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
Figaro56 - Tuesday, March 8, 2011 - link
I have 2 XFX HD 5870 cards for sale. I have a double lifetime warranty on these so you get the use of the second lifetime warranty on these. Interested? They are very great performers I can vouch for that. I am use to upgrading my GPU on an annual basis so I am upgrading to 2 HD 6970. $230 each.Thanny - Tuesday, March 8, 2011 - link
Ignoring the inappropriateness of advertising here, I submit:http://www.newegg.com/Product/Product.aspx?Item=N8...
Why would someone pay you $230 for a used product that can be obtained new at $190?
fausto412 - Tuesday, March 8, 2011 - link
I kinda wanted to see a chart with the most common gaming resolution...and can we benchmark with a Q9550 just for comparison? i would love to know if i'm holding back a video card by not going i5 or i7 and by how much.jabber - Tuesday, March 8, 2011 - link
If you can afford a 6990 why would you be bothering using it with a Q9550 at 1680x1050. Hence why it isnt part of this review.This review is to show how it works for the intended market/customer.
As I said before, this card isnt for folks like you (or me for that matter). Sorry.
7Enigma - Tuesday, March 8, 2011 - link
The most common gaming resolution for this card is the one Ryan tested. It is pointless to test at a lower resolution other than possibly true 24" (1920X1200). And even at that res this card is really not needed.Figaro56 - Tuesday, March 8, 2011 - link
BOYA to both of those resolutions. You should be playing your games at 2560x1600. Now that's what I'm talkin about! You'd be saying hell ya.Jorgisven - Tuesday, March 8, 2011 - link
It seems we're getting into the Pentium IV trap, a bit. Big, hot, power-hungry, noisy chips...personally, I'm going to pass on this generation of GPUs. I'm waiting for a revolution in either manufacturing or coding. It's all well and good to have a fast computer for getting what you need done in minimal, but at the risk of the box taking flight because the fans are now of jet engine proportion in speed and power, I'd rather not be able to hear my fans over my headphones...or risk my cat getting sucked into the intake.jabber - Tuesday, March 8, 2011 - link
Well we've kinda got what we asked for. We've all gamely been buying more and more powerful graphics cards with regards to brute force rendering power.We've shown we love buying 750w+ power supplies with multiple GPU connectors, buying SLI and Xfire setups galore.
So the GPU corps think we love nothing more than just piling on more and more power and wattage to solve the situation.
It works both ways.
What we should have been doing was challenging AMD and Nvidia to develop smarter rendering techniques. Had either of them developed PowerVR to the state we are in today we would be in a far better place. Chances are the most power hungry card we'd have today would be 5770 level.
We need something more efficient like PowerVR to take us to the next level.
Less brute force and more finesse.
therealnickdanger - Tuesday, March 8, 2011 - link
Are you waiting to update your test system until the SATA port issue is corrected? Seems to me that anyone wanting to buy this card would also be using an overclocked 2600K... According to the Bench numbers, the 2600K offers roughly 30% more frames than the 920, depending on the game. That indicates to me that your test system is insufficient to properly test this card.Granted, since the vast majority of displays are fixed at 60Hz, fps counts beyond that don't really matter, but I have to wonder what impact this would have on folks with 120Hz-native LCDs. That extra 30% could make the difference.
... just sayin'. :)
Ryan Smith - Tuesday, March 8, 2011 - link
At this point we're waiting on SNB-E. SNB is very nice, but for a GPU testbed the lack of PCIe bandwidth is an issue.