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
Vertigo2000 - Monday, May 14, 2012 - link
That's some bad, bad math comprehension. 300 and 365 are already the calculated areas.300 mm sq. = 17.3205 mm x 17.3205 mm
365 mm sq. = 19.1050 mm x 19.1050 mm
365 mm sq. is 21.667% larger than 300.
chimaxi83 - Thursday, May 10, 2012 - link
Pay no attention to this known banned from the forums troll. Poor, sad little man you are lol, sucks to be you doesn't it?CeriseCogburn - Thursday, May 10, 2012 - link
I agree, the targeted "forum trusted end user alpha males" amd's pr campaign handlers coddle and coach and follow and email and bribe with free amd hardware and event tickets are gonig to be sorely pressed making up enough fanboy lies to earn their freebies.They are sad, and muted, bringing down the entire amd fanboy morale - the lies need to be too big this time - and evil amd already scapled the crap outa their fanbase, lost in their power perf area, lost in frame rate, lost their 3G future fantasy ram argument - all they have left is blowing chunks about overclocking the unstable poor drivers 7000 series - not to mention the GTX570 slamming their 78xx series, and a single GTX680 outselling their entire 78xx series at the egg...
I've seen more than one reviewer mention the 680 back orders were so great and the 79xx and 78xx sales so stalled, that they hoped for some kind of change... now the egg is full of 670's...
amd fanboys need a whole wheel barrel full verdetrol just to not off themselves
Spunjji - Friday, May 11, 2012 - link
What kind of weird conspiracy crack are you on? :SCeriseCogburn - Friday, May 11, 2012 - link
I guess you're so ignorant you didn't read the fired AMD employees testimonies concerning and explaining their "use" of amd fanboys, preferably the "trusted alpha male" type from long established forums all over the web, including anand's here.If you're entirely clueless, as you are, you have a lot of homework to do to be able to understand.
Now, with your rude attack due to your personal ignorance above, I think all of us can fairly completely ignore your sad, pathetic, amd pro backing smack talk comment above that, where put down the entire comments section as an argumentative and sad place.
Kind of ironic that after trying to defend another amd fanboy who smarted back about not being sad, you wound up saying this place is sad, and yes we've all seen your amd favoritism for some time now.
You are sad, amd has lost, and has been forced down into shame by their superior, nVidia.
Galidou - Saturday, May 12, 2012 - link
LoL Cerise is still here launching inflammatory stuff at everyone who says something about AMD. Look at the mad Nvidia fanboy crying setting fire to the rain...Whoever loses, you don't have to get so MUCH mad when speaking of computer parts, they are freaking computer parts, not women intimate parts... get laid before speaking of ignorance...
Why is it always the nvidia fanboys who gets so mad, are they freaking lacking real life sex so they become obsessed by video cards. I don't even want to think at what kind of picture they look at when they try to evacuate the pressure.... NO I DON'T...
CeriseCogburn - Thursday, May 31, 2012 - link
Amd lost, and lost badly, and all you do is attack me, so you lose too.Gastec - Tuesday, November 13, 2012 - link
Omg he's just a troll and you have fallen into his trap. Went me talking to the walls painted 6 months ago....Ah but maybe the troll reads this and replies. And if he's a mad troll he will copy my comments store it in a .txt. or .doc and use it "against" me in future ..."fights" Pfff! Sayonara suckers!Galidou - Sunday, May 13, 2012 - link
''Now, with your rude attack due to your personal ignorance''LOL, I wonder who's rude?
''your sad, pathetic, amd pro backing smack talk comment above that, where put down the entire comments section as an argumentative and sad place''
I think that just answered my last question. Just sad to see someone who gather this information to use it like you do, good job.
''You are sad, amd has lost, and has been forced down into shame by their superior, nVidia. ''
I think the only one who lost, it's you. You lost your sanity just for speaking about video card companies, according so much importance to it while it's so much irrelevant. 90% of the people will continue going to best buy and get a geforce 8600gt or a Radeon 2600xt for 120$ thinking they got the deal of the year because it has a 30$ rebate on it and allow them to run their game at 720p. GG
CeriseCogburn - Sunday, May 13, 2012 - link
Ah yes, another whose only "information" in 2 consecutive posts is 100% personal attack.Thank me when you grow up enough to realize rebutting lies and fibs by others is an adult and responsible thing to do. Something you obviously are absolutely incapable of, since expertise is not in your arsenal.
BTW - the gentleman who had to be corrected on his lackluster amd excuse concerning gpu core size vs amd's ability to for the 3td time in a few months, reduce prices, it has been pointed out, lied about the nVidia core size exaggerating it's dimensions (by mistake of course!).
ROFL hahahha - you pathetic amd fanboys have to lie all the time....
Not 300mm, 294mm sonny boy. ROFL
When it's so bad no lie is too big or too small, this is what happens - this is what we're (well not you of course, you're just an attacking troll) dealing with.