Damien L-G
Damien L-G
It is something we would be generally interested in doing. Would you be willing to draft initial support for it plus basic testing? We can see that all headers are...
A `View` can still be *runtime* unmanaged without having that `MemoryUnmanaged` trait. In that case the `View::use_count()` member function will return `0` when called on the host-side to indicate that...
This came up recently when debating whether to exclusively print some backend-specific "IMPL_` macro if it is defined and say nothing otherwise. The reasoning was keeping the output terse and...
Then please open a separate PR for this change
What is this about? Did you open this issue in order to track progress on some upcoming work?
> on https://en.cppreference.com/w/cpp/algorithm/adjacent_difference am i missing it or the precondition is not stated ? It does say > If `op` invalidates any iterator (including any of the end iterators) or...
What do you mean when you say "the other part of the condition [is] missing"? > If `op` invalidates any iterator (including any of the end iterators) **or modify any...
> The only variable exported form kokkos modules at ALCF is KOKKOS_HOME OLCF prefixed them with `OLCF_` so no issue
> We just don't treat this variable in `nvcc_wrapper`. Any good reason why we couldn't/shouldn't add support for it? https://docs.nvidia.com/cuda/cuda-compiler-driver-nvcc/index.html#disable-warnings-w
The format check passed. Ignoring everything else