Compute Performance

In our recently added Compute benchmarks I use the Folding @ Home benchmark in both single and double precision tests to see how well the cards perform. In the single precision test the GTX 980 Ti came in just three nanoseconds per day less than the Titan X that tops that chart. In the double precision test the numbers were again right behind the Titan X but this time around the older GTX 780 still tops the chart by a large margin. This is because Nvidia has pulled back on the double precision performance of their new cards to prevent people from buying cards like the Titan X for MUCH cheaper than their Quadro and getting close to the same performance. With their Quadro cards requiring their own development and software work people snatching up the consumer cards was hurting their bottom line. It’s not a huge deal considering nearly all double precision applications should be looking into commercial quality cards anyhow.