[FEA]: Use `__builtin_bitreverseg` in `cuda::bit_reverse`

[davebayer]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

libcu++

Is your feature request related to a problem? Please describe.

clang-23 implements __builtin_bitreverseg builtin that we should use in our cuda::bit_reverse implementation.

Describe the solution you'd like

Use the builtin.

Describe alternatives you've considered

No response

Additional context

No response

[Revaj]

Hello, I'm interested to take this issue :)

[davebayer]

Hi @Revaj, thanks for you interest! Feel free to make a PR, you can follow the implementation of cuda::std::countl where we use __builtin_clzg.

However, Clang 23 is far from being supported as a CUDA host compiler (it's not even released), so we wouldn't be able to test & merge the PR until then. In fact, Clang 22 was released just days earlier.

#7778 has a similar nature and we could test it out at least locally, wouldn't you rather be interested in that one?

[Revaj]

Hi @Revaj, thanks for you interest! Feel free to make a PR, you can follow the implementation of cuda::std::countl where we use __builtin_clzg.

However, Clang 23 is far from being supported as a CUDA host compiler (it's not even released), so we wouldn't be able to test & merge the PR until then. In fact, Clang 22 was released just days earlier.

#7778 has a similar nature and we could test it out at least locally, wouldn't you rather be interested in that one?

Hi, I'd be happy to take a look at #7778 instead. :)

[syan-h]

i m new and i wanna take this as my first issue , is it open for me to grab?

[fbusato]

Hi @syan-h, yes, it would be great! My understanding is that @Revaj focuses on #7778

[Revaj]

Hi @syan-h yes, please feel free to grab it.