gahow/aituner
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add Frontier linear profiler smoke

Browse Source
This commit is contained in:
Gahow Wang
2026-07-16 22:11:04 +08:00
parent 32ab584789
commit 48d2c2fb80
2 changed files with 73 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
runs/frontier-qwen30-vllm020-profile-v1/jobs_frontier_linear_smoke.toml Normal file
View File

@@ -0,0 +1,18 @@
version = 1
[[jobs]]
name = "qwen30-vllm020-frontier-linear-smoke-20260716-v1"
gpus = 1
gpu_model = "H20"
hosts = ["dash0"]
command = "cd /home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-v1/runs/frontier-qwen30-vllm020-profile-v1 && timeout --signal=TERM --kill-after=30s 1020 bash run_frontier_linear_smoke.sh"
artifacts = ["artifacts/frontier-linear-smoke-v1"]
[jobs.env]
HOME = "/tmp/wjh"
XDG_CACHE_HOME = "/tmp/wjh/.cache"
VLLM_CACHE_ROOT = "/tmp/wjh/.cache/vllm"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/frontier-linear-smoke-v1"
VENV_ROOT = "/tmp/wjh/venvs/vllm-0.20.0-cu129-profiler-v1"
FRONTIER_ROOT = "/home/admin/cpfs/wjh/frontier-qwen30-vllm020-profile-v1/Frontier"
VLLM_SOURCE = "/home/admin/cpfs/wjh/agentic-kv/third_party/vllm_v20_build"

55
runs/frontier-qwen30-vllm020-profile-v1/run_frontier_linear_smoke.sh Normal file
View File

@@ -0,0 +1,55 @@
#!/usr/bin/env bash
set -euo pipefail
OUTPUT_ROOT="${OUTPUT_ROOT:?OUTPUT_ROOT must be set}"
VENV_ROOT="${VENV_ROOT:-/tmp/wjh/venvs/vllm-0.20.0-cu129-profiler-v1}"
FRONTIER_ROOT="${FRONTIER_ROOT:-/home/admin/cpfs/wjh/frontier-qwen30-vllm020-profile-v1/Frontier}"
VLLM_SOURCE="${VLLM_SOURCE:-/home/admin/cpfs/wjh/agentic-kv/third_party/vllm_v20_build}"
mkdir -p "${OUTPUT_ROOT}/logs" "${OUTPUT_ROOT}/provenance" "${OUTPUT_ROOT}/profiles"
exec > >(tee -a "${OUTPUT_ROOT}/logs/smoke.log") 2>&1
IFS=',' read -r -a GPU_IDS <<< "${CUDA_VISIBLE_DEVICES:?fleet GPU is required}"
if [[ "${#GPU_IDS[@]}" -ne 1 ]]; then
echo "ERROR: expected exactly one GPU, got ${CUDA_VISIBLE_DEVICES}" >&2
exit 1
fi
echo "PROFILE_LAUNCH_ECHO host=$(hostname) gpu=${CUDA_VISIBLE_DEVICES} model=qwen3-a3b-30b-moe runtime=vLLM-0.20.0+cu129 profiler=Frontier@d9cfeb6 operator=linear,norm,rope,embedding tp=1,2,4 tokens=8 precision=BF16 method=cuda_event output=${OUTPUT_ROOT} expected_wall=2-6m hard_wall=900s hard_gpu_cap=0.25_H20h"
date -u +"START_UTC=%Y-%m-%dT%H:%M:%SZ"
nvidia-smi --query-gpu=index,name,driver_version,memory.used,utilization.gpu --format=csv,noheader
test "$(git -C "${FRONTIER_ROOT}" rev-parse HEAD)" = "d9cfeb6d8791fbf2f295dd9744c56a666171776e"
test "$(git -C "${VLLM_SOURCE}" rev-parse HEAD)" = "88d34c6409e9fb3c7b8ca0c04756f061d2099eb1"
git rev-parse HEAD > "${OUTPUT_ROOT}/provenance/aituner.commit"
git -C "${FRONTIER_ROOT}" rev-parse HEAD > "${OUTPUT_ROOT}/provenance/frontier.commit"
git -C "${VLLM_SOURCE}" rev-parse HEAD > "${OUTPUT_ROOT}/provenance/vllm-source.commit"
sha256sum run_frontier_linear_smoke.sh \
> "${OUTPUT_ROOT}/provenance/source.sha256"
uv pip freeze --python "${VENV_ROOT}/bin/python" \
> "${OUTPUT_ROOT}/provenance/pip-freeze.txt"
nvidia-smi --query-gpu=index,uuid,name,driver_version,memory.total \
--format=csv,noheader > "${OUTPUT_ROOT}/provenance/gpus.csv"
cd "${FRONTIER_ROOT}"
timeout --signal=TERM --kill-after=30s 780 \
"${VENV_ROOT}/bin/python" -m frontier.profiling.linear_op.main \
--disable_ray \
--num_gpus 1 \
--output_dir "${OUTPUT_ROOT}/profiles" \
--device h20 \
--models qwen3-a3b-30b-moe \
--num_tensor_parallel_workers 1 2 4 \
--num_tokens_list 8 \
--profile_method cuda_event \
--precision BF16 \
--is_moe \
--yes
find "${OUTPUT_ROOT}/profiles" -type f -size +0c -print \
> "${OUTPUT_ROOT}/provenance/profile-files.txt"
test -s "${OUTPUT_ROOT}/provenance/profile-files.txt"
find "${OUTPUT_ROOT}/profiles" -type f -print0 | sort -z | xargs -0 sha256sum \
> "${OUTPUT_ROOT}/profiles.sha256"
date -u +"END_UTC=%Y-%m-%dT%H:%M:%SZ"
echo "FRONTIER_LINEAR_SMOKE_COMPLETE"