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
Comments Locked

259 Comments

View All Comments

  • Prosthetic Head - Friday, February 3, 2012 - link

    I purchased a second hand 4850 and put a nice custom cooler on it about 3.5 years ago. It runs near silent and performs more than adequately in all the games I play. I honestly don't see the excitement in GPU announcements unless they really are game changers (59XX, 9800XT --> 6800, X800 type leaps). I realise I'm not a bleeding edge gamer & some of you are and thats great - It means second hand high end GPUs available regularly for the same price as low end offerings less than a year later.

    The only thing that tempts me about the new architectures is the compute performance. I actually use this for some computational chemistry work and the upgrade looks well worth it if I do more of this type of work.

    The improving performance / power ratio is also of interest since I don't like loud fans, hot computers or excessive electricity consumption.
  • Galidou - Saturday, February 4, 2012 - link

    Don't worry, chizow is a fanboy, there's nothing to do against greenies, they have a closed mind :P

    He never said anything wrong about nvidia but sure has a ton of things to say about AMD. He speaks of overclockability of GTX 580, get an answer about how overclockable the 7970 and then changes his mind and says he cares about OC but not much about stock cards OC(which the 7970 seems to do VERY well).

    Makes no sense when you say things and then right after what you say has no value for the other team.....
  • chizow - Saturday, February 4, 2012 - link

    No, I said it makes no sense to compare one stock part to one that's overclocked. If you want to compare apples to apples, you overclock both, I have no problem with that and I think you'll find that both parts overclock similarly so the original difference in stock performance holds true.

    Only a fanboy would try to compare stock to OC results as you and others apparently tried to do.
  • Galidou - Sunday, February 5, 2012 - link

    I never tried to compare any video card, I'm a proud owner of both camps video card, from geforce 2 gts all the way up to 6850 in crossfire and gtx 560 ti in other rig... thing is you never said anything wrong about Nvidia and I can say wrong things and good things about both camps being an owner of MULTIPLE of their cards...

    Someone with only one side of a story shouldn't be used as a reference for righteousness in judgement... Like taking a judge for an affair of murder while the judge himself is in the family of the murdered one.....

    Anyway just my two cents...
  • chizow - Sunday, February 5, 2012 - link

    If you never tried to compare any video card, why are you defending someone who did then?

    As usual with these kinds of threads arguing with ignorant fanboys, I'm simply keeping things honest and correcting erroneous flaws in facts or logic.

    Does it make sense to you to make a comparison about one overclocked part compared to another stock part, then declare the OC'd part the winner based on those results?

    Of course not!

    As for your two cents, keep them, you will need to save every penny to make the jump from 6850 or GTX 560ti to 7950s at these prices.
  • Galidou - Sunday, February 5, 2012 - link

    There was bad things about AMD and Nvidia in history and thing is you can only mention bad things from one side. I'd never listen to someone that already has a choosen side to speak about ANYTHING in my whole life, simply because you won't be able to hear the absolute truth from the defender of only one side...
  • chizow - Saturday, February 4, 2012 - link

    No, but you were talking about me. So, please explain what you think determines pricing in this market since you insist I have no clue, or YOU shut up.

    Thanks!
  • chizow - Saturday, February 4, 2012 - link

    Yes of course, because it would be negligent not to from any perspective.

    Anyone who buys something at any point needs to project the market, needs to weigh the potential for obsolescence. That's a large reason why companies do NOT disclose unannounced products as not to erode their own sales before a product goes EOL.

    Similarly, it would be negligent if AMD didn't project Nvidia's next-gen parts given they use the same fab and process. And right now, AMD is basically telling you with this price that they either 1) are ignoring Nvidia's 28nm partsor 2) think Nvidia's 28nm parts perform the same as their 40nm parts or 3) don't think their customers are smart enough to realize 1 or 2.

    Either way, they're going to look the fool once Nvidia does release their 40nm parts.
  • chizow - Saturday, February 4, 2012 - link

    You mean the super-OC'd Classified 3GB versions that perform within a few % points of the 7970? Yeah of course there are. You'd be a moron not to realize the difference, or you're just being dishonest. With you it could be either.

    Most 580s are in the same price range as the 7950, $440-500, because once again, they deserved that price tag when they launched 14 months ago and the release of the 7950 has done nothing to make Nvidia drop the price on them.

    But as I said, this price and performance level will be retired soon, but obviously not as a result of the "next-gen" Tahiti parts. We'll most likely have to wait for the real "next-gen" parts with Kepler to shift this stagnant price:performance metric.
  • chizow - Saturday, February 4, 2012 - link

    No, only a moron would blow $500 on another card a year later that performs the *SAME* as the card they paid $500 for a year earlier. And that's why the 7950's pricing sucks.

    There's light at the end of the tunnel, you may finally be getting it lol.

Log in

Don't have an account? Sign up now