Daniel Arndt
Follow-up to https://github.com/kokkos/kokkos/pull/8017. Addressing https://github.com/kokkos/kokkos/pull/8017#pullrequestreview-2799703870: > Since this basically passed testing: I am ok merging it, but Daniel open a follow on PR to take away the #ifdef HPX. Since...
Fixes #7946. Based on #7959.
There doesn't seem to be any good reason to both have `View::stride([0-7])` and `View::stride_[0-7]`. https://godbolt.org/z/nxKsfPM7T shows that they compile to the same assembly with compile-time input.
https://github.com/kokkos/kokkos/pull/7509 caused a regression for lighweight kernels for parallel_reduce (like a dot product) in the range of 10^7 to 10^9 elements. Previous to that pull request we were using a...
https://github.com/kokkos/kokkos/pull/7775 was partly motivated by making sure that code like ``` DualView a("a", 10); A.modify(); Kokkos::parallel_for([=](int i){ a(i)=1; }); A.sync(); std::cout
ViewUniformType appears to be unused even in Trilinos and its purpose is unclear.
https://github.com/kokkos/kokkos/blob/850b88e5259eba514e15829b5f76cf20b581e5f6/core/src/View/Kokkos_ViewCommonType.hpp#L109-L119 says that we should be able to get rid off `common_view_alloc_prop` once `View` is based on `mdspan`. Related to https://github.com/kokkos/kokkos/pull/7427/files#r1796207018.
Related to https://github.com/kokkos/kokkos-tutorials/pull/101/files#r1854601705. The point is to write the interfaces so that the calls to `parallel_reduce` and `parallel_scan` compile and are correct. Hence, we shouldn't actually call them in `Begin`...
We discussed in the developer meeting last week to remove Makefile support in the examples. This should for the most part be straightforward but there are a couple of examples...
In response to https://github.com/kokkos/mdspan/pull/413#issuecomment-2967400086 and https://github.com/kokkos/mdspan/pull/285#discussion_r1369043961, this pull request enforces the repository to be formatted with clang-format 16 using the same settings as for https://github.com/kokkos/kokkos.