[moe training] update benchmarks using more reliable b200 gpu

[danielvegamyhre]

GPU 0 on my devgpu seems to have perf issues, we get better performance benchmarks by pinning other GPUs:

CUDA_VISIBLE_DEVICES=7 python benchmark_scaled_grouped_mm_dq.py --compile

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/3174

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 Cancelled Job, 1 Pending

As of commit e41c878 with merge base b633f89 ():

CANCELLED JOB - The following job was cancelled. Please retry:

• Run 1xH100 Tests / test (H100, linux.aws.h100, --pre torch torchvision torchaudio fbgemm-gpu-genai --index-url https... / linux-job (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.