Michael Schellenberger Costa comments

Results 319 comments of


                                            Michael Schellenberger Costa

Add workflow to build RAPIDS from source with local CCCL

I am super excited about this thanks a lot for working on this 😻

Enable including atomics and friends in TUs that do not support them.

> I agree with @miscco that I don't think this is functionality we can advertise until we have a way to test it. I believe properly testing this requires us...

Enable including atomics and friends in TUs that do not support them.

> > use NV_IF_TARGET to test this within the current framework > > Sure, you'd just need to rewrite all of the current `atomic` tests that use `pre-sm-70` to instead...

Drop more of thrust type traits

/ok to test

Add dimensions description functionality to CUDA Experimental library

/ok to test

Ensure that {cr}begin works with types that pull in namespace std via ADL

@Artem-B could you verify that this fixes your issues?

Ensure that {cr}begin works with types that pull in namespace std via ADL

The obvious wrinkle here is that those types wont find `cuda::std::begin` but AFAIK there is nothing we can do there and it is better than always breaking compilation

Ensure that {cr}begin works with types that pull in namespace std via ADL

I also added a workaround for `swap` which suffers from the same issue

Ensure that {cr}begin works with types that pull in namespace std via ADL

@Artem-B I have pushed backported fixes here https://github.com/miscco/cccl/tree/backport_adl_fixes

Ensure that {cr}begin works with types that pull in namespace std via ADL

I love how fixing ambiguities with begin and friend brings up errors like: >thrust/system/cuda/detail/internal/copy_cross_system.h(94): error #694: reinterpret_cast cannot cast away const or other type qualifiers