This lab teaches how to use NVIDIA's Nsight tools for analyzing and optimizing CUDA applications. Attendees will be using Nsight Systems to analyze the overall application structure and explore parallelization opportunities. Nsight Compute will be used to analyze and optimize CUDA kernels, using an online machine learning code for 5G.
This repository contains training content for the NVIDIA Nsight GTC 2021 lab. Follow the link to watch the session recording on the GTC website, or take the training by simply using the instructions in the jupyter notebooks.
After cloning this repo move into the task1 directory and then do:
docker-compose up -d
If you need to rebuild containers instead do:
docker-compose up -d --build
To check on running containers (for example to get their ports) do:
docker ps
To get logs for all running containers do:
docker-compose logs
If you want to follow logs add the -f flag:
docker-compose logs -f
If you only want logs for one container, get its service name from docker-compose.yml and do:
docker logs -f <service-name> e.g. docker-logs -f nsight
When finished, please spin down containers from within the task1 directory with:
docker-compose down
After running docker-compose up -d (see above) and confirming the containers have spun up, visit 127.0.0.1:9333/lab in your browser. You can modify the mapped port (e.g. 9333) via task1/.env by modifying NGINX_DEV_PORT.
The remote desktop can be accessed at the /nsight/ endpoint of the same host and port that the Jupyter notebook is running on. For example, if Jupyter is accessed at 127.0.0.1:9333 than the remote desktop is at 127.0.0.1:9333/nsight/.
The password is nvidia.
There is a shared file system mount between the lab (Jupyter) container and the nsight (remote desktop) container where the lab container's /dli/task directory is mounted into the nsight container's /root/Desktop/reports/ directory. See task1/docker-compose.yml for details or to edit.