[CI][CPU] Split CPU-Distributed Tests into per-scenario labels (#41203)
Signed-off-by: haosdent <haosdent@gmail.com>
This commit is contained in:
@@ -69,11 +69,11 @@ steps:
|
|||||||
pytest -x -v -s tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_logprobs
|
pytest -x -v -s tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_logprobs
|
||||||
pytest -x -v -s tests/quantization/test_cpu_wna16.py"
|
pytest -x -v -s tests/quantization/test_cpu_wna16.py"
|
||||||
|
|
||||||
- label: CPU-Distributed Tests
|
- label: CPU-Distributed Tests (PP+TP)
|
||||||
depends_on: []
|
depends_on: []
|
||||||
device: intel_cpu
|
device: intel_cpu
|
||||||
no_plugin: true
|
no_plugin: true
|
||||||
source_file_dependencies:
|
source_file_dependencies: &cpu_distributed_deps
|
||||||
- csrc/cpu/shm.cpp
|
- csrc/cpu/shm.cpp
|
||||||
- vllm/v1/worker/cpu_worker.py
|
- vllm/v1/worker/cpu_worker.py
|
||||||
- vllm/v1/worker/gpu_worker.py
|
- vllm/v1/worker/gpu_worker.py
|
||||||
@@ -82,10 +82,21 @@ steps:
|
|||||||
- vllm/platforms/cpu.py
|
- vllm/platforms/cpu.py
|
||||||
- vllm/distributed/parallel_state.py
|
- vllm/distributed/parallel_state.py
|
||||||
- vllm/distributed/device_communicators/cpu_communicator.py
|
- vllm/distributed/device_communicators/cpu_communicator.py
|
||||||
|
- .buildkite/scripts/hardware_ci/run-cpu-distributed-smoke-test.sh
|
||||||
commands:
|
commands:
|
||||||
- |
|
- |
|
||||||
bash .buildkite/scripts/hardware_ci/run-cpu-test.sh 10m "
|
bash .buildkite/scripts/hardware_ci/run-cpu-test.sh 10m "
|
||||||
bash .buildkite/scripts/hardware_ci/run-cpu-distributed-smoke-test.sh"
|
bash .buildkite/scripts/hardware_ci/run-cpu-distributed-smoke-test.sh tp_pp"
|
||||||
|
|
||||||
|
- label: CPU-Distributed Tests (DP+TP)
|
||||||
|
depends_on: []
|
||||||
|
device: intel_cpu
|
||||||
|
no_plugin: true
|
||||||
|
source_file_dependencies: *cpu_distributed_deps
|
||||||
|
commands:
|
||||||
|
- |
|
||||||
|
bash .buildkite/scripts/hardware_ci/run-cpu-test.sh 10m "
|
||||||
|
bash .buildkite/scripts/hardware_ci/run-cpu-distributed-smoke-test.sh dp_tp"
|
||||||
|
|
||||||
- label: CPU-Multi-Modal Model Tests %N
|
- label: CPU-Multi-Modal Model Tests %N
|
||||||
depends_on: []
|
depends_on: []
|
||||||
|
|||||||
@@ -3,42 +3,37 @@ set -euox pipefail
|
|||||||
export VLLM_CPU_CI_ENV=0
|
export VLLM_CPU_CI_ENV=0
|
||||||
export VLLM_CPU_KVCACHE_SPACE=1 # avoid OOM
|
export VLLM_CPU_KVCACHE_SPACE=1 # avoid OOM
|
||||||
|
|
||||||
echo "--- PP+TP"
|
MODE=${1:-all}
|
||||||
vllm serve meta-llama/Llama-3.2-3B-Instruct -tp=2 -pp=2 --max-model-len=4096 &
|
|
||||||
server_pid=$!
|
|
||||||
timeout 600 bash -c "until curl localhost:8000/v1/models > /dev/null 2>&1; do sleep 1; done" || exit 1
|
|
||||||
vllm bench serve \
|
|
||||||
--backend vllm \
|
|
||||||
--dataset-name random \
|
|
||||||
--model meta-llama/Llama-3.2-3B-Instruct \
|
|
||||||
--num-prompts 20 \
|
|
||||||
--result-dir ./test_results \
|
|
||||||
--result-filename tp_pp.json \
|
|
||||||
--save-result \
|
|
||||||
--endpoint /v1/completions
|
|
||||||
kill -s SIGTERM $server_pid; wait $server_pid || true
|
|
||||||
failed_req=$(jq '.failed' ./test_results/tp_pp.json)
|
|
||||||
if [ "$failed_req" -ne 0 ]; then
|
|
||||||
echo "Some requests were failed!"
|
|
||||||
exit 1
|
|
||||||
fi
|
|
||||||
|
|
||||||
echo "--- DP+TP"
|
run_scenario() {
|
||||||
vllm serve meta-llama/Llama-3.2-3B-Instruct -tp=2 -dp=2 --max-model-len=4096 &
|
local label="$1" result_file="$2"
|
||||||
server_pid=$!
|
shift 2
|
||||||
timeout 600 bash -c "until curl localhost:8000/v1/models > /dev/null 2>&1; do sleep 1; done" || exit 1
|
echo "--- $label"
|
||||||
vllm bench serve \
|
vllm serve meta-llama/Llama-3.2-3B-Instruct "$@" --max-model-len=4096 &
|
||||||
--backend vllm \
|
local server_pid=$!
|
||||||
--dataset-name random \
|
timeout 600 bash -c "until curl localhost:8000/v1/models > /dev/null 2>&1; do sleep 1; done" || exit 1
|
||||||
--model meta-llama/Llama-3.2-3B-Instruct \
|
vllm bench serve \
|
||||||
--num-prompts 20 \
|
--backend vllm \
|
||||||
--result-dir ./test_results \
|
--dataset-name random \
|
||||||
--result-filename dp_pp.json \
|
--model meta-llama/Llama-3.2-3B-Instruct \
|
||||||
--save-result \
|
--num-prompts 20 \
|
||||||
--endpoint /v1/completions
|
--result-dir ./test_results \
|
||||||
kill -s SIGTERM $server_pid; wait $server_pid || true
|
--result-filename "$result_file" \
|
||||||
failed_req=$(jq '.failed' ./test_results/dp_pp.json)
|
--save-result \
|
||||||
if [ "$failed_req" -ne 0 ]; then
|
--endpoint /v1/completions
|
||||||
echo "Some requests were failed!"
|
kill -s SIGTERM "$server_pid"; wait "$server_pid" || true
|
||||||
exit 1
|
if [ "$(jq '.failed' "./test_results/$result_file")" -ne 0 ]; then
|
||||||
fi
|
echo "Some requests were failed in $label!"
|
||||||
|
exit 1
|
||||||
|
fi
|
||||||
|
}
|
||||||
|
|
||||||
|
case "$MODE" in
|
||||||
|
tp_pp) run_scenario "PP+TP" tp_pp.json -tp=2 -pp=2 ;;
|
||||||
|
dp_tp) run_scenario "DP+TP" dp_tp.json -tp=2 -dp=2 ;;
|
||||||
|
all)
|
||||||
|
run_scenario "PP+TP" tp_pp.json -tp=2 -pp=2
|
||||||
|
run_scenario "DP+TP" dp_tp.json -tp=2 -dp=2
|
||||||
|
;;
|
||||||
|
*) echo "ERROR: unknown mode '$MODE' (expected: tp_pp | dp_tp | all)" >&2; exit 1 ;;
|
||||||
|
esac
|
||||||
|
|||||||
Reference in New Issue
Block a user