Allison Piper
Allison Piper
nvcc defaults to rdc-off, nvc++ defaults to rdc-on. We need to explicitly enable or disable these flags for each CUDA target, rather than just enabling them when needed.
Reported via NVBug 3298282. Reproduced locally. Appears to be a regression (see comments in test case). Only happens when launching from the device, suggesting that the issue is in the...
The `par_nosync` execution policy is missing from the new documentation site. We need to investigate why [this docstring](https://github.com/NVIDIA/thrust/blob/9ca1210129541f676a373c85f97bd5bfb19b921e/thrust/system/cuda/detail/par.h#L190-L236) isn't being parsed.
Naively trying to add these tests caused issues on windows due to CMAKE_OBJECT_PATH_MAX -- the paths to the object files exceed a system-dependent maximum number of characters. We should find...
Some of the Thrust unit tests check that static assertions fire under particular situations. For example, [generate_const_iterators](https://github.com/NVIDIA/thrust/blob/81d683e42047913dcdcf7239df282806e2dc019c/testing/generate_const_iterators.cu) tests that `thrust::generate` and `thrust::fill` trigger static assertions when called with const iterators....
`thrust/system/cuda/detail/reduce.h` contains a complete reduction implementation in the `__reduce` namespace. In 8702bfe89640c9efcf68cf3ad2b89d4d00e96494, the `thrust::cuda_cub::reduce` entry points were changed to just call `cub::DeviceReduce`, but the old implementation remains. The `__reduce` implementation...
Requires NVIDIA/cub#425. This branch makes the following changes to gpuCI: * Splits `main` integration builds from `prb` pull request builds. * Split `gpu` builds into `cuda` and `nvhpc` subsets. *...