NVIDIA CUDA 8.0 Benchmarks For Ending 2016 + cuDNN Caffe AlexNet/Googlenet
And the CUDA Mini-Nbody results.
Now for arguably the most interesting tests, Caffe.
The AlexNet results from the GTX 680 to GTX 1080... Over five times faster using this high-end Pascal GPU compared to what was great for Kepler back in the day.
And equally impressive results for Googlenet with these CUDA+cuDNN runs on the different GPUs.
Well, those are our latest NVIDIA CUDA Linux benchmarks as we end out 2016. If you wish to compare your own CUDA-enabled Linux system's performance against the numbers found in this article, simply install the Phoronix Test Suite and run phoronix-test-suite benchmark 1612261-TA-CUDA2016N88.
For those interested in just open standards, recently I also did some Blender / Darktable OpenCL benchmarks as well as some AMD vs. NVIDIA Darktable 2.2 OpenCL tests.
If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.