If geeks love it, we’re on it

Perspective: Radeon HD 5870 vs FirePro V8750

Perspective: Radeon HD 5870 vs FirePro V8750

ati_firepro_logoOur review of the ATI FirePro V8750 workstation GPU showed that the card is blazing fast at DCC. We tested the V8750 against the venerable NVIDIA GeForce 8800 GTX, and the difference in performance was staggering. But the 8800 GTX is almost four years old, and the question kept pouring in: How would the V8750 fare against the fastest single GPU on earth?

It just so happens that Icrontic recently fitted its test bench with such a GPU, the ATI Radeon HD 5870. And with that, it’s time to finally put the question to rest: How does the best workstation card AMD offers compare to the Radeon HD 5870, a desktop card that’s 100% faster on paper? The results may shock you.

The ATI Radeon HD 5870 is the flagship DirectX 11 GPU. Released on September 23, it is the fastest of the four available GPUs compatible with DirectX 11. As with any new flagship GPU release, gamers and enthusiasts have flocked to NewEgg and other retailers to buy it. It’s fast, it’s efficient, and it answers “yes” to the legendary “Can it play Crysis?” question. But will it blend?

Before jumping into the benchmarks, let’s take a look at the specs for the Radeon HD 5870 and the FirePro V8750:

Radeon HD 5870 specs, FirePro V8750 specs

Testing methodology for the Radeon HD 5870 is identical to what we used with our original FirePro V8750 review. We benchmarked the Radeon HD 5870 in Cinebench R10, SPECviewperf 3ds Max, and SPECviewperf Maya.

The only test we excluded was the general performance in Autodesk Softimage. This is to be a battle of numbers, and the Softimage testing would not communicate the competition well enough.

Enough banter, let’s see the results!
cinebench_r10_8750v5870
We opened with Cinebench R10, and the results were rather surprising. The HD 5870 performed admirably, coming within a stone’s throw of the V8750 in the render tests. We began to question the V8750’s prowess when pitted against a monster of a desktop GPU, but all of those concerns were blown away when we saw how the HD 5870 fared in SPECviewperf’s Maya and 3ds Max tests.

specview_perf_3dsmax_8750v5870
The FirePro V8750 handily overpowered the Radeon HD 5870 by more than twice the average framerate. The Radeon HD 5870 followed with a strong showing of its own, as it doubled the average framerate put up by the GeForce 8800 GTX.

specview_perf_maya_8750v5870
Results in Maya were nothing short of shocking. While the HD 5870 was in a virtual dead heat with the 8800 GTX, the FirePro V8750 quite simply facepounded both of them. In fact, the V8750’s average performance increase over the big, bad Radeon came out to about 650%. Six hundred fifty percent. Madness.

This is the real performance difference between desktop and workstation GPUs. All of the additional technical talk associated with the FirePro line is not minutiae, it yields serious performance.

These tests should put any doubts about the value of a workstation card to rest. The world’s fastest desktop GPU was absolutely crushed by a workstation board with half the theoretical horsepower. For any artist who was considering buying a high-end desktop GPU to cut corners, we think these numbers call for a new plan.

If you’re serious about 3D art and digital content creation, it’s all about making the most of your precious time. As we’ve seen here today, nothing will give you more time to work than the blazing fast technology found in the FirePro.

Comments

  1. Winfrey
    Winfrey Wow! I can understand the premium people are willing to pay for workstation GPUs. When it comes to 3D art and content creation the FirePro is in a league of its own.
  2. Jon The more interesting question is WHERE is the big performance difference coming from? Is it just the extra memory, or even just driver tweaks?
  3. Thrax
    Thrax Jon: We have a complete writeup on the way in the next week, with videos of ATI reps & engineers, explaining just that. Stay tuned!
  4. Jason404 That's a pretty stupid article, as it does not even mention that the FirePro has special optimised drivers for those prgrams. The certified drivers are exactly what you are paying for when you buy a FirePro/Quadro. The hardware itself is the same as the reference designs for the equivalent Radeon/GeForce cards.
  5. Butters
    Butters Intersting results. It would interesting to see how a Quadro FX 4800 compares.

    Also, in the mention writeup, will it also show the V8750 running 3Mark and other "game" related benches to see the inverse side of things.
  6. Thrax
    Thrax No, the hardware is not the same. That is where you're entirely wrong. There are chip and board-level optimizations made to FirePro cores, and driver R&D is only a small portion of the total cost. Certified hardware support, application support, and DCC app R&D plays a significantly larger portion of the overall cost.

    Have the courtesy to know what the facts are before putting your fingers on the keyboard.
  7. Thrax
    Thrax
    Butters wrote:
    Intersting results. It would interesting to see how a Quadro FX 4800 compares.

    Also, in the mention writeup, will it also show the V8750 running 3Mark and other "game" related benches to see the inverse side of things.

    The original V8750 review includes a few game references, but nothing on a chart. Suffice it to say, the V8750 is just barely playable in Team Fortress 2. It gets stomped by the 8800 GTX.
  8. Jason404 I have a couple of FireGL and Quadro cards that are physically exactly the same as equivalent Radeons and Quadros. I looked at pictures of the reference models made by ATi and PNY and saw no difference. The article does not mention the optimised drivers which is what makes the cards so much faster. I do not see any improvement at all when using apps that do not have certified drivers (eg. ArchiCAD). I have never owned a FirePro, but from what I can see, the drivers are what the makes such a difference.
  9. Thrax
    Thrax No, it doesn't mention that because we are preparing a separate article which explores and explains the differences. It's pretty silly to even talk about the drivers when the thesis question is "What happens?" not "Why is it happening?"

    It's a basic case of serving the curiosities and questions of our readers.

    But I can tell you, the FirePro V8750 and the Radeon HD 4870, which are both based on the RV770XT core, are not physically the same. :)

    That's straight from ATI's mouth, and they courteously let us tape the session, which will be a part of the "Why?" article we're preparing.
  10. UPSLynx
    UPSLynx Jason - if you click on the link to the original v8750 piece, it does in fact go into a little more detail on WHY the two types are different.

    Re-stating it for the sake of an article that was meant to showcase performance between ultra-high-end and ultra-high-end would have been a bit needless.
  11. mertesn
    mertesn
    Jason404 wrote:
    The certified drivers are exactly what you are paying for when you buy a FirePro/Quadro.
    Really? Because I just downloaded them for FREE at AMD's web site.
  12. Jason404 @Thrax
    When you say that the Radeon HD 4870 is physically different to the FirePro V8750, are you sure that you are using a reference design from ATi as comparison, and not one from a thrid party manufacturer? I look forward to this article, but I would be sceptical to what the ATi marketing department are telling you. They have to get the money to develop the drivers and would not want people to use them by softmodding.

    @Bobby
    In your previous article you said "PC gaming isn’t impossible on the FirePro V8750, but it certainly isn’t ideal, and it is hardly the task that this GPU was designed for.". Sorry, but this is a fail-statement. The GPU is the same, (and I still all the hardware is too). It is very expensive to develop GPUs, and they are not going to make a special version for such a limited market. It already costs a lot to develop the special drivers.

    @mertesn
    I had to laugh. You really cannot be that dumb.
  13. primesuspect
    primesuspect Jason, if it was so simple as driver optimizations, why wouldn't AMD just let gamers play games on the FirePro series? They're "the same" after all, right? I mean, just tell the driver to go into "game mode" or whatever, right?

    It's not so simple, my friend. Where are you getting your "facts" from?
  14. UPSLynx
    UPSLynx
    Jason404 wrote:
    @Bobby
    In your previous article you said "PC gaming isn’t impossible on the FirePro V8750, but it certainly isn’t ideal, and it is hardly the task that this GPU was designed for.". Sorry, but this is a fail-statement. The GPU is the same, (and I still all the hardware is too).


    Fail-statement? Do you realize what you just said? Even at a driver level, FirePro GPUs are STILL not designed for gaming. Trust me, ATI does not develop this hardware so people can softmod. Not to mention, a soft-modded v8750 will STILL not compare in numbers to it's desktop counterpart. Why? Because at a hardware level, like it or not, they are in fact, different.
    Jason404 wrote:
    It is very expensive to develop GPUs, and they are not going to make a special version for such a limited market. It already costs a lot to develop the special drivers.

    Limited market? Are you serious? Do you even understand how many studios that are in this country, heck, in California alone, that have copious workstations for animators, modelers, artists, that are all powered by workstation GPUs? There is no 'small market' for workstation GPUs. And this hardware isn't limited to animation and VFX studios. Television studios, military, medical imaging (which is one of the biggest industries for graphics today), architecture and machining firms all rely on workstation hardware. The enthusiast GPU market is huge, but I'm willing to bet that the workstation GPU market is not far behind.

    They make money. Oh, yes, they do. It makes workstation GPU development a worthy cause. Studios are replacing hardware regularly. The film industry is still seeing a boom, and studios are still sprouting up all over. They're selling units, and trust me, they have every reason to manufacture a seperate piece of hardware for this market.


    Oh, and leave Mert alone! He's a good guy.
  15. Butters
    Butters Supposing the cards have major differences. Perhaps someone can volunteer their 8800 and softmod it. It would have less ram than the Quadro counterpart, but what would happen?

    http://www.techarp.com/showarticle.aspx?artno=539
  16. primesuspect
    primesuspect Content creation is not the only market for workstation cards, Jason.
  17. mirage
    mirage Nice article and very informative, thanks!

    The discussion about hacking/modding a gaming card into a professional is so unnecessary and irrelevant. I tried softmodding before and it is so problematic that buying a professional card for professional work and gaming card for gaming is the best solution. The manufacturer will always say the hardware is different, hackers will say they are the same. Whatever is correct, softmod is never as good as the real card. Try updating your drivers and be ready for suprises in every version of the driver and software. If you are a professional whose time is really valuable, FireGL/Pro/Stream or Quadro is the only solution *PERIOD*
  18. Jason404 @primesuspect
    I am an architect (just loving this recession), and have been using workstation cards for over a decade. Once I moved from using 3DLabs cards to FireGL I realised that the hardware was the same by taking the HSF off and looking at the reference designs. As a CAD consultant, I have made purchase decisions for dozens of FireGL and Quadro cards.

    The workstation cards do not perform as well in games, as even with the default driver (and not the optimised ones for certain apps), the driver sacrifices frame rates for accuracy. It is okay for games to show glitches and artefacts, but not when modelling in 3D apps.

    @Bobby
    Of course on the driver level FirePros are not designed for gaming. GPU means Graphics Processing Unit. The RT770XT in this case. I do not want to seem condescending, but your comment does seem to make any sense unless you do not know this. Also, the GPUs are the same. They are exactly the same, and that is what essentially makes a graphics card. This is the part that is sold to other companies to make their Radeon cards. They do not make special GPUs for the workstation market. If the workstation cards differ, it is in minor areas. They used to differ in areas like the DAC (Digital to Analogue Converter - 3DLabs cards had very good DACs) but that is not so important these days.

    I know that workstation cards sell in high numbers. I used to work at a firm that bought them by the hundreds, but it is still very small compared to the gaming market. I remember reading an article just about this very subject a few years ago. I think it might have been at Anandtech.

    Guys, it's all about the GPUs, and it is all about the drivers. The hardware is either exactly the same, or essentially the same.
  19. ReD-SpideR
    ReD-SpideR
    Thrax wrote:
    Have the courtesy to know what the facts are before putting your fingers on the keyboard.

    Way to tell him what to do, you're his Mommy I take it?

    Have the common sense to know that the facts are some people don't always know everything fully, unless there taught/told/or research - mind you they don't have to research a subject entirely to comment, hence the a discussion thread.

    He may lack courtesy, but you the common sense.

    Ignorance is shown most by the expectations of those who claim to know more then others. <topic>

    /Rant

    \Topic

    Thank you for the wonderful and insightful last few articles you've put up, I really appreciate them. But the lingering question in my mind, even though they are not targeted at the same level, Will we be seeing a comparison of the ATI FirePro V8750 vs NVIDIA Quadro FX 5800? and if I'm not mistaken, it was mentioned that they may be a small Gaming benchmark, just for kicks?</topic>
  20. pragtastic
  21. Thrax
    Thrax
    ReD-SpideR wrote:

    Have the common sense to know that the facts are some people don't always know everything fully, unless there taught/told/or research - mind you they don't have to research a subject entirely to comment, hence the a discussion thread.

    Defaming the content of an author's hard work based on ignorance or blatant mistruths is a pretty poor way to spur discussion or perform research. ;)

    I do, however, appreciate that you took the time to register and comment. Thank you. Seriously.
  22. QCH
    QCH
    ReD-SpideR wrote:
    <topic>Thank you for the wonderful and insightful last few articles you've put up, I really appreciate them. But the lingering question in my mind, even though they are not targeted at the same level, Will we be seeing a comparison of the ATI FirePro V8750 vs NVIDIA Quadro FX 5800? and if I'm not mistaken, it was mentioned that they may be a small Gaming benchmark, just for kicks?</topic>
    You just wait.... ;)
  23. UPSLynx
    UPSLynx
    Jason404 wrote:

    @Bobby
    Of course on the driver level FirePros are not designed for gaming. GPU means Graphics Processing Unit. The RT770XT in this case. I do not want to seem condescending, but your comment does seem to make any sense unless you do not know this. Also, the GPUs are the same. They are exactly the same, and that is what essentially makes a graphics card. This is the part that is sold to other companies to make their Radeon cards. They do not make special GPUs for the workstation market. If the workstation cards differ, it is in minor areas. They used to differ in areas like the DAC (Digital to Analogue Converter - 3DLabs cards had very good DACs) but that is not so important these days.

    I know that workstation cards sell in high numbers. I used to work at a firm that bought them by the hundreds, but it is still very small compared to the gaming market. I remember reading an article just about this very subject a few years ago. I think it might have been at Anandtech.

    Guys, it's all about the GPUs, and it is all about the drivers. The hardware is either exactly the same, or essentially the same.


    So if you've softmodded yourself, then you know that the performance is not that of an acutal Radeon with the "same" hardware.

    If that was in fact true, it still wouldn't matter. Professional studios would not have time or capacity to deal with the troublesome modding, driver updates, and unreliable performance that comes from softmodding. It doesn't make sense for any professional outlet to take such a risk to cut a corner when they can buy a legitimate card that's built for them.
  24. Cliff_Forster
    Cliff_Forster I'm just curious. What are the architectural differences in the hardware? Is there a good architecture blueprint for each design that explains the hardware optimizations from one to the next?
  25. Ryder
    Ryder
    Thrax wrote:
    But I can tell you, the FirePro V8750 and the Radeon HD 4870, which are both based on the RV770XT core, are not physically the same. :)

    That's straight from ATI's mouth, and they courteously let us tape the session, which will be a part of the "Why?" article we're preparing.
    Let me leave this here :)
  26. Butters
    Butters Its "fascinating" how polarizing the results are with consumer vs workstation graphics cards. I hope the writeup digs deep into both hardware and driver differences. I'm not expecting any type of schematic, but when its that drastic of a difference in performance with arguable hardware differences, flags have to be raised. I'll be patiently waiting.
  27. _k
    _k If they are the same card then why doesn't one of you flash your firepro with a radeon bios, since they are the same it will work. Just sayin.
  28. ReD-SpideR
    ReD-SpideR
    _k_ wrote:
    If they are the same card then why doesn't one of you flash your firepro with a radeon bios, since they are the same it will work. Just sayin.

    If. Which apparently, They are not psychically the same? We'll just sit tight and wait for the interview and figure out if the interviewee is telling the truth.

    I'm more eager to see the FirePro up against the latest Quadro. and the gaming benchmarks against gaming/workstations as well.

    I have the dream that most people in my business have, to be able to work on their 3D models and then be able to play the game there intended on for debugging and testing without the need to swap out cards or use a different machine... but then again... there would be no market in that now would there.
  29. chrisWhite
    chrisWhite I think historically the workstation cards vs consumer cards have often been very similar in hardware. I remember when people were modding the drivers on consumer GPU's to "soft-quadro" their cards. It seems pretty clear that this is no longer the case, but does it really make that big of a difference in the end? nvidia and AMD are selling different products to two different markets that have been engineered and optimized to do two different things, if a lot of that is happening at a driver level then that does not illegitimatize the differences in my opinion.
  30. chrisWhite
    chrisWhite
    ReD-SpideR wrote:
    I'm more eager to see the FirePro up against the latest Quadro. and the gaming benchmarks against gaming/workstations as well.

    Hmmm, I have a feeling the crew has thought about that. ;) I do think this is a very interesting comparison though because a lot of people don't understand the differences between workstation and gaming cards and performance. I find the article and the benchmarks very valuable in making the comparison.
  31. johnfoster fair is fair. where is the Doom vs Doom benchmark. but I kid...

    one thing to remember is that the cost of the card is immaterial compared to the cost of the people running the hardware. if the artist is waiting for screen re-draws that's productivity lost. and if you don't run the settings at most most and instead focus on snap snap you've taken a frustrating part of the jorb (waiting) and made it a non-issue.

    the thing in working with creatives is keeping the bitch factor low. if that means buying them new hardware every nine months so they are happy that's cheaper than higher new knobs because the old stars quit. the new MacBook Pro may cost $2500 but that's cheap comparatively. especially given that the now happy the artist cranks after installing the new.

    you do have to be careful about performance. the writer for example has different needs. he doesn't need FPS to be creative. in fact we don't even need hardware to keep that guy going. that same $1600 buys lots of beer, steaks and cigars which will keep that guy writing for months without complaining.

    (moderator: sorry, I've never ever meet a writer that didn't complain…)
  32. primesuspect
  33. Anand I have a question . What if you compare this 400 dollar 5870 to a workstation card that has a 400 dollar price point. Studios might not have an issue with going fr a 1800 card. But someone like me who is just starting to freelance 1800 is to big. So i would like to see thae same test done with a card theat has the same price point as 5870 and probably one that has the same price point as 4890 the card i have right now
  34. chrisWhite
    chrisWhite That's a good question Anand, I'm not sure what the answer is but I'd tend to go with the gaming card over a super cheap workstation card, I think you'll get a lot more bang for your buck. For one thing you'll be getting a lot less on-board memory on the card and, at least with the Quadro's I've used, a lot less reliability. I'm not the expert around here though and I don't have experience with AMD workstation cards.
  35. Garg
    Garg
    Anand wrote:
    What if you compare this 400 dollar 5870 to a workstation card that has a 400 dollar price point.

    That's a review I'd really like to see, since that's more like the choice I'd have to make if I decide to delve into more 3D work. I don't think a $2000 card is ever going to be in my budget.
  36. UPSLynx
    UPSLynx You guys mean like a Quadro 3800 series workstation GPU?

    Which happens to be enroute to our test bench as we speak.

    Oh, you'll have the numbers you hunger for.

    Also, John Foster, you always have the right words for any occasion. Though I am a write who is also a 3D artist, so I am one of the few who needs that extra umf.

    But I certainly know how to complain, and beer shuts me up, so yes, you nailed it.
  37. Garg
    Garg
    UPSLynx wrote:
    You guys mean like a Quadro 3800 series workstation GPU?

    Oh yeah, I'd like to see that, but I'd still like to see something cheaper, too.

    Check out the variation in the $400 - $500 range on Newegg.
  38. Anand A quadro 3800 would be nice if it falls in the range. Another comparison would be a Ati 5850 and a quadro 17 or 1800. I tihnk they fall in the same place.
  39. chrisWhite
    chrisWhite Note to NVIDIA and AMD, we would like to test your full line of workstation cards :)
  40. ecthelionv I have a nice gaming rig with a Radeon HD 4870. I was thinking about buying a workstation card because my video card causes SolidWorks to crash all the time. I don't really want to build a separate workstation PC. Can I use both cards at the same time in one rig? Or would I have swap the cards each time I wanted to play a game or use SolidWorks?
  41. UPSLynx
    UPSLynx Ecthelionv, you would have to swap the hardware in and out depending on what you want to do. As far as I know, there is no other method of doing it. I keep the desktop card installed most of the time, and swapping to the FirePro isn't too bad with a tool-less chassis.
  42. chrisWhite
    chrisWhite Until he zaps the FirePro and kills it one of these days :)
  43. UPSLynx
    UPSLynx
    chrisWhite wrote:
    Until he zaps the FirePro and kills it one of these days :)

    I turn the thing off and unplug it man! come on, what kind of barbarian do you think I am?! :tongue:
  44. Garg
    Garg
    UPSLynx wrote:
    Ecthelionv, you would have to swap the hardware in and out depending on what you want to do. As far as I know, there is no other method of doing it. I keep the desktop card installed most of the time, and swapping to the FirePro isn't too bad with a tool-less chassis.

    What if you have two monitors? Would the card attached to each monitor behave as expected? Or is that a maybe/sometimes kind of thing?
  45. _k
    _k Driver issue. Both use an ATI driver and most likely that will cause major issues with the system not knowing which driver goes to which card. If you plan to do that I would shoot for a Quadro because you can get drivers from both companies to co-habitat.
  46. Dare
    Dare Great breakdown.
  47. chrisWhite
    chrisWhite
    UPSLynx wrote:
    I turn the thing off and unplug it man! come on, what kind of barbarian do you think I am?! :tongue:

    Trust me, being unplugged does not mean you're not going to zap it with a little static electricity, I'm not too paranoid about it (though I have been through mandatory electrostatic discharge training) but rotating the cards in and out all the time have to increase your chances.

    Really hoping the oven works :(
  48. Thrax
  49. chrisWhite
    chrisWhite Wooooohoooo.

    I wonder if I should try baking my three 7800GT cards that are all on the fritz. Can you kill your card this way?
  50. primesuspect
    primesuspect Hey.....

    /me looks over at Jackie's old dead 7900GT
  51. lordbean
    lordbean
    chrisWhite wrote:
    Wooooohoooo.

    I wonder if I should try baking my three 7800GT cards that are all on the fritz. Can you kill your card this way?

    I would say if they're already on the fritz, there's no harm in trying. You can't really use them when they're unstable anyway, so is it really going to matter if you DO kill them completely?
  52. Annes
    Annes What temp/time for the baking?
  53. primesuspect
    primesuspect Complete writeup coming soon
  54. UPSLynx
    UPSLynx yeesh, give me a little bit to write the aritcle!

    And Chris, you totally jinxed me man. Just a few hours after I posted the witty retort... ZAP.

    I ground myself extensively and everything, but some time, the crap just needs to hit the fan.
  55. photodude
    photodude Thanks for this great article, this really shows the difference between the two levels of video cards.

    I hope to see a future article pitting the top Nvidia gaming GPU against the QuadroFX line. And one pitting the QuadroFX line against the FirePro line.

    I also hope to see benchmarks for some Adobe CS4 products and Autodesk Revit products, and video encoding/decoding.

    I look forward to your future article next week, with videos of ATI reps & engineers, explaining the differences.

    Again Thanks for doing what no one else has done!!! =D
  56. loklomedia3d I'm still waiting for a good directX (3ds max) comparison of desktop vs workstation cards at similar pricepoints - ie HD4870 vs v3750. I've seen charts from each which suggest the v3750 is better, but I want some more proof before I consider switching. And softmodding the 4870 to 8750 isn't as good as a real 8750, but would it be better than a 3750? (Obviously I've got a 4870 & Vista 64).
  57. Digiphix
    Digiphix WOW, this is a really good post. I just wantedto add my 2 cents.

    First of all content creation normally favors the open gl language to render shadows, textures and all the other mumbo jumbo. Open GL use to be the premier toolkit back in the day. Any Graphics card that has the Open GL implementation will work with Revit and AutoCAD and by the sounds of it Solid Works too.

    Now, the video game card can render the Open GL language but the Workstation card has been engineered to carry out the job with much greater quality and performance. Create a model in Revit turn shadows on and and change the views quickly, the shadows will go away and will be quickly re-rendered on the game card but the workstation card will always give you better speed and higher resolution and shadows will always be displayed.

    -Can't go wrong with the HD 5770, open GL 3.0, low power consumption (25w idle and 118w at full load), plays Crysis (Don't get carried away with the settings) and DirectX 11 for gaming (will be the standard for the next 2 years)

    -Any gigabyte mobo with the Ultra Durable 3 feature. Power effecient and less heat.

    -Stuff as much RAM as yo can

    -80+ Silver PSU. I recommend a 600w unit from Corsair

    -go with any processor with a L3 cache. the differences with Intel and AMD nowadays are trivial really. You will like either one just be sure to get the L3 cache.

    -If you decide not to game, the Quadro FX 580 (512mb) will handle the load and help your renderings pop.The Quality and performance will definitely be higher than a gaming card.
  58. photodude
    photodude
    Digiphix wrote:
    Any Graphics card that has the Open GL implementation will work with Revit and AutoCAD and by the sounds of it Solid Works too.

    As of version 2010 Revit uses D3D and DirectX10 for the primary rendering. Open GL which was big up to revit 2009 was reduced in the amount of openGL acceleration 2010 due to a plague of video card problems thanks to openGL. Directx10 is the way forward in revit.
    Digiphix wrote:
    --Can't go wrong with the HD 5770, open GL 3.0, low power consumption (25w idle and 118w at full load), plays Crysis (Don't get carried away with the settings) and DirectX 11 for gaming (will be the standard for the next 2 years)

    Thanks to different hardware and program specific drivers Workstation cards have a big leg up over any gaming card even the HD 5870 wouldn't perform as well a a lower end workstation card for Digital content creation or CAD work.

    keep gaming cards in the gaming world, and workstation cards in the work world. it's not advisable to cross use the cards (although it can be done there is a big performance hit in cross use)
  59. Erol The SPECviewperf benchmark is not real it seems. It's probably highly optimized in the drivers like ATI and NVIDIA also did with the 3DMark benchmark a few years ago.

    I have a Quadro FX 570 and a ATI HD 2900 XT. When I run SPECviewperf the Quardo FX 570 really smokes the HD 2900 XT. However. I tested the 2 cards on my own 3ds max projects the HD 2900 XT has always a higher fps. I tested a scene where the camera is rotating around a mechanical device with moving parts. The Quadro FX 570 got to max 22 fps and the HD 2900 XT almost double with 41 fps.

    So when it comes to practical use I prefer a cheaper gamecard and invest in a faster cpu to cut down rendering times. Graphics cards don't affect the rendering times.

    The 3ds max viewport quality is more or less the same on both cards.

    I used the nVidia 3ds max performance driver for the Quadro FX 570 and the default DirectX 9 3ds max driver for the HD 2900 XT.

    I softmodded the HD 2900 XT to a FireGL V7600 too but that only added a few fps so not really worth the trouble. I did very well in SPECviewperf tho. Maybe a real V7600 is faster but I don't know.

    Cinebench is strange too. My HD 2900 XT scores +8500 points in the OpenGL test. A lot higher than the cards tested here.
  60. photodude
    photodude
    Erol wrote:
    So when it comes to practical use I prefer a cheaper gamecard and invest in a faster cpu to cut down rendering times. Graphics cards don't affect the rendering times.

    depends on the program, Some programs Rendering is programed for graphics cards, other programs rendering is primarily CPU. You have to know which is better for your program (sometimes it's hard to find accurate information to know which is better)

    FYI: the QuadroFX 570 is a really low end card the workstation line. With only 16 cuda cores and 256MB of memory; I'm not too surprised your HD 2900 XT is kicking it's trash. The 2900 has 320 stream processing units and double the memory at 512MB. The larger memory is your friend in this case.

    Softmodding the 2900 doesn't really get you a FireGL V7600, your only forcing the FireGL V7600 driver to run on the 2900 but the hardware and GPU are different on the two cards. the only thing you got out of softmodding is a glitchy graphics card that wouldn't perform as well as a real FireGL V7600.

    I personally set the entry level of workstation cards at the QuadroFX1800 or FirePro v7750
  61. UPSLynx
    UPSLynx FirePro v7750 for entry level? No way dude, not at all. The v7750 is a high end solution. I'm working on a review for it right now, and that card has topped the Quadro FX 3800 in every single benchmark so far. It is an outstanding product. At $899, it's not an entry level piece. The Quadro FX 1800 is an accurate entry level product though.

    You're right on the Quadro FX 570 though, Photodude. That kind of GPU is primarily used for 2D graphics production and applications like nonlinear video editors. Hardly a good solution for 3D graphics, which is exactly as you said, the reason why the gaming card is beating it.
  62. photodude
    photodude
    UPSLynx wrote:
    FirePro v7750 for entry level? No way dude, not at all. The v7750 is a high end solution.

    ya, I've been trying to figure what a good entry level for the Firepro line would be. I chose the v7750 based on it's 1GB GDDR5 of memory which was the closest to the QuadroFX1800 which has 768MB GDDR3. My other option was the v5700 with 512MB GDDR3

    I still wouldn't call the v7750 a high end solution, Based on your comment; I would put it as midrange with the QuadroFX3800 (even if v7750 is beating the fx3800)
    UPSLynx wrote:
    At $899, it's not an entry level piece.
    You can get a v7750 for about $600 which is more on the entry level "workstation" price with better then entry level performance.
  63. chrisWhite
    chrisWhite
    photodude wrote:
    depends on the program, Some programs Rendering is programed for graphics cards, other programs rendering is primarily CPU. You have to know which is better for your program (sometimes it's hard to find accurate information to know which is better)

    Right now there are very few GPU accelerated renderers, Gelato was the only real player until the last year or two and while there are a bunch of GPU based render engines in the works there still aren't that many that have actually hit the market yet. Especially non-proprietry hardware based ones like the Caustic Graphics solution.
  64. Martijn Hey Bobby Miller,
    You wrote that you didn't test it with Softimage. Do you happen to know which SPECViewperf benchmark would resemble Softimage's performance best that of Max or Maya? Any clue?
  65. UPSLynx
    UPSLynx Hello Martijn,

    I was able to find a user script that runs an FPS benchmark in XSI, so if you have the software, you can benchmark with it:

    http://www.xsibase.com/forum/index.php?board=15;action=display;threadid=36978;start=0

    I got it to work in XSI 6.5 without problem. Pretty sweet benchmark. If you try to run the script in Softimage 7 or higher, some edits need to be made to the script. Edits are listed in that forum thread a few posts down.

    Now, if you don't have Softimage, It's hard to say which benchmark on SPEC Viewperf would be closest. Out of all of the included benchmarks, 3DS Max and Maya are the two that are going to be closest. Now, Maya has been extremely refined, as it is the defacto standard in 3D. From my own experiences in Softimage with workstation GPUs, I have a feeling benchmark results would be closer to 3DS Max than they would to the astronomical results in Maya.

    That is, of course, all speculation.
  66. UNBIASED @ Photodude, no kidding the Quadro 570 is a low end card you hack- obviously it's in the same price range though so wouldn't it suffice to compare the 2? instead of a $400 card vs a $1,500 card? This article is lame and biased! Next time try comparing things by their price tag as most consumer reports would do and not by their meaningless manufactering number!
  67. Martijn Hi Bobby,

    Thanks for your response. The tool on XSI Base will only let me compare scores of Softimage on different configurations. I'm more interested in the benchmarks which are run in different tests like SpecView so it's easier to compare different GPU's instead of pc-setups.

    I have a FireGL V7600 right now and nowadays you can get a FireGL v7700 or a newer Quadro FX 3700 for prices along those of high-end gamecards (with twice the RAM and GB/s). I reckon a current generation workstation card performs better than a current gamecard (like your test shows) but I'm looking for the best bang for the buck and a HD 5870 might outperform the V7700 and FX3500 even though the drivers and OpenGL performance are not optimized. Stability is also a factor of course but there are even 3d companies that run on gamecard in stead of Quadro's and FirePro's and gamecards do support later shadermodels and DirectX.
  68. UPSLynx
    UPSLynx Martijn,

    You may find interest in the numbers on our recently benchmarked FirePro v7750. The v7750 is the incremental step from the v7700. It performed incredibly well in Maya.

    http://tech.icrontic.com/articles//ati-firepro-v7750-review/

    It's an outstanding product with great all around performance, and has become the go-to product I recommend to anyone sitting on the fence.

    Though, then again, if you aren't looking to pay close to one thousand dollars, you may be better off with a desktop grade card. Myself, I wouldn't skimp myself as an atist after testing these GPUs. I do, however, understand the issues that the financial implications can bring.

    I'm still looking for a better Softimage solution for benchmarks...
  69. Dantist Купил карту HD 5870 скажу, что она довольно таки быстрая, но у меня стояла задача работать в 3dMAX и при этом наслаждаться новыми играми. Так вот скорость и качество рендеринга в 3dMAX меня порадовала особенно под Windows 7, соответственно игры тоже летают. Но при моделировании сложных объектов все равно наступает придел, когда работать уже невозможно - поэтому все-таки прокарты тут выигрывают, но только за счет драйверов. Надеятся на STREAM нельзя, а самому прописать дрова для оптимизации 3dMAX - это потянет на Нобелевку)
  70. primesuspect
    primesuspect
    Dantist wrote:
    Купил карту HD 5870 скажу, что она довольно таки быстрая, но у меня стояла задача работать в 3dMAX и при этом наслаждаться новыми играми. Так вот скорость и качество рендеринга в 3dMAX меня порадовала особенно под Windows 7, соответственно игры тоже летают. Но при моделировании сложных объектов все равно наступает придел, когда работать уже невозможно - поэтому все-таки прокарты тут выигрывают, но только за счет драйверов. Надеятся на STREAM нельзя, а самому прописать дрова для оптимизации 3dMAX - это потянет на Нобелевку)

    Translation:
    I bought a HD 5870 card will say that it is still pretty fast, but I had a goal to work in 3dMAX while enjoying the new games. So the speed and quality of rendering 3dMAX pleased me especially under Windows 7, respectively, the game also fly. But when modeling complex objects still comes aisle when work is no longer possible - so still prokarty win here, but only at the expense of drivers. STREAM can hope for, and prescribe the most wood for optimization 3dMAX - it will pull on Nobelevku)
  71. Martijn Hi Bobby, I'm sure the FirePro 7750 performs well but the FireGL V7700 had a price cut recently and is available for less than 200 dollar which seems like really good value.
    Still curious if a HD5870 would out perform a V7700 in DCC though.
  72. haronharon
    haronharon
    Anand wrote:
    I have a question . What if you compare this 400 dollar 5870 to a workstation card that has a 400 dollar price point. Studios might not have an issue with going fr a 1800 card. But someone like me who is just starting to freelance 1800 is to big. So i would like to see thae same test done with a card theat has the same price point as 5870 and probably one that has the same price point as 4890 the card i have right now

    http://www.youtube.com/watch?v=Y99mD_A1uug

    BTW Guru3d has a great thread about softmodding Radeon cards to FireGL/PRO.
    just google it up...

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!