Numba/CUDA compatible?

[blooop]

Hi,

I'm interested in running say 1000 independent IK queries on a gpu. In the readme you mention that this is a pure python library which means it theory it should work with numba relatively easily. Are there any caveats or thing I should be aware of if I'm going to try this?

Something similar to this: https://ipython-books.github.io/58-writing-massively-parallel-code-for-nvidia-graphics-cards-gpus-with-cuda/

Thanks

[Phylliade]

Hey @blooop,

That's a good idea indeed! There is no caveat a priori for the inference part (forward kinematics) For the inverse kinematics, the optimization is done in scipy, out of the python loop, so it won't be done on the GPU.

FYI, there is a plan to use Jax (https://github.com/google/jax) to do the inverse kinematics on GPU

[blooop]

Thanks, jax also looks very interesting.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.