clemisch

Wow, thank you for the very detailed answer @mattjj ! After reading your explanation, it makes sense that multiple kernels are built and are controlled from the CPU, creating some...

I've noticed the K4000 GPU is pretty old and only support CUDA compute capability 3.0. Could this be a problem? I noticed no errors in the compilation of `jaxlib`. **If**...

Thanks for the info @hawkinsp. Not great for me right now but might save others some time if the future :+1:

Any news on this?

Can't you just use this? ``` ./mcrcon -H localhost -P 25575 -p $(cat /path/to/pw/file) ```

Are the current best practices for caching documented somewhere?

Any progress on this?

Thanks for your quick response! I think there are already some issues concerning linear ops, but not specifically eigendecomp or batched SVD. Also thanks a lot for the explanation! I'll...

Hey @hawkinsp, thank you for getting back on this! Tbh I have not looked into this so far. It would be great if you could have a look too!

Thank you @hawkinsp, this is great! Non-symmetric eigendecomposition is not very urgent for me, especially if it's so cumbersome to add to jax. Concerning batched SVD I have a question...