Back to the Shapespark home page

GeForce RTX 2080 bake question

I have a GeForce RTX 2080 with Max-Q Design for my modified MSI gaming laptop.

ShapeSpark displays it as recognized, but is only using 0-0.2% of the GPU capacity. This is still 33% faster than running on my CPU: Intel i7-9750 @2.60 Ghz and 32 GB ram.

I have tried loading the older nvidia toolkit version 10.1 as recommended on your forum for this topic, no success. So I reverted back to the latest nvidia drivers… Same results occur…

How do I get shapespark to use all of my GPU’s capabilities?

Hi @Digiwip,

RTX 2080 Max-Q is based on the same architecture and the same chip as RTX 2080 (desktop & mobile), so it should work out of the box.

Only the baking phase - when the 'Baking lightmap X of Y" progress bar is displayed - is computed on the GPU, the other phases like denoising or calculating lightmap UVs are always computed on the CPU. If you checked the GPU load during the other phases, it might have been as low as you write - 0-0.2%.

In general, for small scenes or scenes using a low number of samples, using a powerful GPU may not significantly outperform a CPU, because the baking phase is short in comparison to the other phases. However, for large scenes or high number of samples the difference should be much more noticeable.

What do you mean by 33% faster? Do you refer to the total time - from pressing the “Bake” button to the end?