Azure-Samples / Azure-Samples/aks-labs

[LAB] AKS GPU Cluster on AML

Open
#92 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
TypeScript
Stars
60
Forks
70
PR merge metrics
No merged PRs in 30d

Description

**Please describe the lab you are proposing**
This Lab will be focusing on creating an AML (Azure Machine Learning) compute backed by an AKS (Azure Kubernetes Service) GPU with IB cluster and running a distributed AML job on it with Infiniband support.

For more details:
https://github.com/maljazaery/AML_AKS_GPU_IB_CLUSTER/blob/main/main.ipynb

**If possible, provide a rough outline of the proposed lab**
- Prepare Azure resources
- Deploy and configure the AKS gpu cluster with Infiniband
- Install required Kubernetes and AzureML components
- Attach the cluster to AzureML and validate Infiniband performance
- Submit AML Job for NCCL test.

**Potential category. Please check one.**

- [ ] Networking
- [ ] Security
- [ ] Storage
- [ ] Operations
- [X ] AI/ML
- [ ] Platform Engineering
- [ ] Other. Please specify in a new line below if none of the categories above fit.

**Are you willing to be added to the CODEOWNERS file for this lab?**

- [ X] Yes

Contributor guide

Open the contributing guide

Research direction

Start by reading and running main.ipynb from the linked AML_AKS_GPU_IB_CLUSTER repository. Compare its Azure resource preparation, AKS GPU and Infiniband setup, AzureML attachment, and NCCL test steps with this lab proposal. Done means the lab documents and demonstrates a distributed AML job with Infiniband support.

Written by the indexing model from the issue text.

Assessment

Tech stack
azure, jupyter-notebook, kubernetes
Domain
cloud, distributed-systems, infrastructure, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.