[CI] Move compatible 1xL4 jobs to H200 35GB MIG (#43024)

Signed-off-by: Simon Mo <simon@inferact.ai>
Co-authored-by: Simon Mo <simon@inferact.ai>
Co-authored-by: OpenAI Codex <noreply@openai.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
This commit is contained in:
Kevin H. Luu
2026-07-19 19:21:25 -07:00
committed by GitHub
co-authored by Simon Mo OpenAI Codex OpenAI Codex
parent ace9fda495
commit 1dcbbd9cac
17 changed files with 129 additions and 15 deletions
@@ -384,8 +384,13 @@ def test_fp32_cache_state(
example_prompts, max_tokens, num_logprobs
)
# Leave enough headroom for repeated engine initialization on a
# 32.5 GiB MIG.
with vllm_runner(
model, max_num_seqs=MAX_NUM_SEQS, **{cache_dtype_param: "float32"}
model,
max_num_seqs=MAX_NUM_SEQS,
gpu_memory_utilization=0.9,
**{cache_dtype_param: "float32"},
) as vllm_model:
vllm_outputs = vllm_model.generate_greedy_logprobs(
example_prompts, max_tokens, num_logprobs