NVIDIA GeForce GTX 670 Review Feat. EVGA: Bringing GK104 Down To $400
by Ryan Smith on May 10, 2012 9:00 AM ESTCompute
Shifting gears, as always our final set of benchmarks is a look at compute performance. As we have seen with GTX 680, GK104 appears to be significantly less balanced between rendering and compute performance than GF110 or GF114 were, and as a result compute performance suffers. Cache and register file pressure in particular seem to give GK104 grief, which means that GK104 can still do well in certain scenarios, but falls well short in others.
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.
It’s quite shocking to see the GTX 670 do so well here. For sure it’s struggling relative to the Radeon HD 7900 series and the GTX 500 series, but compared to the GTX 680 it’s only trailing by 4%. This is a test that should cause the gap between the two cards to open up due to the lack of shader performance, but clearly that this not the case. Perhaps we’ve been underestimating the memory bandwidth needs of this test? If that’s the case, given AMD’s significant memory bandwidth advantage it certainly helps to cement the 7970’s lead.
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.
SmallLuxGPU on the other hand finally shows us that larger gap we’ve been expecting between the GTX 670 and GTX 680. The GTX 680’s larger number of SMXes and higher clockspeed cause the GTX 670 to fall behind by 10%, performing worse than the GTX 570 or even the GTX 470. More so than any other test, this is the test that drives home the point that GK104 isn’t a strong compute GPU while AMD offers nothing short of incredible compute performance.
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.
Once again the GTX 670 has a weak showing here, although not as bad as with SmallLuxGPU. Still, it’s enough to fall behind the GTX 570; but at least it’s enough to beat the 7950. Clockspeeds help as showcased by the EVGA GTX 670SC but nothing really makes up for the missing SMX.
Our foruth 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.
For reasons we’ve yet to determine, this benchmark strongly dislikes GTX 670 in particular. There doesn’t seem to be a performance regression in NVIDIA’s drivers, and there’s not an incredible gap due to TDP, it just struggles on the GTX 670. As a result performance of the GTC 670 only hits 42% of the GTX 680, which is well below what the GTX 670 should theoretically be getting. Barring some kind of esoteric reaction between this program and the unbalanced GPC a driver issue is still the most likely culprit, but it looks to only affect the GTX 670.
Finally, we’re adding one last benchmark to our compute run. NVIDIA and the Folding@Home group have sent over a benchmarkable version of the client with preliminary optimizations for GK104. 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.
Whenever NVIDIA sends over a benchmark you can expect they have good reason to, and this is certainly the case for Folding@Home. GK104 is still a slouch given its resources compared to GF110, but at least it can surpass the GTX 580. At 970 nanoseconds per day the GTX 670 can tie the GTX 580, while the GTX 680 can pull ahead by 6%. Interestingly this benchmark appears to be far more constrained by clockspeed than the number of shaders, as the EVGA GTX 670SC outperforms the GTX 680 thanks to its 1188MHz boost clock, which it manages to stick to the entire time.
Facebook
Twitter
RSS
414 Comments
View All Comments
Galidou - Sunday, May 13, 2012 - link
LoL you're still so much into it's almost like you wanna make me feel that you're gonna save the world with your knowledge. Pollution will be eradicated by the light of your mesurements(in milimeters that is).Sorry if my english ain't at your level, maybe why you beleive I'm young, but it is in fact my third language. The only fanboy here is you, I know that this 670 is AMAZING, if I were in the market for a 300$ video card with what I see now, I'd dish an extra 100 to get it without hesitation.
OMG sonny boy you had to mention this like it was all the supernatural 6mm difference... who gives a darn but you... If you're that old and ''responsible'' commenting like you do about video cards, WOW, it's even worse than I think, the word responsible might even have to change definition just because of you.
I have never seen someone so irresponsible in forums when speaking about video cards. And I'm not the only one who might think that way for sure.
You'Ve been doing personnal attacks on this forum on a regular basis and all that because of what, because of your knowledge in video cards. I'd prefer to be the most stupid man on earth instead of using any form of knowledge the way you do.
CeriseCogburn - Sunday, May 13, 2012 - link
Another gigantic wall of text with nothing on the topic, and nothing but attacking. Goodbye, you missed all the discussions, and all you've got to add now is your hatred.Galidou - Monday, May 14, 2012 - link
I never said your discussions were fulled with lies or anything, it's just the way you bring up your argumentation. It makes me feel like not everyone will listen to you because of that. You do it in a harsh way and bring everyone down with you at the same time.You think calling everyone an ignorant is really necessary to explain anything related to the topic even if their argumentation is flawed? I don't think so. I already know Nvidia won for this gen, and they won big time. It's nice to see some nice bang for your bucks at the top, and I mean, if it wasn't for the fact that I only run in 1080p and the most anticipated game I'll play is diablo 3, I'd get one right now. But darn it dude, calm down...
CeriseCogburn - Thursday, May 31, 2012 - link
You calm down amd fanboy, you LOST, amd LOST, and is LOSING, and amd is near broke.And you're very upset over it.
Gastec - Tuesday, November 13, 2012 - link
Aha! Busted! You are looking forward for AMD to get broke, closed so we would all be forced to buy video cards from only one corporation: nVidia. Because that way your share of money will get bigger. You greedy bastard!jamyryals - Friday, May 11, 2012 - link
While highly entertaining to me, your comments are actually a bit disturbing when one thinks about what you are like in real life.CeriseCogburn - Friday, May 11, 2012 - link
Yes, don't forget to personally attack me, use the usual blabber the most ignorant and clueless fools on the net use to do so, it's a very short list, so it won't be too hard for you unless you try to remember how to spell basement.Now back to the topic I brought up.
It's a sad time for amd fanboys and no amount of lies can help.
After the reviewer smacked down all 79xx CF setups as not able to recommend, we have this very next follow up review - and it's easy to say everyone is absolutely amazed by the massive performance of this next step nVidia GTX670 - beating amd and their fanboys at the very heart and I do mean their dark little love it to death fanboy talking point amd heart - die size / power use / price perf / fan noise...
It's a total and complete smackdown, exceeding even in muliti monitor with the nVidia 3+1surf - a complete smackdown - no area left for the and fanboy to grab onto - extreme sadness the era has ended.
Consolation prize is vehemently claiming amd "OC's better", but it's a very difficult and voltage increasing road of heat and instability to that little nugget - while the nVidia fan comfortably uses a new technology for OC, and "enjoys some OC anyway" even without touching a thing, if we are to believe the angry and defeated amd fans protestations..
SlyNine - Saturday, May 12, 2012 - link
You need to take a step back and read your comments. You seem to think you're the only person with a valid opinion.CeriseCogburn - Sunday, May 13, 2012 - link
Oh more personal attacking ?You can have your opinion, even if it's ridiculous and stupid, and rest assured, I may use the facts to prove that it is, and thus, you may whine I seem to think only my opinion is valid.
I see now you went on a half page excuse rant for amd in the prior page trying to justify it's terrible loss with much spinning and distortions - look buddy, why don't you take some of your own advice ?
It's obviously very important to you, given your actions - so it would be better if you made more attempts like the one I just referred to, and also had the guts to correct those spinning and lying for amd, as then I wouldn't be so busy.
We can thank snakefist for pointing out past page 25 here that the nVidia core size is 294mm on a side not 300mm - so there we go again, another lie by an amd fan corrected...
( to be fair there was some misinformation concerning that comporting to the error the poster made)
Thanks for telling me to read my own posts, I assure you I do, although now I've skipped reading your entire rant on the prior page. It's laughable BTW.
haakon_k - Saturday, May 12, 2012 - link
While a bit entertaining to me, your comments are actually highly disturbing when one thinks about what you are like in real life.*my version