Compute Benchmarks

The next batch of benchmarks won’t apply to everyone, but I like to include them just in case anyone needs them. These benchmarks are all compute benchmarks. I start with Folding at Home to take a look at both single and double precision performance. This applies to gaming some because async is starting to become important with DX12, the games are only starting to trickle in but a heavily optimized game can really benefit. Take a look at Doom results with Vulkan with the new RX 480 for example. So anyhow in Folding at Home, the EVGA GTX 1070 Superclocked performed really well in the single precision benchmark. In fact, it pulled ahead of the other GTX 1070’s by a surprising margin, even with the MSI being overclocked. In double precision, it dropped as expected. Currently, Nvidia has been focusing on clock speed and memory improvements rather than double precision with their consumer cards. This helps with efficiency and also helps protect their higher margin Quadro lineup as well. So here we are actually below the old GTX 780 and the much cheaper RX 480.

The other half of my compute testing was in CompuBenchCL. Here I focused on the Video Composition and Bitcoin results. These are both situations that the average user might get into. When rendering video the EVGA performed really well, putting up just behind the GTX 1080 and above everything else. In the Bitcoin mining benchmark, the result was similar but with the MSI edging ahead slightly.