Playground

Playground tasks for newcomers!

1. Setup

First, fork this repository to your own github account:

Create a docker container from image pjlab/deeplearning:v2.2.3-cuda12.8.0-torch2.7.0, for example:

# Params:
#   <container-name>: Your container name. Can be anything.
#   <container-port>: A port that nobody has used. For example: 1145.
#   <project-home>: A directory that stores all your projects. This directory should be shared by all containers.
#   <proxy-addr>: Your proxy address.
#   --cap-add=SYS_ADMIN: This is required for using Nsight Compute to profiling in docker.
docker run --gpus all --name <container-name> -it \
    -p <container-port>:22 --entrypoint /bin/bash \
    --cap-add=SYS_ADMIN \
    -v ~/<project-home>:/root/<project-home>  \
    -e HTTP_PROXY=<proxy-addr>  \
    -e HTTPS_PROXY=<proxy-addr>  \
    -e http_proxy=<proxy-addr>  \
    -e https_proxy=<proxy-addr>  \
    pjlab/deeplearning:v2.2.3-cuda12.8.0-torch2.7.0

Inside the container, clone the repo you forked:

cd /root/<project-home>
git clone https://github.com/<your-github-account-name>/playground.git

Remember to change your git config:

git config --global user.name "<your-name>"
git config --global user.email "<your-email>"

2. Tasks

No.	Discription	Directory	Document
1	CUDA Programming: GEMM	./task-1	./task-1/READEME.md