Peter Lindstrom

I ran across [this paper](https://arxiv.org/pdf/2109.05410.pdf) that seems to have tackled this problem. Not sure if their code is available.

@data-panda No, this release does not include the latest CUDA and HIP work we have been doing. That will end up in the next release. Regarding CUDA streams specifically, that...

I'm not sure what `pip install` does when installing from source. If you want to build zfpy with CMake, you have to add `-DBUILD_ZFPY=ON`. Maybe there's a way to tell...

Great. Does `pip install` still fail when you add `-DBUILD_ZFPY=ON`?

Thanks for the suggestion. I do not have any experience with pkg-config, but from a quick glance at the documentation I would suggest adding this to the repo: % cat...

I'd like to go ahead and close out this issue, but am unsure how to best address it. @ast, if you could submit a .pc file that works with CMake,...

That does seem awfully slow (4 MB in over a second). We are aware that our HIP implementation is quite a bit slower than the CUDA implementation, though we're still...

@pkoosha A brief update: We've done some work to improve the performance of the HIP implementation. In particular, the high compression (but not decompression) overhead you're seeing is likely due...

@drdnichols I agree that it makes sense to support this functionality. I would have to think a little about the semantics of `view::get()`, as some of the views maintain their...

Since zfp already supports parallel CUDA decompression, I assume you're referring to the OpenMP implementation of `zfp_decompress` (and not the compressed-array C++ classes). We have plans to add support for...