pika
pika copied to clipboard
pika-org
pika builds on C++ std::execution with fiber, CUDA, HIP, and MPI support.
While using `async_rw_mutex` a few additional use cases for `async_rw_mutex` have popped up. Implementing them in pika would simplify implementations in DLA-Future and/or allow certain functionality that would otherwise not...
Builds with Clang on macOS
Compiler error: https://cdash.cscs.ch/test/66321134. Triggered by triggered by the call to `vm.count("seed")` (https://github.com/pika-org/pika/blob/87b10a0b70973cbaf164c46118444839501e0d56/libs/pika/async_cuda/tests/unit/cuda_sender.cu#L129).
The compilation of synchronize.cu is very slow to compile, this part of the code is the problem: https://github.com/pika-org/pika/blob/8e6ac7e251a75e5e468de7c987f2f1d074151943/libs/pika/async_cuda/tests/performance/synchronize.cu#L150-L168 A minimal version of that test limiting the loop unrolling to only...
`--pika:print-bind` is currently handled at a very late stage, inside the `run_helper` task before `pika_main` is run. This is presumably done so that it happens after "late" command line options...
`std::move_only_function` is available from C++23 onwards.
Removing `pika::init` could simplify starting the runtime noticeably and make spawning of work on the runtime more explicit (usually a good thing). It would remove the special casing of `pika_main`...
We could initially refer to hpx's documentation and provide only the things that differs from it. But it might diverge too much from it in the future to keep it...
We've disabled many CMake options by default, even in debug mode, for performance. However, sometimes it's still useful to enable those debug options. Instead of needing to know which debug...