[ROCm Test-First CI]: Day 0 FlyDSL MI455X UALoE72 Helios Rack open source github CI by AMD Advancing AI July 22 2026
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 282
- Forks
- 120
- Avg merge
- 1d 21h
- Merged PRs (30d)
- 67
Description
Suggestion Description
hi @andyluo7 @sunway513
+viz @chunfangamd @powderluv
after discussion with AMD senior leader, in an effort to start MI455 gfx1250 helios off on the correct path, the goal is to have MI455X UALoE72 Helios rack CI by AMD Advancing AI on July 22 2026.
Day 0 MI455X UALoE72 Helios rack open source github upstream CI should aim for:
- at least nightly coverage of all the FlyDSL kernels of at least 2 major 2026 frontier OSS model (i.e. kimi k2.6, minimax m3, deepseekv4 pro, etc) for both FP4 & FP8
- nightly accuracy & performance regression testing
- nightly or per PR testing of MegaMoE kernels that are at least EP16
- FlyDSL SGLang integration testing similar to https://github.com/ROCm/FlyDSL/blob/main/.github/workflows/flydsl-sglang-integration.yaml
- FlyDSL vLLM integration testing similar to https://github.com/ROCm/FlyDSL/blob/main/.github/workflows/flydsl-vllm-integration.yaml
Operating System
No response
GPU
No response
ROCm Component
No response
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reading .github/workflows/flydsl-sglang-integration.yaml and .github/workflows/flydsl-vllm-integration.yaml, which the issue identifies as integration-testing examples. Clarify the initial scope for MI455X UALoE72 Helios coverage, model and kernel targets, and nightly versus per-PR runs; done means the agreed FlyDSL, SGLang, vLLM, accuracy, and performance checks run in upstream GitHub CI.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- github-actions, python
- Domain
- ci-cd, machine-learning, testing-qa
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100