AMDGPU-PRO 17.20 Benchmarking vs. RadeonSI/RADV

Thanks to this week's Radeon Vega Frontier Edition launch, AMD pushed out a new build of their hybrid driver stack for Linux, AMDGPU-PRO. This new release is marketed as AMDGPU-PRO 17.20 and is only found when looking for the Frontier driver, but it's been working out fine so far in my Polaris/Fiji GPU testing. Here are some benchmarks compared to their current stable series, AMDGPU-PRO 17.10, as well as the newest open-source AMDGPU+RadeonSI/RADV driver stack.

Aside from the AMDGPU-PRO 17.20 driver having initial Vega support, it also now incorporates the ROCm OpenCL/compute component with this driver. But aside from that, no other official changes were mentioned. But given the ROCm changes, I ran some fresh benchmarks not only with OpenCL but also OpenGL and Vulkan.

On a Radeon RX 470 Polaris system I ran the AMDGPU-PRO 17.10 driver as is the current latest stable release, the AMDGPU-PRO 17.20 release geared for Vega Frontier, and then the latest open-source driver stack. The open-source stack was Mesa 17.2-dev built against LLVM 5.0 SVN via the Padoka PPA and also the Linux 4.12 Git kernel. Given the open-source ROCm not yet running on the upstream Linux kernel and the Clover Gallium3D OpenCL support being unmaintained for RadeonSI, the OpenCL benchmarks were done on just the PRO driver stacks.

All the OpenCL/OpenCL/Vulkan benchmarks were carried out in a fully-automated and reproducible manner using the open-source Phoronix Test Suite benchmarking software.