Compute Performance

As always our final set of real-world benchmarks is composed of a look at compute performance. As we have seen with GTX 680 other Kepler cards, Kepler appears to be significantly less balanced between rendering and compute performance than GF110 or GF114/GF116 were, and as a result compute performance suffers.  On the other hand, relative to the GTX 660 the GTX 650 Ti sacrifices a smaller portion of its compute performance than its ROP/L2/memory performance, so this may bode better for computer performance.

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.

There really isn’t a lot to say here. The GTX 650 Ti is only slightly ahead of the GTX 550 Ti, never mind the GTX 560. Worse, it’s tied with the 7770 and well behind the 7850. Given the nature of the test, with cards on memory busses this small I believe we’ve run into a proxy test for memory bandwidth rather than compute throughput. Which just goes to show that all of that compute throughput is meaningless without the memory bandwidth and cache to feed the best.

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.

Not surprisingly, the GTX 650 Ti loses to just about everything. SmallLuxGPU’s OpenCL renderer just doesn’t mesh well with Kepler and NVIDIA’s drivers. The resulting lead for the 7850 is nothing short of massive.

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.

Unlike our previous OpenCL benchmark the GTX 650 Ti’s showing isn’t nearly as bad, but neither is it great. Both the GTX 560 and the 7770 are in the lead, but at least the improvement over the GTX 550 Ti is nothing short of amazing. At times NVIDIA’s problem isn’t where GTX 650 Ti is compared to last-generation cards, but rather it is compared to AMD’s strong Radeon HD 7000 series lineup.

Our fourth 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.

Once more the GTX 650 Ti is in trouble. It can beat the GeForce 500 series, but even the 7770 is faster.

Finally, we’ll take a look at one last benchmark to our compute run with the benchmarkable version of the Folding@Home client. Folding@Home and similar initiatives are still one of the most popular consumer compute workloads, so it’s something NVIDIA wants their GPUs to do well at.

Here’s another case where memory bandwidth and L2 cache appear to be a problem. The GTX 650 Ti is much farther behind the GTX 660 than we would have expected, and even the GTX 560 can take a lead here. On the other hand memory bandwidth bottlenecking isn’t so bad that EVGA’s GTX 650 Ti can’t still take the lead over the other factory overclocked cards.

Civilization V Synthetics
Comments Locked

91 Comments

View All Comments

  • Hades16x - Tuesday, October 9, 2012 - link

    A little bit saucy while reading this review on the page "Meet the Gigabyte Gefore GTX 650 TI OC 2GB Windforce" the second to last paragraph reads:

    "Rounding out the package is the usual collection of power adapters and a quick start guide. While it’s not included in the box or listed on the box, the Gigabyte GeForce GTX 660 Ti OC...."

    Shouldn't that read "the Gigabyte GeForce GTX 650 Ti OC" ?

    Thanks for the review Ryan!
  • Hrel - Tuesday, October 9, 2012 - link

    First, card makers: If the card doesn't have FULL size HDMI, I won't even consider it. I get mini on smartphones, makes no damn sense on a GPU that goes into a 20lb desktop. Fuck everyone who does that. Second, Every display I own uses HDMI, most of them ONLY use HDMI. I want to see cards with 3 or 4 HDMI ports on them so I can run 3/4 displays without having to chain together a bunch of fucking adapters. HDMI or GTFO. I really don't understand why any other video cable even exists anymore, DVI is dumb and old, VGA, psh. Display Port? Never even seen it on a monitor/TV. I don't spend stupid amounts of money on stupid resolution displays where NO media is even produced at that resolution; but last I checked HDMI supports 8K video.

    Next: I bought my GTX460 for 130, or 135 bucks. This was a few months after it was released and with a rebate and weekend sale on newegg. Still, that card can MAX out every game I play at 1080p with no issues. I get that they're putting more RAM in the cards now, but that can't really justify more than a 10$ difference; of actual cost. I don't see the GTX660 EVER dropping down to 150 bucks or lower, WTF? Why is the GPU industry getting DRAMATICALLY more expensive and no one seems to be saying a thing? Remember the system RAM price fixing thing? Yeah, that sucked didn't it. I'd really hate to see that happen to GPU's.

    It's good to finally see a tangible improvement in performance in GPU's. From GT8800 to GTX560 improvements were very incremental; seems like an actual gain has been achieved beyond just generational improvements. Hoping consoles have at least 2GB of GDDR5 and at least 4GB of DDR3 system RAM for next gen. Seems like RAM is becoming much more important, based on Skyrim. With that said, I can buy 8GB of system ram for like 30 or 40 bucks. Puts actual cost at a few dollars. No reason at all these cards/consoles can't have shit tons of RAM all over the damn place. RAM is cheap, doesn't cost anything anymore. You can charge 10 bucks/4GB and still turn a stupid profit. Do the right thing Microsoft/Nvidia and everyone else; put shit tons of RAM in AT COST. Make money on the GPU/Console/Games.
  • maximumGPU - Wednesday, October 10, 2012 - link

    we should all be pushing and asking for royalty-free display ports!
    and just so you'd know quite a few high end monitors don't have hdmi, the dell ultrasharp U2312hm comes to mind.
    DP should be the standard.
  • Hrel - Thursday, October 11, 2012 - link

    DP doesn't support audio, as far as I know. Also offers no advantage at all for video. So why?
  • maximumGPU - Thursday, October 11, 2012 - link

    It does support audio!
    with all else being equal the fact that it's royalty free means it's preferable to hdmi.
  • TheJian - Wednesday, October 10, 2012 - link

    I'm not sure of physx in AC3, but yeah odd they put this in there. I would have figured a much cheaper game. When you factor in phsyx in games like Borderlands2 it changes the game quite a lot. You can interact with object in a way you can't on AMD:
    "One of the cool things about PhysX is that you can interact with these objects. In this screenshot we are firing a shot at the flag. The bullets go through the flag, causing it to blow a hole in the middle of it. After the actual flag tears apart, the entire string of flags fell down. This happens with flags and other cloth objects that are hanging around, the "Porta-John's" that are scattered across the world, blood and explosive objects. You can not destroy any of these objects without PhysX enabled on at least Medium. "
    http://hardocp.com/article/2012/10/01/borderlands_...

    I don't know why more sites don't talk about the physx stuff. I also like hardocp ALWAYS showing minimums as that is more important than anything else IMHO. I need to know a game is playable or not, not that it can hit 100fps here and there. Their graphs always show how LONG they stay low also. Much more useful info than a max fps shot in time (or even avg to me, I want min numbers). Anandtech only puts mins in where it makes an AMD look good it seems. Not sure other than that why they wouldn't include them in EVERY game with a graph like hardocp showing how long their there. If you read hardocp it's because they dip a lot, but maybe I'm just a cynic. At least they brought back SC2 :) Cuda is even starting to be used in games like just cause 2 (for water).
    http://www.geforce.com/games-applications/pc-games...
    Interesting :)
  • jtenorj - Wednesday, October 10, 2012 - link

    You can run medium physx on a radeon without much loss of performance.
  • Magnus101 - Wednesday, October 10, 2012 - link

    Why suddenly the race for 60 FPS?
    It used to be 30 FPS average and minimums not going under 18 in Crysis that was considered good.
    Movies are at 24 FPS and stuttering isn't recognisable until you hit 16-17 FPS.
    Pal TV in Europe was at 25 FPS.

    It looks like everybody is buying into Carmacks 60 FPS mantra, which is insane.
    For me minimums above 20 FPS is enough for a game to be perfectly playable.
    This is the snobby debate with audiphiles all over again where they swear they can tell the difference between 96 and 44.1 khz, just substitue the samplerate with FPS.

    But I guess the Nvidia and ATI are happy that you for no reason just raise the bar of acceptance!
  • ionis - Wednesday, October 10, 2012 - link

    60 FPS has been the target for the past 3-4 years. I'm happy with 25-30 but this min 30 ave 60 FPS target has been going on for quite a while now.
  • CeriseCogburn - Friday, October 12, 2012 - link

    You'll be happy until you play the same games on a cranked SB system with a high end capable videocard and an SSD (on a good low ping connection if multi).

    Until then you have no idea what you are missing. You're telling yourself there isn't more, but there is a lot, lot more.
    Quality
    Fluidity
    Immersion
    Perception in game
    Precision
    Timing
    CONTROL of your game.

    Yes it is snobby to anyone lower than whatever the snob build is - well, sort of, because the price to get there is not much at all really.

    You may not need it, you may "be fine" with what you have, but there is exactly zero chance there is isn't a huge, huge difference.

Log in

Don't have an account? Sign up now