Paul Taylor
Paul Taylor
> And I believe CUDA handles IPC handle GC via client bindings tying them to lang GCs CUDA will clean these up on shutdown, otherwise you'll be at the whim...
In order to IPC buffers that back a (zero-copy) sliced Python Series, you'd have to IPC the entire original buffer + communicate the offset + length along with the IPC...
@Luxcium not entirely sure what you've tried, but generally the 3 things you will need (in addition to the driver) are: * [docker-ce](https://docs.docker.com/get-docker/) * [nvidia-container-toolkit](https://github.com/NVIDIA/nvidia-container-toolkit) * [nvidia-container-runtime](https://github.com/NVIDIA/nvidia-container-runtime) I know it's...
@Luxcium does this work for you? https://github.com/NVIDIA/nvidia-docker/issues/706#issuecomment-851816502
@nicolasgere are you running the WebGL demo from inside a docker container? If so, are you passing the `--runtime=nvidia` flag (and have the [`nvidia-container-toolkit` installed](https://github.com/rapidsai/node/blob/8430aa9c68a28dc5158d504cd630696b0821b98b/docs/docker/installation.md#installing-the-nvidia-container-toolkit))? Our dev and runtime `docker-compose`...
Looks like I copy/pasted the dependency list from the docker container, but the [comments](https://github.com/rapidsai/node/blob/main/modules/core/bin/install-deps/debian.sh#L78-L85) here aren't valid in a bash script. If you delete the comment lines, `yarn dev:install-cpp-dependencies` should...
That's strange, `yarn docker:build:devel` should've built and cached the `base` image before building `main`. We don't publish to dockerhub at the moment, but we do publish to the GitHub container...
@brycelelbach does libcu++ have plans to load the system timezone database? I recall @ogiroux saying that might not be supported yet.
@BradReesWork could you also take a look at https://github.com/rapidsai/cugraph-ops/pull/198? :pray:
This also works: ```shell # This is for make/ninja SKBUILD_BUILD_OPTIONS="-j${PARALLEL_LEVEL}" \ pip wheel \ # and this is for `setup.py build_ext` i.e. for use by `cythonize()` --global-option="build_ext" \ --global-option="-j${PARALLEL_LEVEL}" \...