NVIDIA's GeForce GTX 560: The Top To Bottom Factory Overclockby Ryan Smith on May 17, 2011 9:00 AM EST
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 GTX 560 is at a bit of a disadvantage here. Whereas the higher clocked ROPs would offset some of the GTX 560 Ti’s gaming lead, here the GTX 560 has nothing to stand on besides raw shader power.
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.
This benchmark threw us for a bit of a loop, and we ended up running it several more times just to be sure. Under most circumstances the texture decompression test should be shader bound, but here it’s not – or at least not entirely. The GTX 560 Mid and GTX 560 Ti basically tie here, which is quite odd since the GTX 560 Ti has quite a shader advantage due to the 8th SM. As best as we can tell, either texture decompression doesn’t scale well with SMs, or it’s memory bandwidth bound at some point, with the latter case being the more likely of the two.
In any case as a result the GTX 560 Mid does quite well here, and the ASUS GTX 560 with its factory overclock shoots right past everything short of the 480/570/580. We’ll have to see if this weird outcome holds up in our other compute benchmark.
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.
Under SmallLuxGPU the GTX 560 behaves much more like we’d expect it to. The GTX 560 Mid does well, but it still falls behind the GTX 560 Ti as we’d expect. Keeping in mind that the theoretical difference should be around 10% in favor of the GTX 560 Ti however, the GTX 560 Mid still does a bit better than we were expecting, letting the GTX 560 take off with only a 6% lead. Compared to the GTX 460 however the difference is almost exactly what we’d expect it to be: the GTX 560 Mid is ahead by 23%.
Overall our compute results leave us a bit surprised: while extra shaders are normally extremely beneficial on compute benchmarks, that’s not entirely the case here. The GTX 560 series was never meant to be a compute powerhouse like the GTX 570/580, but if compute performance is a factor in your purchasing decision then it looks like you aren’t giving up much moving from a GTX 560 Ti to a GTX 560 Mid.