to the performance scores I get with it. Anyone who donates testing material would of course get full credit for it in the article.”

You can contact Jason directly at jason@cgchannel.com

By: Jim Thacker

Jim Thacker — Mon, 13 Dec 2010 22:15:36 +0000

Posted on behalf of Jason Lewis, who is having login problems (long response, broken into separate posts):

“In reply to Jaymie – I apologize, but yes, the glitches are not completely absent with the pro cards, they are just far less frequent, and when glitches are identified, they are usually fixed for the pro cards quickly. This holds true for both the ATI FirePro cards, and Nvidia Quadro Cards in my experiences. An example of this was a viewport display bug in MAYA from versions 2008 + that caused massive viewport display corruption when used with Nvidia Geforce 8800 and higher cards. (I personally saw it happen with 8800 / 9800 / GTX260 / & GTX280 cards.) It has since been fixed, but it took quite some time to implement as Nvidia does not certify any GeForce card to work with DCC applications. Geforce & Radeon cards will work with them, you just won’t usually get any support if you encounter any issues as the response to your issue will usually be: that cards is not certified for use with that application. Again, this holds true for both Nvidia & AMD.”

By: Jim Thacker

Jim Thacker — Mon, 13 Dec 2010 22:05:36 +0000

Posted on behalf of Jason Lewis, who is having login problems (long response, broken into separate posts):

“First of all, several of you have noted how much faster Softimage is than Max & MAYA, and this is not a mistake or typo. Now I am not a Softimage power user, but it has always been a lot faster than other DCC apps since I started looking at back with version 7. They have some kind of viewport display technology that makes it very fast, on both professional & consumer graphics cards. Myself and others in the business speculate that this is partially why Autodesk bought Softimage. There is the rumor going around that they intend to implement that technology in Max & MAYA at some point, but again, this is just rumor & speculation.