https://github.com/megengine/megray
A communication library for deep learning
https://github.com/megengine/megray
Last synced: 5 months ago
JSON representation
A communication library for deep learning
- Host: GitHub
- URL: https://github.com/megengine/megray
- Owner: MegEngine
- License: other
- Created: 2020-03-18T10:07:29.000Z (almost 6 years ago)
- Default Branch: master
- Last Pushed: 2024-08-08T08:02:33.000Z (over 1 year ago)
- Last Synced: 2024-12-09T11:52:50.423Z (about 1 year ago)
- Language: C++
- Homepage:
- Size: 211 KB
- Stars: 50
- Watchers: 11
- Forks: 9
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
Awesome Lists containing this project
README
# MegRay
MegRay is a cross-platform communication library providing point-to-point and collective communication methods, such as send, recv, all\_gather, all\_reduce, reduce\_scatter, reduce, broadcast, gather, scatter and all\_to\_all. In the area of deep learning, these methods can be utilized for implementing distributed training framework, including data parallel and model parallel. Currently there are three backends, nccl and ucx for cuda platform, rccl for rocm platform. In the future, algorithms on more platforms will be added.
## Build
0. prepare third party repositories.
```
./third_party/prepare.sh
```
1. Make a directory for build.
```
mkdir build
cd build
```
2. Generate build configurations by `CMake`.
```
cmake .. -DMEGRAY_WITH_TEST=ON
```
3. Start to build
```
make
```