AMDGPU-PRO vs. NVIDIA OpenCL Performance With ArrayFire Using 18 GPUs

Written by Michael Larabel in Graphics Cards on 25 January 2017 at 09:00 PM EST. Page 1 of 4. 17 Comments.

With now being able to benchmark ArrayFire via the Phoronix Test Suite, I've been having fun running a number of OpenCL graphics card tests with the 300+ available AF tests. The tests over the past week have been using the NVIDIA Linux driver while here are our first Radeon benchmark results using the AMDGPU-PRO driver stack.

As our latest OpenCL benchmarking look is seeing how this ArrayFire compute library compares between NVIDIA and Radeon under Linux. The NVIDIA tests were obviously done with the latest official driver (375.26) given the lack of other viable OpenCL driver options. On the Radeon side, the AMDGPU-PRO 16.50 driver for OpenCL testing was the only logical choice given the open-source Clover-based Gallium3D stack for it is basically abandoned and incomplete while the newer ROCm OpenCL stack still leaves a lot to be desired in its current stage. (For additional reference was the recent NVIDIA Linux OpenCL Performance vs. Radeon ROCm / AMDGPU-PRO.)

ArrayFire NVIDIA GeForce AMD Radeon OpenCL Linux Testing

The NVIDIA cards tested for this comparison were the GeForce GTX 750 Ti, GTX 760, GTX 780 Ti, GTX 950, GTX 960, GTX 970, GTX 980, GTX 980 Ti, GTX 1050, GTX 1050 Ti, GTX 1060, GTX 1070, and GTX 1080. On the Radeon side the selection was more limited due to AMDGPU-PRO only supporting newer cards and having less cards available to me, but tested were the most interesting options of: Radeon R9 285, R9 290, RX 460, RX 480, and the R9 Fury.

All of these 300+ ArrayFire benchmarks were done in a fully-automated and reproducible manner using the open-source Phoronix Test Suite benchmarking software.


Related Articles