AMD’s Radeon HD 6670 & Radeon HD 6570: Two’s Company, Sub-$100’s A Crowd
by Ryan Smith on April 19, 2011 12:01 AM ESTCompute & Tessellation 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 benchmark.
While Turks and the 6670/6570 are quite a bit faster than the 6450, both are still so-so as compute cards. Although the unit of measurement (fps) is effectively arbitrary, what isn’t arbitrary is that the 5750 is nearly 33% ahead of the 6670. Juniper (5750/5770) seems to be the floor for good compute performance on AMD cards.
Our second compute benchmark of the day is SmallLuxGPU, the GPU ray tracing branch of the open source LuxRender renderer. While it’s still in beta, SmallLuxGPU recently hit a milestone by implementing a complete ray tracing engine in OpenCL, allowing them to fully offload the process to the GPU. It’s this ray tracing engine we’re testing. AMD in turn recently hit a milestone by offering the OpenCL runtime with their drivers, so SmallLuxGPU will now run out of the box without the AMD Compute SDK.
At 3200K rays/sec, the 6670 actually beats the Radeon HD 4870 by about 10%, showcasing just how far AMD’s compute performance has come compared to the 4000 series. The 6670 also enjoys a similar lead over the 5670 thanks to its additional shaders, but the 720SP 5750 looms large above it.
At the other end of the spectrum from GPU computing performance is GPU tessellation performance, used exclusively for graphical purposes. Barts’ tessellation improvements are inherited by Turks, which should give it an edge over the 5670.
What ended up surprising us the most with our DX11 Detail Tessellation Sample test wasn’t so much the raw performance of the 6670, it was the performance relative to some other AMD cards. At a high tessellation factor the Radeon HD 5870 isn’t all that far ahead, and meanwhile the 5750 is edged out by the 6670 in all cases. Given that our most tessellation-intensive games struggled at 1680x1050 on the 6670 it may not be powerful enough to really use its tessellation abilities in full on top of a heavy graphical workload, but the basis for respectable tessellation performance is there.
Facebook
Twitter
RSS
53 Comments
View All Comments
Ryan Smith - Tuesday, April 19, 2011 - link
Our primary focus on reviews is for the North American market. I'm not sure about Europe and Asia, but in North America the 6450 does (did?) not go on sale in retail until today. It has been available to OEMs for a couple of months however.mino - Tuesday, April 19, 2011 - link
Even ignoring your global audience, if a card was available for MONTHS in the OEM channel, calling it "paper launch" is absurd.But; who pays, he gets. :(
AstroGuardian - Thursday, April 21, 2011 - link
Consider opening an European branch. Many of us do not agree with many things in the reviews.Targon - Tuesday, April 19, 2011 - link
If a game is CPU limited after that many years, that hints that the game is not multi-threaded by design. Both Intel and AMD have really been more focused on multi-core designs, rather than really pushing the performance of individual cores. Yes, there have been improvements, but it has not been the real focus of CPU development. AMD looks to be working on getting the core design improved to be more competitive with Intel, but that is pretty much it at this point. Going to a 32nm CPU design should also help.vavutsikarios - Tuesday, April 19, 2011 - link
I guess it's not really multithreaded, and even if it is, it definitely won't use more than 2 cores. Afterall, it's a 2005 game. But this is besides the point. The game is CPU limited because of its nature. It is not a design flaw or anything. To clarify: the CPU is the limiting factor not from a performance POV but from a gaming experience POV. After you do whatever you have to do you click the "end turn" button. Then you have to wait for the AI to make their move. This takes awhile. It was minutes, worst case, on the PC I had when I first played the game, it may be less than a minute on the 3GHz quadcore I use now. Still, in order to have smooth gameplay, I need this to become 100x faster.SlyNine1 - Wednesday, April 20, 2011 - link
Thats like saying ChessMaster is CPU limited. Or saying Every game out there is HDD/SDD limited. Yes you will have to wait for complexe operations, but it doesn't interfer with gameplay. Because of that its hard pressed to say its CPU limited, as It doesn't limit gameplay.vavutsikarios - Wednesday, April 20, 2011 - link
I understand what you re saying. The CPU speed, at these levels of CPU performance, doesnt interfere with the functions of the game, that is true. But it does affect the gaming experience. Imagine having to wait a couple of hours for the AI. Wouldnt that be really frustrating? More than that, wouldnt it render the game unplayable? The way you define gameplay, having to wait any amount of time doesnt matter. So, I guess, it is a matter of semantics, of definition of what gameplay is.In the broader sense, which, IMHO, is what matters, things like that are important. They directly affect the pleasures we make for ourselves in the precious little time we have. So, yes, obviously, Every Game out there is HDD Limited! -nice line that one :)
Btw: Chessmaster is not CPU limited, although it should be. It is not though, because having to wait for your opponent to move is part of the normal chess experience. Truth is, chessmaster moves way too fast sometimes, and sometimes it thinks a lot when what it has to do is obvious, but still.
AstroGuardian - Thursday, April 21, 2011 - link
+1fic2 - Tuesday, April 19, 2011 - link
CPU limited on an i7-2600k Sandy Bridge? Or CPU limited on the cpu you bought 6 years ago to play Heroes 5?vavutsikarios - Tuesday, April 19, 2011 - link
See my reply to Targon above. An i7-2600 Sandy would probably be a nice improvement over my phenom2, but still a long way from not being the limiting factor.