Eric Niebler
Eric Niebler
The symptoms of this problem are described in NVIDIA/cccl#705. Really, there are two problems with the way Thrust's iterator tags are currently defined: 1. The `*_device_*` iterator tags are erroneously...
If you navigate to [Submitting a PR / Configure CMake Builds](https://nvidia.github.io/thrust/contributing/submitting_a_pr.html#configure-cmake-builds), you'll find a broken link to "CMake Options". The URL points to: ```html https://nvidia.github.io/thrust/contributing/setup/cmake_options.md ``` There are two things...
These are the scheduler affinity changes from the unstable branch, with minor changes. I hope to refine these changes to make them ready for trunk (see below). I will use...
For example, the following shows how to do it specifically for `unifex::task`: https://godbolt.org/z/a5cqTEYze Generalize this to typed senders. ```c++ #include #include #include #include #include #include #include #include #include #include namespace...
Additionally, make the following "magical" in a `task`-returning coroutine just like `co_await schedule(sched);` currently is: ```c++ co_await schedule_at(sched, time_point); co_await schedule_after(sched, duration); ```
Those queries will be punched through a stack of coroutine the way stop tokens are currently. Implement `task` in terms of `basic_task` (but probably not as a simple alias because...