Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Benchmark] Fix parameters mismatch #247

Merged
merged 3 commits into from
Jun 7, 2024
Merged

[Benchmark] Fix parameters mismatch #247

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7dfa689
fix parameters mismatch
KepingYan Jun 6, 2024
edf99a1
fix parameters mismatch
KepingYan Jun 6, 2024
f4e7318
disable benchmark ci in pr
KepingYan Jun 7, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 0 additions & 4 deletions .github/workflows/workflow_orders_on_pr.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,3 @@ jobs:
Finetune:
needs: Lint
uses: ./.github/workflows/workflow_finetune.yml

Benchmark:
needs: Lint
uses: ./.github/workflows/workflow_test_benchmark.yml
10 changes: 5 additions & 5 deletions benchmarks/run_benchmark.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,7 +61,7 @@ get_peak_throughpt(){
echo "RUN llm-on-ray with vllm"
echo "RUN bs ${vllm_bs}"
# server:
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_concurrent_queries $VALUE_INF --vllm_max_num_seqs $vllm_bs
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_ongoing_requests $VALUE_INF --max_num_seqs $vllm_bs
# client:
$NUMA_CLIENT_COMMAND python $BENCHMARK_SCRIPT --model-endpoint-base $MODEL_ENDPOINT --model-name $MODEL_NAME --dataset $DATASET_SHAREGPT_PATH --num-prompts $num_prompts --dataset-format ShareGPT --vllm-engine --simple --results-dir $bs_dir_vllm
done
Expand All @@ -80,7 +80,7 @@ metric_bs(){
echo "RUN llm-on-ray with vllm"
echo "RUN bs ${vllm_bs}"
# server:
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_concurrent_queries $VALUE_INF --vllm_max_num_seqs $vllm_bs
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_ongoing_requests $VALUE_INF --max_num_seqs $vllm_bs
# client:
$NUMA_CLIENT_COMMAND python $BENCHMARK_SCRIPT --model-endpoint-base $MODEL_ENDPOINT --model-name $MODEL_NAME --dataset $DATASET_SHAREGPT_PATH --num-prompts $num_prompts --dataset-format ShareGPT --vllm-engine --simple --results-dir $bs_dir_vllm
done
Expand All @@ -90,7 +90,7 @@ metric_bs(){
echo "RUN bs ${wo_vllm_bs}"
bs_dir_wo_vllm=$choice_dir_wo_vllm"/bs_"$wo_vllm_bs
# server:
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WO_VLLM_CONFIG_FILE --simple --max_concurrent_queries $wo_vllm_bs
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WO_VLLM_CONFIG_FILE --simple --max_ongoing_requests $wo_vllm_bs
# client:
$NUMA_CLIENT_COMMAND python $BENCHMARK_SCRIPT --model-endpoint-base $MODEL_ENDPOINT --model-name $MODEL_NAME --dataset $DATASET_SHAREGPT_PATH --num-prompts $num_prompts --dataset-format ShareGPT --simple --results-dir $bs_dir_wo_vllm
done
Expand All @@ -107,7 +107,7 @@ latency_throughput(){
tokens_dir=$choice_dir"/tokens_"$input_tokens_length"_"$output_tokens_length

# server
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_concurrent_queries $VALUE_INF --vllm_max_num_seqs $VALUE_INF
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_ongoing_requests $VALUE_INF --max_num_seqs $VALUE_INF

# client
for i in $(seq 1 $num_iter)
Expand All @@ -133,7 +133,7 @@ get_best_latency(){
choice_dir=${4}

# server
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_concurrent_queries $VALUE_INF --vllm_max_num_seqs $VALUE_INF
$NUMA_SERVER_COMMAND llm_on_ray-serve --config_file $WITH_VLLM_CONFIG_FILE --simple --max_ongoing_requests $VALUE_INF --max_num_seqs $VALUE_INF

# client
for i in $(seq 1 $num_iter)
Expand Down
2 changes: 1 addition & 1 deletion docs/benchmark.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,7 +113,7 @@ cpus_per_worker: 24
```
3. Deploy server
```cmd
OMP_NUM_THREAD=24 numactl -N 0 -m 0 -C 0-47 python -u inference/serve.py --config_file inference/models/vllm/llama-2-7b-chat-hf-vllm.yaml --keep_serve_terminal --max_concurrent_queries 1000 --vllm_max_num_seqs 256 --simple
OMP_NUM_THREAD=24 numactl -N 0 -m 0 -C 0-47 python -u inference/serve.py --config_file inference/models/vllm/llama-2-7b-chat-hf-vllm.yaml --keep_serve_terminal --max_ongoing_requests 1000 --max_num_seqs 256 --simple
```
4. Send requests
```cmd
Expand Down
Loading