• What
    is this?
    You've landed on the AMD Portal on AnandTech. This section is sponsored by AMD. It features a collection of all of our independent AMD content, as well as Tweets & News from AMD directly. AMD will also be running a couple of huge giveaways here so check back for those.
    PRESENTED BY

Compute Performance

Moving on from our look at gaming performance, we have our customary look at compute performance. Since compute performance is by definition shader bound, the 7950 is at a bit of a disadvantage here compared to gaming performance. Whereas ROP performance scales with the core clock, shader performance is hit by both the reduction in the core clock and the disabled CU array.

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. Note that this is a DX11 DirectCompute benchmark.

AMD’s greatly improved compute performance continues to shine here, though in the case of Civilization V it’s largely consumed by just closing the previously large gap between the GTX 500 series and the Radeon HD 6000 series. As a result the 7950 falls ever so short of the GTX 580, while the factory overclocked Sapphire and XFX cards give the 7950 enough of a push to come within 5% of the 7970.

Our next benchmark is SmallLuxGPU, the GPU ray tracing branch of the open source LuxRender renderer. We’re now using a development build from the version 2.0 branch, and we’ve moved on to a more complex scene that hopefully will provide a greater challenge to our GPUs.

Under SmallLuxGPU the 7970 enjoyed a large lead over the GTX 580, and this continues with the 7950. Even though the 7950 is well behind the 7970—to the tune of 24%—it’s still 33% ahead of the GTX 580 and the lead only grows from there. Meanwhile the XFX and Sapphire cards can catch up to the 7970 somewhat, but as this is truly a shader-bound test, you can’t make up for the lack of shaders units on the 7950.

For our next benchmark we’re looking at AESEncryptDecrypt, an OpenCL AES encryption routine that AES encrypts/decrypts an 8K x 8K pixel square image file. The results of this benchmark are the average time to encrypt the image over a number of iterations of the AES cypher.

In spite of being a compute benchmark, AESEncryptDecrypt is not particularly sensitive to GPU performance, showcasing the impact that setup times can have. The 7950 trails the 7970 by 10%, and overclocking doesn’t change this much. Unfortunately for AMD NVIDIA is still the leader here, showing that AMD’s compute performance still has room to grow.

Finally, our last benchmark is once again looking at compute shader performance, this time through the Fluid simulation sample in the DirectX SDK. This program simulates the motion and interactions of a 16k particle fluid using a compute shader, with a choice of several different algorithms. In this case we’re using an (O)n^2 nearest neighbor method that is optimized by using shared memory to cache data.

With the compute shader fluid simulation we once again shift back into a compute task that’s much more shader-bound. The 7950 only reaches 80% of the performance of the 7970, once more proving the real impact of losing a CU array. This is still enough to handily surpass the GTX 580 however, with the 7950 taking a 15% lead.

Civilization V Power, Temperature, & Noise
POST A COMMENT

259 Comments

View All Comments

  • MattM_Super - Tuesday, January 31, 2012 - link

    Yes I would enjoy more performance (esp at a lower power draw). I like to get ~60fps minimum @1080p. In games with fancy lighting effects, 4xAA, high res textures, like Witcher 2, metro 2033, crysis 2 dx11 current cards cant deliver that. Its a luxury sure, but one I am willing to pay for.
    I also think there is still plenty of room for improvement in graphics.
    Reply
  • sleepeeg3 - Wednesday, February 15, 2012 - link

    Bingo. The fanboys are too focused on bickering to realize these GPUs are blowing past 100fps on popular games. Who cares? Game companies stopped pushing the limits years ago. Reply
  • chizow - Tuesday, January 31, 2012 - link

    Its funny because next-gen single-GPU performance *USED* to be measured against previous gen x2 GPU card or CF/SLI where we would expect 50+% increase over last-gen single-GPU.

    But yes its easy to ignore the fact the 580 also overclocks quite well. I don't know what would be worst though for a new 7950 or 7970 owner. Kepler launching and make these prices look like tragic comedy, or AMD releasing the HD89xx not too long after to make them feel even worst about their purchase.

    Either way, you make a compelling argument against buying one of these cards today.
    Reply
  • Master_shake_ - Tuesday, January 31, 2012 - link

    you are obviously a fanboy,

    obviously you missed the part where the 7970 can oc to 1125 on the cvore and max out the vram clock with STOCK voltiage, and when you do so it beats BEATS the gtx 590 a current gen dual gpu card.
    Reply
  • chizow - Tuesday, January 31, 2012 - link

    While I value OC'ability, I don't consider OC to stock comparisons relevant because frankly, it makes no sense.

    Every card can OC to some degree so an apples-to-apples comparison should be the primary comparison with overclockability secondary.
    Reply
  • Tchamber - Wednesday, February 01, 2012 - link

    Rather than call you a fanboy, I would ask for an example of a high end card going against the previous gen x2 card...and bear in mind that ONCE does NOT consitute a trend. I think evolution of performance is always incremental, at least is has been since gtx200 and radeon 4000 series. Reply
  • chizow - Wednesday, February 01, 2012 - link

    -GTX 280 was very close to as fast as 9800GX2
    Filter prevented a link for this, but google computerbase GTX 280 review and look at performance rating summary.

    -GTX 480/580 was faster than GTX 295
    Same deal, google 480 review for that site.

    -GTX 680 should be very close to GTX 590 performance

    What makes this easier on the next-gen GPU flagship is the last-gen X2 parts generally have to make sacrifices in either bandwidth, functional units, or clockspeeds to stay within TDP envelopes along with imperfect multi-GPU scaling, so you generally see only 50-75% scaling over the top single-GPU of that generation.

    So you can look at it two ways, either next-gen should be ~50% faster than the top last-gen GPU, or it is very close to the last-gen X2 GPU card.

    I'm trying not to take any shots at AMD fans, but their reactions to these Tahiti parts in light of historical price and performance leads me to believe they either have very low standards/expectations or they just haven't been paying attention to the industry.
    Reply
  • Galidou - Saturday, February 04, 2012 - link

    ''they just haven't been paying attention to the industry''

    Wait, if I buy a GPU now, and I want the top single gpu for multiple screens, whatever the people in the world think about what I WANT, I realize that I have more power for my money by buying a 7970 and I go with it, it turns out that because I have done that, I'm an AMD fanboy because I haven'T been paying attention to the industry?

    OMG I'm such a tard, I forgot I haven'T paid attention to the industry so I should of paid more for a less performing card..... darn that's how stupid I am.....
    Reply
  • chizow - Saturday, February 04, 2012 - link

    If you absolutely had to have a new card today, sure you'd have some reason to buy a 7970, but still not at these prices imo.

    If you wanted to maximize your return on the dollar however, you'd be much better off waiting for Kepler because even if the 7970 still suits your needs better at that point, Kepler will almost assuredly force downward pressure on all Tahiti AND Fermi parts and force AMD and Nvidia to adjust prices accordingly. In the past AMD has done this when they launched a next-gen part first (See: 5850/5870 launch), but not this time.

    This actually reminds me a lot of the X1950XTX launch. Great performance when it launched and price perfectly justified compared to Nvidia's last-gen G70/G71 designs like 7900GTX and 7950GT. But as soon as G80/8800GTX launched a few months later, it made that buy decision look horrible in retrospect.

    So yes, while a X1950XTX today, er I mean 7970 today looks perfectly justified, it would be a mistake to ignore what the future holds and what the past has told us when making a buying decision.
    Reply
  • Galidou - Sunday, February 05, 2012 - link

    What about the pricing of gtx 280 at launch vs the price of 8800gt were selling? There's always an example about Ati making bad pricing decisions, but nvidia... not out of your mouth?

    I'm no ATI fanboy, I'm just waiting for you to show us you see the other side of the medal...

    Those GTX 280 650$ at launch while the 8800gt was about 55-65% of the performance of that card but HEY it was three times cheaper... No you haven't heard of? Oh I forgot, Nvidia makes no mistakes, it's ATI that did with the 4870 pricing it so low, but Nvidia... they make no mistake.

    But hey the price of the 7970 accordingly to it'S performance is SO BAD compared to the triple price of that GTX 280 for not even double the performance over that nice 8800gt...
    Reply

Log in

Don't have an account? Sign up now