Bradley Dice
Bradley Dice
For context, I made [this PR to conda-forge](https://github.com/conda-forge/conda-forge-ci-setup-feedstock/pull/144) to make it possible to do autoselection of CPU/GPU builds using the `__cuda` metapackage that is present for new-ish versions of conda...
@jaimergp Can you give me more information on the reasoning behind this comment? I’m unaware of the context. > Problem is that we don't want to distribute CPU only builds....
It may help to investigate these errors with https://docs.nvidia.com/compute-sanitizer/ComputeSanitizer/index.html#stream-ordered-race-detection: `compute-sanitizer memcheck --track-stream-ordered-races all`
I'm putting this on pause until after I get #20800 and some other projects finished up first.
@wence- Can you add links to the relevant sections of code that need changed, to make this more actionable? Are all the changes we need in https://github.com/rapidsai/cudf/blob/branch-23.08/python/cudf/cudf/_lib/copying.pyx , or do...
It would be good for this global configuration to support changing the RMM memory resource. It would be sufficient to have some coarse control like what we have for [libcudf...
@lamarrr What does fdd5458 actually do to fix the problem? Did we identify a root cause?
If there's some kind of locking behavior causing deadlocks or an issue with NVRTC, we need to get deeper, because that means this feature isn't stable enough to use yet....
Readers of `cpp/tests/CMakeLists.txt` cannot tell that this problem exists, so the percent values used by those tests are now "magic". Please open a cuDF issue explaining what we know about...