mirror of
https://github.com/saymrwulf/pytorch.git
synced 2026-05-15 21:00:47 +00:00
9de2085806
Summary: * purge hcSPARSE now that rocSPARSE is available * integrate a custom hcc and HIP * hcc brings two important compiler fixes (fixes hundreds of unit tests) * HIP brings a smart dispatcher that allows us to avoid a lot of static_casts (we haven't yet removed the automatic static_casts but this catches some occurrences the script did not catch) * mark 5 unit tests skipping that have regressed w/ the new hcc (we don't know yet what is at fault) * optimize bitonic sort - the comparator is always an empty struct - therefore passing it by value saves at least 3 bytes. It also removes an ambiguity around passing references to `__global__` functions Pull Request resolved: https://github.com/pytorch/pytorch/pull/11198 Differential Revision: D9652340 Pulled By: ezyang fbshipit-source-id: f5af1d891189da820e3d13b7bed91a7a43154690 |
||
|---|---|---|
| .. | ||
| jenkins | ||
| ubuntu-14.04-cpu-all-options | ||
| ubuntu-14.04-cpu-minimal | ||
| ubuntu-16.04-cpu-all-options | ||
| ubuntu-16.04-cpu-minimal | ||
| ubuntu-16.04-cuda8-cudnn6-all-options | ||
| ubuntu-16.04-cuda8-cudnn7-all-options | ||
| ubuntu-16.04-gpu-tutorial | ||
| readme.md | ||
Docker & Caffe2
Note: use nvidia-docker to run all GPU builds.
To get the latest source, rerun the docker builds using the Dockerfiles.
Docker images at https://hub.docker.com/r/caffe2ai/caffe2/ are a few months old, but will be refreshed soon.
Build like: docker build -t caffe2:cuda8-cudnn6-all-options .
Run like: nvidia-docker run --rm -it caffe2:cuda8-cudnn6-all-options python -m caffe2.python.operator_test.relu_op_test
For Docker on USB related instructions you can find some help on the gh-pages branch here