Jake Hemstad

[BUG] libcudf must customize the Thrust/CUB namespace

12

**Describe the bug** For header-only libraries like Thrust/CUB, it can be problematic when an application inadvertently ends up with multiple or mismatched versions of the headers. Thrust/CUB provide a way...

[REQUEST]: Add descriptions of PTX assembly instructions for CUDA C++

2

### Is your feature request related to a problem? Please describe Today, for Standard C++, in the assembly pane you can hover over instructions to get a tool tip that...

Relax input type requirements for thrust::merge

1

`thrust::merge` states this requirement: > InputIterator2 and InputIterator1 have the same value_type https://thrust.github.io/doc/group__merging_ga3d2776685a00dca265399f411784acec.html cppreference for `std::merge` has a slightly more relaxed requirement: > The signature of the comparison function should...

transform_iterator fails with callables whose arguments are passed by non-const l-value ref

2

Minimal reproducer: https://godbolt.org/z/s4oPhqx3W This fails because it cannot deduce the return type of the callable. A workaround is to explicitly define the `result_type` in the callable: https://godbolt.org/z/MTM8b1zW9 However, I believe...

While debugging https://github.com/NVIDIA/thrust/issues/1588 I found that a transform_iterator with a reference `result_type` would fail: https://godbolt.org/z/o6aTaE8Tz I traced this down to the fact that `thrust::detail::has_result_type` fails with reference `result_type`s: https://godbolt.org/z/b1d8PMjbq This...

Need docs for `atomic_ref`

1

We have docs for [`cuda::atomic`](https://nvidia.github.io/libcudacxx/extended_api/synchronization_primitives/atomic.html), but none for `cuda::atomic_ref`. We should fix that. I think this can largely be a copy/paste of the `cuda::atomic` docs as a starting point.

Make `cuda::aligned_size_t` available in a more appropriate header than `<cuda/barrier>`

3

The [`cuda::aligned_size_t`](https://github.com/NVIDIA/libcudacxx/blob/4f42427dfe5fd88672c29c279637e0ccf5b47478/include/cuda/std/barrier#L28-L36) type is currently defined in ``. This requires me to include `` any time I wish to use `cuda::aligned_size_t`. This is especially problematic as merely including `cuda/barrier` prevents...

Update `cuda::std::invoke_result` to emit diagnostic when attempting to query return type of extended lambda in host code

1

https://github.com/NVIDIA/thrust/issues/1688 made me think we should put this check directly into `cuda::std::invoke_result`. ``` #if defined(__NVCC__) && defined(__CUDACC_EXTENDED_LAMBDA__) && !defined(__CUDA_ARCH__) static_assert(!__nv_is_extended_device_lambda_closure_type(F), "Attempt to use an extended __device__ lambda in a context...

Add support for <chrono> Time Zone functionality

2

I would like support for the timezone database and time zone APIs added in [`` in C++20](https://en.cppreference.com/w/cpp/chrono). This functionality should also be backported to C++14.

cuda::(counting/binary)_semaphore unusable in __shared__ memory

1

Attempting to use a `__shared__ cuda::(counting/binary)_sempahore` will result in a hard compile error. ``` error: initializer not allowed for __shared__ variable ``` https://godbolt.org/z/68Wq3Eon1 In contrast, a `__shared__ cuda::barrier` only issues...