NVIDIA GeForce RTX 4090/4080 Linux Compute CUDA & OpenCL Benchmarks, Blender Performance

Right off the bat when running Blender 3.4 with the CUDA back-end the GeForce RTX 4080 and RTX 4090 show off their lofty performance abilities over prior generation RTX 30 and RTX 20 GPUs. The RTX 4090 took a third of the render time of the TITAN RTX for handling the demanding Classroom scene.

But, of course, the CUDA back-end for NVIDIA isn't even the most ideal... When firing up Blender 3.4 with OptiX to make use of the RTX capabilities, the render times drop further. The RTX 4090 with OptiX took around half the time to render the Classroom scene as the prior generation GeForce RTX 3090 graphics card.

The GeForce RTX 4080 and 4090 graphics cards were not only performing much better than the RTX 3080/3090 but for this benchmark even managed to do so while consuming less power than the Ampere GPUs.

The RTX 4080 and RTX 4090 graphics cards are quite large in their physical size, but their large heatsinks do a great job at keeping the Ada Lovelace GPUs operating cool.

The RTX 4080/4090 graphics cards were offering terrific generational improvements for Blender.

Especially if you are on an RTX 20 series card or older, it may be worthwhile considering the RTX 40 series upgrade.