AMD's Radeon HD 6990: The New Single Card King
by Ryan Smith on March 8, 2011 12:01 AM EST- Posted in
- AMD
- Radeon HD 6990
- GPUs
Compute Performance
Moving on from our look at gaming performance, we have our customary look at compute performance. With AMD’s architectural changes from the 5000 series to the 6000 series, focusing particularly on compute performance, this can help define the 6990 compared to the 5970. However at the same time, neither benchmark here benefits from the dual-GPU design of the 6990 very much.
Our first compute benchmark comes from Civilization V, which uses DirectCompute to decompress textures on the fly. Civ V includes a sub-benchmark that exclusively tests the speed of their texture decompression algorithm by repeatedly decompressing the textures required for one of the game’s leader scenes.
New as of Catalyst 11.4, AMD’s performance in our Civilization V DirectCompute benchmark now scales with CrossFire at least marginally. This leads to the 6990 leaping ahead of the 6970, however the Cayman architecture/compiler still looks to be sub-optimal for this test. The 5970 has a 10% lead even with its core clock disadvantage. This also lets NVIDIA and their Fermi architecture establish a solid lead over the 6990, even without the benefit of SLI scaling.
Our second GPU compute benchmark is SmallLuxGPU, the GPU ray tracing branch of the open source LuxRender renderer. While it’s still in beta, SmallLuxGPU recently hit a milestone by implementing a complete ray tracing engine in OpenCL, allowing them to fully offload the process to the GPU. It’s this ray tracing engine we’re testing.
There’s no CrossFire scaling to speak of in SmallLuxGPU, so this test is all about the performance of GPU1, and its shader/compute performance at that. At default clocks this leads to the 6990 slightly trailing the 6970, while overclocked this leads to perfect parity with it. Unfortunately for AMD this is a test where NVIDIA’s focus on compute performance has really paid off; coupled with the lack of CF scaling and even a $240 GTX 560 Ti can edge out the $700 6990.
Ultimately the take-away from this is that for most desktop GPU computing workloads, the benefit of multiple GPU cores is still unrealized. As a result the 6990 shines as a gaming card, but is out of its element as a GPU computing card unless you have an embarrassingly parallel task to feed it.
Facebook
Twitter
RSS
130 Comments
View All Comments
MarkLuvsCS - Tuesday, March 8, 2011 - link
Thanks for an awesome article!Minor typo in section "ONCE AGAIN THE CARD THEY BEG YOU TO OVERCLOCK" second to last paragraph second sentence says "...the 6690OC’s core clock is only 6% faster and the memory clock is the same, versus..."
Figaro56 - Tuesday, March 8, 2011 - link
Yes this is the article I as waiting for. Time to get rid of my 2 HD 5870 cards and purchase 2 HD 6970 ones. I wouldn't get an HD 6990. That is pretty clear.Thanks AnAndTech!
mino - Tuesday, March 8, 2011 - link
AT has CHOSEN to not overclock the card based on its THEORETHICAL (Furmark) load temperatures ...Go bash AT for writing "OC" on the slides while they enabled ONLY the performance BIOS. Not doing ANY overclocking whatsoever by fear of Furmark ...
In effect what they have done was in effect a factory OC, not a traditional OC of the what-it-can-handle kind.
Great, so Furmark has achieved one more evil goal: it prevents (AT?) journalists to do overclocking reviews ...
mino - Tuesday, March 8, 2011 - link
Here come some real OC numbers: www.legitreviews.com/article/1566/14BTW, they did not even bother with the #1 BIOS option to achieve it ... so, lets talk about biased reviewing, shall we?
RaistlinZ - Tuesday, March 8, 2011 - link
Looks like the 2x6950 is a much better option, given you'll have much less noise to deal with and that they can be flashed to 6970 shaders.If this card had been $599 I probably would have picked one up. But at $699 I think I'll just wait for 28nm generation of cards.
Thanks for trying, AMD.
MarcHFR - Tuesday, March 8, 2011 - link
Hi,Drivers used are :
NVIDIA ForceWare 262.99
NVIDIA ForceWare 266.56 Beta
NVIDIA ForceWare 266.58
AMD Catalyst 10.10e
AMD Catalyst 11.1a Hotfix
AMD Catalyst 11.4 Preview
Is it possible to know wich driver is used for each card ?
Thanks
jcandle - Tuesday, March 8, 2011 - link
Ryan, any chance you'll be doing a thermal compound review soon? 8% against their stock compound. How much better is it than current performance aftermarket compounds?IanCutress - Tuesday, March 8, 2011 - link
Quite difficult to get accurate thermal compound numbers. There's no way you can guarantee that the compound will be spread evenly and accurately every time. Any big 8ºC differences will show sure, but you're always playing with statistics to +/- 3ºC. Then there's the inevitable argument about the right way to apply the paste...7Enigma - Tuesday, March 8, 2011 - link
More importantly is the normal compound most manufacturers use is junk compared to a good thermal compound such as arctic silver (don't keep up on the latest brands as I still have Arctic Silver 3 that works great for me). So that 8% might very well be true since the normal stuff is of poor quality.ypsylon - Tuesday, March 8, 2011 - link
But few issues need to be addressed. Noise for starters, nearly 80dBA. Thats like working in a foundry. Also cooling is highly inefficient for card of this size. Need some 3rd party solution or water cooling altogether.Biggest problem for 6990 could be (or rather will be) nVidia. If they price GTX590 at the same level or even below $700 price tag then AMD will be screwed totally. For now waiting for GTX590 and 6990 with some after market coolers as stock solutions are completely unacceptable.
One thing straight - I do not sleep on ca$h and if I buy 6990/590 it will be ma$$ive expense for me, but... What swings things for me with cards like this, is that I do not need uber VGA for 30 monitors. All I want is card with large frame buffer, which will live in my PC for ~10 years without need to upgrade, and it will occupy only 1 PCI-ex x16 slot. SLI/CF is totally misguided if you do have some more hardware installed inside. Sometimes (with all that SLI/CF popularity) I wonder, why 7 slot ATX is still alive and 10-12 slot motherboards are not a standard?