Bram Veenboer

See the attachment for some tuning results on AMD Radeon FuryX (using driver 1800.8). [amd-radeon-furyx.zip](https://github.com/CNugteren/CLBlast/files/994655/amd-radeon-furyx.zip)

I ran tuning using CLBlast 1.5.0 on a NVIDIA Titan RTX (using driver 415.125): [titanrtx-415.125.tar.gz](https://github.com/CNugteren/CLBlast/files/2850387/titanrtx-415.125.tar.gz)

The `NVMLObserver` is covered by these two files: - https://github.com/KernelTuner/kernel_tuner/blob/4dbcb66fad3561f7ee24d270b60d597e68fc4cdf/test/test_observers.py#L4 - https://github.com/KernelTuner/kernel_tuner/blob/4dbcb66fad3561f7ee24d270b60d597e68fc4cdf/examples/cuda/vector_add_observers.py#L9 Regarding the former, you would need something similar to `@skip_if_no_pynvml`. Maybe it's enough to check the Linux kernel...

> I'm getting a 404 now on https://git.astron.nl/RD/recruit/nvmetrics but I was able to access the page last Friday, what happened? There is something going on at the ASTRON GitLab server,...

I have pushed `nvmetrics` to a new repository: https://github.com/nlesc-recruit/nvmetrics

> Luckily I already cloned it on Friday. I just managed to get around the permission issues related to reading performance counters on WSL, about to test the NCUObserver :-)...

> I'll make some changes to see that this would look like These changes indeed make the code more consistent with the existing code, but it also makes it harder...

> Please use the repository to download and build the source, which are available from https://github.com/NVIDIA/cuda-python/releases. I have rewritten the package file to download and install from source.

I noticed that `cuda-python` depends on `pyclibrary`, which is not available in Spack yet. A new package file for this library is added. While looking into this, I stumbled upon...

Now that this PR has been migrated to https://github.com/spack/spack-packages/pull/173, I think this PR can be closed.