NVIDIA GeForce RTX 3090 - Windows vs. Linux GPU Compute Performance

Following the recent RTX 30 series Linux gaming benchmarks and RTX 30 compute comparison, I was curious how the Linux performance for the flagship GeForce RTX 3090 graphics card compares to the Windows 10 performance in various GPU compute workloads. Well, here are those benchmarks for those wondering about Vulkan / OpenCL / CUDA / OptiX compute performance between Windows and Linux with the very latest NVIDIA drivers.

With the official NVIDIA Windows and Linux (and FreeBSD and Solari) drivers being comprised from largely the same sources, historically the performance has been close to the same across platforms when it comes to the binary driver speed. It mostly though comes down to a function of the application/game under test how well it is optimized for a given platform and if it's relying upon any emulation / additional abstraction layers primarily on Linux. For today's article is a fresh look at the Windows vs. Linux NVIDIA performance when focusing in on the GPU compute performance.

The same AMD Ryzen 9 5900X system was used for all of the testing with the NVIDIA GeForce RTX 3090. On the Windows 10 Pro side was the NVIDIA 466.27 driver as the latest packaged driver as of testing. On the Linux side was Ubuntu 21.04 with the Linux 5.11 kernel and using the NVIDIA 465.27 driver as the latest public Linux driver as of testing time. Both of these NVIDIA driver builds were exposing Vulkan 1.2 / OpenCL 3.0 / CUDA 11.3.