Skip to content

Commit 3406355

Browse files
cquil11claude
andcommitted
runners(h100-dgxc-slurm): wire AIPERF mmap cache mount + env
Matches the same pattern as launch_b200-dgxc, launch_h200-dgxc-slurm, launch_gb300-{nv,cw}, launch_mi355x-amds, launch_h200-{nb,cw}: define AIPERF_MMAP_CACHE_HOST_PATH on the host, bind-mount it to /aiperf_mmap_cache in the container, and expose AIPERF_DATASET_MMAP_CACHE_DIR=/aiperf_mmap_cache via --export. Host path: /mnt/nfs/sa-shared/gharunners/ai-perf-cache (sibling of the existing hf-hub-cache mount on the same NFS volume). Needed for the new qwen3.5-fp8-h100-sglang-agentic recipe to reuse the pre-built mmap dataset cache across runs rather than re-mmaping every job. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> Signed-off-by: Cam Quilici <cjquilici@gmail.com>
1 parent 72cf856 commit 3406355

1 file changed

Lines changed: 3 additions & 2 deletions

File tree

runners/launch_h100-dgxc-slurm.sh

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -280,6 +280,7 @@ EOF
280280
else
281281

282282
HF_HUB_CACHE_MOUNT="/mnt/nfs/sa-shared/gharunners/hf-hub-cache/"
283+
AIPERF_MMAP_CACHE_HOST_PATH="/mnt/nfs/sa-shared/gharunners/ai-perf-cache"
283284
SQUASH_FILE="/mnt/nfs/lustre/containers/$(echo "$IMAGE" | sed 's/[\/:@#]/_/g').sqsh"
284285
LOCK_FILE="${SQUASH_FILE}.lock"
285286

@@ -306,10 +307,10 @@ else
306307

307308
srun --jobid=$JOB_ID \
308309
--container-image=$SQUASH_FILE \
309-
--container-mounts=$GITHUB_WORKSPACE:/workspace/,$HF_HUB_CACHE_MOUNT:$HF_HUB_CACHE \
310+
--container-mounts=$GITHUB_WORKSPACE:/workspace/,$HF_HUB_CACHE_MOUNT:$HF_HUB_CACHE,$AIPERF_MMAP_CACHE_HOST_PATH:/aiperf_mmap_cache \
310311
--no-container-mount-home \
311312
--container-workdir=/workspace/ \
312-
--no-container-entrypoint --export=ALL,PORT=8888 \
313+
--no-container-entrypoint --export=ALL,PORT=8888,AIPERF_DATASET_MMAP_CACHE_DIR=/aiperf_mmap_cache \
313314
bash benchmarks/single_node/${SCENARIO_SUBDIR}${EXP_NAME%%_*}_${PRECISION}_h100.sh
314315

315316
scancel $JOB_ID

0 commit comments

Comments
 (0)