Michał Dominiak comments

Results 55 comments of


                                            Michał Dominiak

enabling limits for ppc64le

Hi folks, Can you try to use libcu++ that's included in CUDA 11.0 to test if it works? This repo isn't maintained, we'll try to push the libcu++ repo to...

Can a pool memory resource be used to avoid cudaMalloc on each async call?

Can you provide a full compilable reproducer?

[BUG] atomic is not trivially copyable when used inside template class

The difference for here is probably which of the compiler frontends (i.e. nvcc or the host compiler) is instantiating the static assertion. For the record, atomics are not copyable, and...

[BUG] atomic is not trivially copyable when used inside template class

Oh. Just noticed you said it compiles on Linux, I wonder why that is.

[BUG] atomic is not trivially copyable when used inside template class

We seem to have lost the deleted copy ctor declaration. That's the real bug here.

[BUG] atomic is not trivially copyable when used inside template class

Yes; see https://en.cppreference.com/w/cpp/atomic/atomic/atomic.

[BUG] atomic is not trivially copyable when used inside template class

cc @cwharris

Instead of running all tests for each SM architecture, add specific variants of tests for each SM architecture

I don't understand. Do you want to create a test per SM per every test we have? Why. If that's what this is asking for, that create a massive maintenance...

Document why combining pipeline<thread_scope_thread>/barrier<thread_scope_thread> with memcpy_async is legal

I don't think our current docs imply that there's inter-thread synchronization, because I don't think we mention any side threads in memcpy_async docs; we only say that we issue an...

Rename `.upstream-tests`

I'm planning to do away with the subdirectory entirely and just promote the symlinks to be the actual directories inside `.upstream-tests`.