[FEA]: constexpr log functions

[RAMitchell]

Is this a duplicate?

• [x] 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.

I was blocked making std::normal_distribution constexpr because cuda::std::log is not yet constexpr. This is part of C++26.

Describe the solution you'd like

.

Describe alternatives you've considered

No response

Additional context

No response

[miscco]

we do have constexpr ilogb is that sufficient for you?

constexpr math functions are out of our hands because the generally need compiler and runtime support

[RAMitchell]

I need natural log unfortunately. Is it theoretically possible to use std::is_constant_evaluated() to "overload" for constexpr?

[miscco]

There are some things we can do, but there are issues wrt to denormals, so I would not hold my breath