Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

ENH: Add RISC-V Vector V1.0 Support #26200

Open
luyahan opened this issue Apr 3, 2024 · 4 comments
Open

ENH: Add RISC-V Vector V1.0 Support #26200

luyahan opened this issue Apr 3, 2024 · 4 comments
Labels
01 - Enhancement component: SIMD Issues in SIMD (fast instruction sets) code or machinery

Comments

@luyahan
Copy link

luyahan commented Apr 3, 2024

I'm currently planing in add RISC-V Vector support for NumPy.

Originally, I want to port rvv into _simd. But I found a great solution NEP-054.
I read NEP-054 (https://numpy.org/neps/nep-0054-simd-cpp-highway.html#discussion) and would like to get more information about its current status.
I noticed some PR had been opened ( #25781 and #25934 ) for the NEP-054 but still under review and discussion.
I wonder if there are any specific tasks or contributions I can undertake to assist with its progress.

And whether I should plan RVV support following the current NPY_SIMD implementation style or in the NEP-054 way?
Any suggestion or reference for this topic?

@luyahan luyahan closed this as completed Apr 3, 2024
@luyahan luyahan changed the title ENH: How about the schedule for NEP-054 ENH: Add RISC-V Vector Support Apr 3, 2024
@luyahan luyahan changed the title ENH: Add RISC-V Vector Support ENH: Add RISC-V Vector V1.0 Support Apr 3, 2024
@luyahan luyahan reopened this Apr 3, 2024
@mattip
Copy link
Member

mattip commented Apr 3, 2024

One way to get involved would be to join the optimization slack channel

@luyahan
Copy link
Author

luyahan commented Apr 3, 2024

optimization slack channel

Thanks~
I send a email to numpy-team@googlegroups.com

@rgommers rgommers added 01 - Enhancement component: SIMD Issues in SIMD (fast instruction sets) code or machinery labels Apr 3, 2024
@r-devulap
Copy link
Member

I am not really familiar with RISC-V myself, but Google highway already supports RISC-V: RVV (1.0). Is that what you are looking for?

Looks like NumPy currently lacks both compiler time and run time cpuid detection of RISC-V SIMD features. We will need this to leverage RISC-V intrinsics (be it highway or NPY intrinsics). Perhaps you could start with adding those?

@mattip
Copy link
Member

mattip commented Apr 3, 2024

The place to add that would be in the vendored meson submodule, in the features component of meson.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
01 - Enhancement component: SIMD Issues in SIMD (fast instruction sets) code or machinery
Projects
None yet
Development

No branches or pull requests

4 participants