NVIDIA GeForce RTX 4090/4080 Linux Compute CUDA & OpenCL Benchmarks, Blender Performance
When taking the geometric mean of all the compute/render tests carried out for today's article, the GeForce RTX 4090 was 1.78x the speed of the prior generation GeForce RTX 3090 or 2.66x the speed of the two generations old TITAN RTX. The RTX 4080 was 1.43x the speed of the prior generation RTX 3080 and 2.52x the speed of the RTX 2080 SUPER.
For as impressive as the raw performance was, it was also surprising to see the RTX 4080 and RTX 4090 consume less power often than the prior generation parts. The RTX 4090 had the highest peak power consumption by far but for the average power consumption rate across the mix of GPU compute and render benchmarks was in very good shape.
The GPU thermals for the RTX 4080 and RTX 4090 with these NVIDIA Founder's Edition cards were also running at the lowest operating temperatures of all the tested NVIDIA graphics cards. These graphics cards are large but their sizable cooling package allows these graphics cards to operate rater efficiently.
Overall the NVIDIA GeForce RTX 4080/4090 performance in this range of Linux tests boasted terrific generational improvements. The NVIDIA GeForce RTX 4080/4090 Linux gaming performance was great while these render and OpenCL/CUDA compute benchmarks showed these high-end Ada Lovelace GPUs to be just as -- or even more -- compelling for Linux creators. If you don't mind NVIDIA's current proprietary driver stack, the GeForce RTX 4080 and RTX 4090 offer great Linux compute potential.
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.