[CI][BugFix] ShellCheck cleanup to remove baseline and preserve runtime behavior (#34514)
Signed-off-by: junuxyz <216036880+junuxyz@users.noreply.github.com>
This commit is contained in:
@@ -46,10 +46,10 @@ echo "VLLM_LOGGING_LEVEL=$VLLM_LOGGING_LEVEL"
|
||||
echo "RESULT_FILE=$RESULT"
|
||||
echo "====================== AUTO TUNEPARAMETERS ===================="
|
||||
|
||||
rm -rf $LOG_FOLDER
|
||||
rm -rf $PROFILE_PATH
|
||||
mkdir -p $LOG_FOLDER
|
||||
mkdir -p $PROFILE_PATH
|
||||
rm -rf "$LOG_FOLDER"
|
||||
rm -rf "$PROFILE_PATH"
|
||||
mkdir -p "$LOG_FOLDER"
|
||||
mkdir -p "$PROFILE_PATH"
|
||||
|
||||
cd "$BASE/vllm"
|
||||
|
||||
@@ -114,7 +114,7 @@ start_server() {
|
||||
|
||||
# wait for 10 minutes...
|
||||
server_started=0
|
||||
for i in {1..60}; do
|
||||
for _ in {1..60}; do
|
||||
# This line checks whether the server is still alive or not,
|
||||
# since that we should always have permission to send signal to the server process.
|
||||
kill -0 $server_pid 2> /dev/null || break
|
||||
@@ -145,12 +145,12 @@ run_benchmark() {
|
||||
local vllm_log="$LOG_FOLDER/vllm_log_${max_num_seqs}_${max_num_batched_tokens}.txt"
|
||||
echo "vllm_log: $vllm_log"
|
||||
echo
|
||||
rm -f $vllm_log
|
||||
rm -f "$vllm_log"
|
||||
pkill -if "vllm serve" || true
|
||||
|
||||
echo "starting server..."
|
||||
# Call start_server without a profile_dir to avoid profiling overhead
|
||||
start_server $gpu_memory_utilization $max_num_seqs $max_num_batched_tokens $vllm_log ""
|
||||
start_server "$gpu_memory_utilization" "$max_num_seqs" "$max_num_batched_tokens" "$vllm_log" ""
|
||||
result=$?
|
||||
if [[ "$result" -eq 1 ]]; then
|
||||
echo "server failed to start. gpu_memory_utilization:$gpu_memory_utilization, max_num_seqs:$max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens"
|
||||
@@ -168,15 +168,15 @@ run_benchmark() {
|
||||
# --profile flag is removed from this call
|
||||
vllm bench serve \
|
||||
--backend vllm \
|
||||
--model $MODEL \
|
||||
--model "$MODEL" \
|
||||
--dataset-name random \
|
||||
--random-input-len $adjusted_input_len \
|
||||
--random-output-len $OUTPUT_LEN \
|
||||
--random-output-len "$OUTPUT_LEN" \
|
||||
--ignore-eos \
|
||||
--disable-tqdm \
|
||||
--request-rate inf \
|
||||
--percentile-metrics ttft,tpot,itl,e2el \
|
||||
--goodput e2el:$MAX_LATENCY_ALLOWED_MS \
|
||||
--goodput e2el:"$MAX_LATENCY_ALLOWED_MS" \
|
||||
--num-prompts 1000 \
|
||||
--random-prefix-len $prefix_len \
|
||||
--host "$HOSTNAME" \
|
||||
@@ -195,20 +195,20 @@ run_benchmark() {
|
||||
request_rate=$((${throughput%.*} + 1))
|
||||
while ((request_rate > 0)); do
|
||||
# clear prefix cache
|
||||
curl -X POST http://${HOSTNAME}:8004/reset_prefix_cache
|
||||
curl -X POST http://"${HOSTNAME}":8004/reset_prefix_cache
|
||||
sleep 5
|
||||
bm_log="$LOG_FOLDER/bm_log_${max_num_seqs}_${max_num_batched_tokens}_requestrate_${request_rate}.txt"
|
||||
vllm bench serve \
|
||||
--backend vllm \
|
||||
--model $MODEL \
|
||||
--model "$MODEL" \
|
||||
--dataset-name random \
|
||||
--random-input-len $adjusted_input_len \
|
||||
--random-output-len $OUTPUT_LEN \
|
||||
--random-output-len "$OUTPUT_LEN" \
|
||||
--ignore-eos \
|
||||
--disable-tqdm \
|
||||
--request-rate $request_rate \
|
||||
--percentile-metrics ttft,tpot,itl,e2el \
|
||||
--goodput e2el:$MAX_LATENCY_ALLOWED_MS \
|
||||
--goodput e2el:"$MAX_LATENCY_ALLOWED_MS" \
|
||||
--num-prompts 100 \
|
||||
--random-prefix-len $prefix_len \
|
||||
--host "$HOSTNAME" \
|
||||
@@ -255,7 +255,7 @@ gpu_memory_utilization=0.98
|
||||
find_gpu_memory_utilization=0
|
||||
while (( $(echo "$gpu_memory_utilization >= 0.9" | bc -l) )); do
|
||||
# Pass empty string for profile_dir argument
|
||||
start_server $gpu_memory_utilization "${num_seqs_list[-1]}" "${num_batched_tokens_list[-1]}" "$LOG_FOLDER/vllm_log_gpu_memory_utilization_$gpu_memory_utilization.log" ""
|
||||
start_server "$gpu_memory_utilization" "${num_seqs_list[-1]}" "${num_batched_tokens_list[-1]}" "$LOG_FOLDER/vllm_log_gpu_memory_utilization_$gpu_memory_utilization.log" ""
|
||||
result=$?
|
||||
if [[ "$result" -eq 0 ]]; then
|
||||
find_gpu_memory_utilization=1
|
||||
@@ -274,7 +274,7 @@ fi
|
||||
|
||||
for num_seqs in "${num_seqs_list[@]}"; do
|
||||
for num_batched_tokens in "${num_batched_tokens_list[@]}"; do
|
||||
run_benchmark $num_seqs $num_batched_tokens $gpu_memory_utilization
|
||||
run_benchmark "$num_seqs" "$num_batched_tokens" "$gpu_memory_utilization"
|
||||
done
|
||||
done
|
||||
echo "finish permutations"
|
||||
@@ -285,7 +285,7 @@ echo "finish permutations"
|
||||
if (( $(echo "$best_throughput > 0" | bc -l) )); then
|
||||
echo
|
||||
echo "Benchmark tuning finished. Now running profiling on the best configuration found..."
|
||||
echo "Best config: max_num_seqs: $best_max_num_seqs, max_num_batched_tokens: $best_num_batched_tokens, throughput: $best_throughput"
|
||||
echo "Best config: max_num_seqs: $best_max_num_seqs, max_num_batched_tokens: $best_num_batched_tokens, throughput: $best_throughput, goodput: $best_goodput"
|
||||
echo
|
||||
|
||||
vllm_log="$LOG_FOLDER/vllm_log_BEST_PROFILE.txt"
|
||||
@@ -293,7 +293,7 @@ if (( $(echo "$best_throughput > 0" | bc -l) )); then
|
||||
|
||||
# Start server with the best params and profiling ENABLED
|
||||
echo "Starting server for profiling..."
|
||||
start_server $gpu_memory_utilization $best_max_num_seqs $best_num_batched_tokens "$vllm_log" "$PROFILE_PATH"
|
||||
start_server "$gpu_memory_utilization" "$best_max_num_seqs" "$best_num_batched_tokens" "$vllm_log" "$PROFILE_PATH"
|
||||
|
||||
# Run benchmark with the best params and the --profile flag
|
||||
echo "Running benchmark with profiling..."
|
||||
@@ -301,15 +301,15 @@ if (( $(echo "$best_throughput > 0" | bc -l) )); then
|
||||
adjusted_input_len=$(( INPUT_LEN - prefix_len ))
|
||||
vllm bench serve \
|
||||
--backend vllm \
|
||||
--model $MODEL \
|
||||
--model "$MODEL" \
|
||||
--dataset-name random \
|
||||
--random-input-len $adjusted_input_len \
|
||||
--random-output-len $OUTPUT_LEN \
|
||||
--random-output-len "$OUTPUT_LEN" \
|
||||
--ignore-eos \
|
||||
--disable-tqdm \
|
||||
--request-rate $best_request_rate \
|
||||
--request-rate "$best_request_rate" \
|
||||
--percentile-metrics ttft,tpot,itl,e2el \
|
||||
--goodput e2el:$MAX_LATENCY_ALLOWED_MS \
|
||||
--goodput e2el:"$MAX_LATENCY_ALLOWED_MS" \
|
||||
--num-prompts 100 \
|
||||
--random-prefix-len $prefix_len \
|
||||
--host "$HOSTNAME" \
|
||||
|
||||
@@ -64,7 +64,7 @@ for i in $(seq 0 $(($num_runs - 1))); do
|
||||
else
|
||||
STATUS="FAILURE"
|
||||
((FAILURE_COUNT++))
|
||||
FAILED_RUNS+=("Run #$((i+1)): $(echo $run_object | jq -c .)")
|
||||
FAILED_RUNS+=("Run #$((i+1)): $(echo "$run_object" | jq -c .)")
|
||||
fi
|
||||
|
||||
RUN_OUTPUT=$(<"$RUN_OUTPUT_FILE")
|
||||
|
||||
@@ -71,7 +71,7 @@ while [[ $# -gt 0 ]]; do
|
||||
usage
|
||||
;;
|
||||
*)
|
||||
echo "Unknown argument: $1\n"
|
||||
printf "Unknown argument: %s\n" "$1"
|
||||
usage
|
||||
;;
|
||||
esac
|
||||
@@ -84,15 +84,17 @@ mkdir -p "$OUTPUT_DIR"
|
||||
QPS_VALUES=(25 20 15 10 5 1)
|
||||
|
||||
# Common parameters
|
||||
COMMON_PARAMS="--backend $BACKEND \
|
||||
--model $MODEL \
|
||||
--dataset $DATASET \
|
||||
--structured-output-ratio $STRUCTURED_OUTPUT_RATIO \
|
||||
--save-results \
|
||||
--result-dir $OUTPUT_DIR \
|
||||
--output-len $MAX_NEW_TOKENS \
|
||||
--port $PORT \
|
||||
--tokenizer-mode $TOKENIZER_MODE"
|
||||
COMMON_PARAMS=(
|
||||
--backend "$BACKEND"
|
||||
--model "$MODEL"
|
||||
--dataset "$DATASET"
|
||||
--structured-output-ratio "$STRUCTURED_OUTPUT_RATIO"
|
||||
--save-results
|
||||
--result-dir "$OUTPUT_DIR"
|
||||
--output-len "$MAX_NEW_TOKENS"
|
||||
--port "$PORT"
|
||||
--tokenizer-mode "$TOKENIZER_MODE"
|
||||
)
|
||||
|
||||
echo "Starting structured output benchmark with model: $MODEL"
|
||||
echo "Backend: $BACKEND"
|
||||
@@ -109,17 +111,17 @@ for qps in "${QPS_VALUES[@]}"; do
|
||||
GIT_BRANCH=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo "unknown")
|
||||
|
||||
# Construct filename for this run
|
||||
FILENAME="${BACKEND}_${qps}qps_$(basename $MODEL)_${DATASET}_${GIT_HASH}.json"
|
||||
FILENAME="${BACKEND}_${qps}qps_$(basename "$MODEL")_${DATASET}_${GIT_HASH}_${GIT_BRANCH}.json"
|
||||
|
||||
NUM_PROMPTS=$(echo "$TOTAL_SECONDS * $qps" | bc)
|
||||
NUM_PROMPTS=${NUM_PROMPTS%.*} # Remove fractional part
|
||||
echo "Running benchmark with $NUM_PROMPTS prompts"
|
||||
|
||||
# Run the benchmark
|
||||
python "$SCRIPT_DIR/benchmark_serving_structured_output.py" $COMMON_PARAMS \
|
||||
--request-rate $qps \
|
||||
python "$SCRIPT_DIR/benchmark_serving_structured_output.py" "${COMMON_PARAMS[@]}" \
|
||||
--request-rate "$qps" \
|
||||
--result-filename "$FILENAME" \
|
||||
--num-prompts $NUM_PROMPTS
|
||||
--num-prompts "$NUM_PROMPTS"
|
||||
|
||||
echo "Completed benchmark with QPS: $qps"
|
||||
echo "----------------------------------------"
|
||||
|
||||
Reference in New Issue
Block a user