Mikael Simberg

Debugging MKL linking

4

Not for merging.

gcc: Add mold variant to use mold for linking

4

Triggered by https://github.com/spack/spack/pull/44034#issuecomment-2096707791. This adds a `mold` variant to `gcc` which, when enabled, adds `mold` to GCC's search paths and `-fuse-ld=mold` to use `mold` without a user having to specify...

Avoid overallocation when underlying allocation is guaranteed to be sufficiently aligned

2

**Is your feature request related to a problem? Please describe.** The underlying allocator may have sufficient alignment, but `aligned_allocate` always overallocates to guarantee the alignment, even if it may not...

`async_rw_mutex` currently allocates many shared states through `shared_ptr`s. Look into reducing the number of allocations.

https://github.com/pika-org/pika/issues/63#issuecomment-1963556821 and #63.

This requires testing in DLA-Future to see that lifetime requirements are satisfied with the reference implementation. This work can already be started with the `PIKA_WITH_P2300_REFERENCE_IMPLEMENTATION` option before e.g. nvcc/nvc++ is...

`pika::detail::small_vector` seems to be significantly slower than `boost::container::small_vector`. It's unclear if it's "just" a bug in the implementation or if it's something more inherent in the use of the standard...

This could be useful in a few places: - Forwarding values for the bulk customization for `cuda_scheduler` (and possibly also for the `thread_pool_scheduler` one) - For creating wrappers where you...