AMD's Radeon HD 5870: Bringing About the Next Generation Of GPUs
by Ryan Smith on September 23, 2009 9:00 AM EST- Posted in
- GPUs
DirectCompute, OpenCL, and the Future of CAL
As a journalist, GPGPU stuff is one of the more frustrating things to cover. The concept is great, but the execution makes it difficult to accurately cover, exacerbated by the fact that until now AMD and NVIDIA each had separate APIs. OpenCL and DirectCompute will unify things, but software will be slow to arrive.
As it stands, neither AMD nor NVIDIA have a complete OpenCL implementation that's shipping to end-users for Windows or Linux. NVIDIA has OpenCL working on the 8-series and later on Mac OS X Snow Leopard, and AMD has it working under the same OS for the 4800 series, but for obvious reasons we can’t test a 5870 in a Mac. As such it won’t be until later this year that we see either side get OpenCL up and running under Windows. Both NVIDIA and AMD have development versions that they're letting developers play with, and both have submitted implementations to Khronos, so hopefully we’ll have something soon.
It’s also worth noting that OpenCL is based around DirectX 10 hardware, so even after someone finally ships an implementation we’re likely to see a new version in short order. AMD is already talking about OpenCL 1.1, which would add support for the hardware features that they have from DirectX 11, such as append/consume buffers and atomic operations.
DirectCompute is in comparatively better shape. NVIDIA already supports it on their DX10 hardware, and the beta drivers we’re using for the 5870 support it on the 5000 series. The missing link at this point is AMD’s DX10 hardware; even the beta drivers we’re using don’t support it on the 2000, 3000, or 4000 series. From what we hear the final Catalyst 9.10 drivers will deliver this feature.
Going forward, one specific issue for DirectCompute development will be that there are three levels of DirectCompute, derived from DX10 (4.0), DX10.1 (4.1), and DX11 (5.0) hardware. The higher the version the more advanced the features, with DirectCompute 5.0 in particular being a big jump as it’s the first hardware generation designed with DirectCompute in mind. Among other notable differences, it’s the first version to offer double precision floating point support and atomic operations.
AMD is convinced that developers should and will target DirectCompute 5.0 due to its feature set, but we’re not sold on the idea. To say that there’s a “lot” of DX10 hardware out there is a gross understatement, and all of that hardware is capable of supporting at a minimum DirectCompute 4.0. Certainly DirectCompute 5.0 is the better API to use, but the first developers testing the waters may end up starting with DirectCompute 4.0. Releasing something written in DirectCompute 5.0 right now won’t do developers much good at the moment due to the low quantity of hardware out there that can support it.
With that in mind, there’s not much of a software situation to speak about when it comes to DirectCompute right now. Cyberlink demoed a version of PowerDirector using DirectCompute for rendering effects, but it’s the same story as most DX11 games: later this year. For AMD there isn’t as much of an incentive to push non-game software as fast or as hard as DX11 games, so we’re expecting any non-game software utilizing DirectCompute to be slow to materialize.
Given that DirectCompute is the only common GPGPU API that is currently working on both vendors’ cards, we wanted to try to use it as the basis of a proper GPGPU comparison. We did get something that would accomplish the task, unfortunately it was an NVIDIA tech demo. We have decided to run it anyhow as it’s quite literally the only thing we have right now that uses DirectCompute, but please take an appropriately sized quantity of salt – it’s not really a fair test.
NVIDIA’s ocean demo is a fairly simple proof of concept program that uses DirectCompute to run Fast Fourier transforms directly on the GPU for better performance. The FFTs in turn are used to generate the wave data, forming the wave action seen on screen as part of the ocean. This is a DirectCompute 4.0 program, as it’s intended to run on NVIDIA’s DX10 hardware.
The 5870 has no problem running the program, and in spite of whatever home field advantage that may exist for NVIDIA it easily outperforms the GTX 285. Things get a little more crazy once we start using SLI/Crossfire; the 5870 picks up speed, but the GTX 295 ends up being slower than the GTX 285. As it’s only a tech demo this shouldn’t be dwelt on too much beyond the fact that it’s proof that DirectCompute is indeed working on the 5800 series.
Wrapping things up, one of the last GPGPU projects AMD presented at their press event was a GPU implementation of Bullet Physics, an open source physics simulation library. Although they’ll never admit it, AMD is probably getting tired of being beaten over the head by NVIDIA and PhysX; Bullet Physics is AMD’s proof that they can do physics too. However we don’t expect it to go anywhere given its very low penetration in existing games and the amount of trouble NVIDIA has had in getting developers to use anything besides Havok. Our expectations for GPGPU physics remains the same: the unification will come from a middleware vendor selling a commercial physics package. If it’s not Havok, then it will be someone else.
Finally, while AMD is hitting the ground running for OpenCL and DirectCompute, their older APIs are being left behind as AMD has chosen to focus all future efforts on OpenCL and DirectCompute. Brook+, AMD’s high level language, has been put out to pasture as a Sourceforge project. Compute Abstract Layer (CAL) lives on since it’s what AMD’s OpenCL support is built upon, however it’s not going to see any further public development with the interface frozen at the current 1.4 standard. AMD is discouraging any CAL development in favor of OpenCL, although it’s likely the High Performance Computing (HPC) crowd will continue to use it in conjunction with AMD’s FireStream cards to squeeze every bit of performance out of AMD’s hardware.
Facebook
Twitter
RSS
327 Comments
View All Comments
poohbear - Wednesday, September 23, 2009 - link
is it just me or is anyone else disappointed? next gen cards used to double the performance of previous gen cards, this card beats em by a measly 30-40%. *sigh* times change i guess.AznBoi36 - Wednesday, September 23, 2009 - link
It's just you.The next generations never doubled in performance. Rather they offered a bump in framerates (15-40%) along with better texture filtering, AA, AF etc...
I'd rather my games look AMAZING at 60fps rather than crappy graphics at 100fps.
SiliconDoc - Monday, September 28, 2009 - link
Golly, another red rooster lie, they just NEVER stop.Let's take it right from this site, so your whining about it being nv zone or fudzilla or whatever shows ati is a failure in the very terms claimed is not your next, dishonest move.
---
NVIDIA w/ GT200 spanks their prior generation by 60.96% !
That's nearly 61% average increase at HIGHEST RESOLUTION and HIGHEST AA AF settings, and it right here @ AT - LOL -
- and they matched the clock settings JUST TO BE OVERTLY UNFAIR ! ROFLMAO AND NVIDIA'S NEXT GEN LEAP STILL BEAT THE CRAP OUT OF THIS LOUSY ati 5870 EPIC FAIL !
http://www.anandtech.com/video/showdoc.aspx?i=3334...">http://www.anandtech.com/video/showdoc.aspx?i=3334...
--
roflmao - that 426.70/7 = 60.96 % INCREASE FROM THE LAST GEN AT THE SAME SPEEDS, MATCHED FOR MAKING CERTAIN IT WOULD BE AS LOW AS POSSIBLE ! ROFLMAO NICE TRY BUT NVIDIA KICKED BUTT !
---
Sorry, the "usual" is not 15-30% - lol
---
NVIDIA's last usual was !!!!!!!!!!!! 60.69% INCREASE AT HIGHEST SETTINGS !
-
Now, once again, please, no lying.
piroroadkill - Wednesday, September 23, 2009 - link
No, it's definitely just youGriswold - Wednesday, September 23, 2009 - link
Its just you. Go buy a clue.ET - Wednesday, September 23, 2009 - link
Should probably be removed...Nice article. The 5870 doesn't really impress. It's the price of two 4890 cards, so for rendering power that's probably the way to go. I'll be looking forward to the 5850 reviews.
Zingam - Wednesday, September 23, 2009 - link
Good but as seen it doesn't play Crysis once again... :DWe shall wait for 8Gb RAM DDR 7, 16 nm Graphics card to play this damned game!
BoFox - Wednesday, September 23, 2009 - link
Great article!Re: Shader Aliasing nowhere to be found in DX9 games--
Shader aliasing is present all over the Unreal3 engine games (UT3, Bioshock, Batman, R6:Vegas, Mass Effect, etc..). I can imagine where SSAA would be extremely useful in those games.
Also, I cannot help but wonder if SSAA would work in games that use deferred shading instead of allowing MSAA to work (examples: Dead Space, STALKER, Wanted, Bionic Commando, etc..), if ATI would implement brute-force SSAA support in the drivers for those games in particular.
I am amazed at the perfectly circular AF method, but would have liked to see 32x AF in addition. With 32x AF, we'd probably be seeing more of a difference. If we're awed by seeing 16x AA or 24x CFAA, then why not 32x AF also (given that the increase from 8 to 16x AF only costs like 1% performance hit)?
Why did ATI make the card so long? It's even longer than a GTX 295 or a 4870X2. I am completely baffled at this. It only has 8 memory chips, uses a 256-bit bus, unlike a more complex 512-bit bus and 16 chips found on a much, much shorter HD2900XT. There seems to be so much space wasted on the end of the PCB. Perhaps some of the vendors will develop non-reference PCB's that are a couple inches shorter real soon. It could be that ATI rushed out the design (hence the extremely long PCB draft design), or that ATI deliberately did this to allow 3rd-party vendors to make far more attractive designs that will keep us interested in the 5870 right around the time of GT300 release.
Regarding the memory bandwidth bottleneck, I completely agree with you that it certainly seems to be a severe bottleneck (although not too severe that it only performs 33% better than a HD4890). A 5870 has exactly 2x the specifications of a 4890, yet it generally performs slower than a 4870X2, let alone dual-4890 in Xfire. A 4870 is slower than a 4890 to begin with, and is dependent on Crossfire.
Overall, ATI is correct in saying that a 5870 is generally 60% faster than a 4870 in current games, but theoretically, a 5870 should be exactly 100% faster than a 4890. Only if ATI could have used 512-bit memory bandwidth with GDDR5 chips (even if it requires the use of a 1024-bit ringbus) would the total memory bandwidth be doubled. The performance would have been at least as good as two 4890's in crossfire, and also at least as good as a GTX295.
I am guessing that ATI wants to roll out the 5870X2 as soon as possible and realized that doing it with a 512-bit bus would take up too much time/resources/cost, etc.. and that it's better to just beat NV to the punch a few months in advance. Perhaps ATI will do a 5970 card with 512-bit memory a few months after a 5870X2 is released, to give GT300 cards a run for its money? Perhaps it is to "pacify" Nvidia's strategy with its upcoming next-gen that carry great promises with a completely revamped architecture and 512 shaders, so that NV does not see the need to make its GT300 exceed the 5870 by far too much? Then ATI would be able to counter right afterwards without having to resort to making a much bigger chip?
Speculation.. speculation...
Lakku - Wednesday, September 23, 2009 - link
Read some of the other 5780 articles that cover SSAA image quality. It actually makes most modern games look worse, but that is through no fault of ATi, just the nature of the SS method that literally AA's everything, and in the process, can/does blur textures.strikeback03 - Wednesday, September 23, 2009 - link
I don't know much about video games, but in photography it is known that reducing the size of an image reduces the appearance of sharpness as well, so final sharpening should be done at the output size.