Skip to content

Conversation

@linamy85
Copy link
Collaborator

Change

  • Make sure bfloat16, float32, float16, float8, float4 is supported for GEMM.
  • Updated gemm_multiple_run* yaml files to include test cases.

Test

apiVersion: v1
kind: Pod
metadata:
  name: microbenchmark
spec:
  restartPolicy: Never
  nodeSelector:
    cloud.google.com/gke-tpu-accelerator: tpu7x
    cloud.google.com/gke-tpu-topology: 2x2x1
  containers:
  - name: tpu-job
    image: python:3.12
    ports:
    - containerPort: 8431
    securityContext:
      privileged: false
    command:
    - bash
    - -c
    - |
      set -ex

      git clone https://github.com/linamy85/accelerator-microbenchmarks.git
      cd accelerator-microbenchmarks
      git checkout feature/new-dtype
      pip install -r requirements.txt

      python3 Ironwood/src/run_benchmark.py --config=Ironwood/configs/training/gemm_multiple_run.yaml

      sleep 36000

    resources:
      requests:
        google.com/tpu: 4
      limits:
        google.com/tpu: 4

Make sure `bfloat16`, `float32`, `float16`, `float8`, `float4` is
supported for GEMM.
Copy link
Collaborator

@chishuen chishuen left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@junjieqian junjieqian merged commit 36482ee into AI-Hypercomputer:main Jan 23, 2026
2 checks passed
@junjieqian
Copy link
Collaborator

Thanks! Merged this one as doing tests.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants