AZ Health Checks v0.4.3

• CodeQL security scan
• Trigger Guest Health Reporting (GHR) automatically from NHC
• Status for HCv4, NCv4, NDv2, ND96amsr v4, and ND96isr H100 v5 from the validation pipeline
• Support AMD MI200/300 SKUs for conf files
• Add RCCL all reduce tests
• Log pkeys and kernel version
• Add NCCL all reduce tests
• Bug fixes

AZ Health Checks v0.4.2

What's Changed

• Docker image now uses MCR
• HC44 now supported
• Bug Fixes

AZ Health Checks v0.4.1

What's Changed

• Introduction of Docker NHC
• Increased visibility into test results
• Increased logging
• Bug fixes

AZ Health Checks v0.2.9

• Adding additional logging and Kusto functionality
• Bug fixes

AZ Health Checks v0.4.0

This is a prerelease of the docker version of AzNHC.

The functionality remains the same. The major differences are the following:

• No installation needed
• Docker is now a prerequisite
• The only set up is pulling the docker image

The run script and checks behave the same way.

AZ Health Checks v0.2.8

What's Changed

• Adding AMD GPU SKU support
• Add feature to extend tests by adding a secondary conf file
• Bug Fixes

AZ Health Checks v0.2.7

What's Changed

• NCv3, NCv4, NCv5, NDv2 support
• Hbv3 smaller sizes support
• NvBandwidth tool from Nvidia added to measure Nvlink and GPU BW
• Refresh install script.

AZ Health Checks v0.2.6

What's Changed

• Bug fixes
• Refactoring of IB write tests
  • NDv5 SKU IB tests no longer communicate between devices. The HCA device loop back to themselves to prevent IB traffic from leaving the node
  • Renaming tests for clarity
• Documentation update

Az Health Checks v0.2.5

Whats Changed:

• Distributed NHC functionality
  • Adds ability to launch NHC in a distributed fashion
  • Adds ability to launch NHC with Slurm
• Changes to support Cycle cloud usage
• Accelerated network checks addition

New Contributors

• @mpwillia made their first contribution in #35

Full Changelog: v0.2.4...v.0.2.5

Az Health Checks v0.2.4

What's Changed

• Bug fixes