From 006e7a34aeb3e905ca4131a3251fe079f0511e2f Mon Sep 17 00:00:00 2001 From: Louie Tsai Date: Fri, 5 Sep 2025 05:08:50 -0700 Subject: [PATCH] Adding int4 and int8 models for CPU benchmarking (#23709) Signed-off-by: Tsai, Louie --- .../tests/serving-tests-cpu-snc2.json | 420 +++++++++++- .../tests/serving-tests-cpu-snc3.json | 627 +++++++++++++++++- docs/contributing/benchmarks.md | 32 +- 3 files changed, 1066 insertions(+), 13 deletions(-) diff --git a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json index 2d88a0b30c4f8..f758097e098e4 100644 --- a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json +++ b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json @@ -1,6 +1,6 @@ [ { - "test_name": "serving_llama8B_tp1_sharegpt", + "test_name": "serving_llama8B_bf16_tp1_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -32,7 +32,7 @@ } }, { - "test_name": "serving_llama8B_tp2_sharegpt", + "test_name": "serving_llama8B_bf16_tp2_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -64,7 +64,7 @@ } }, { - "test_name": "serving_llama8B_tp4_sharegpt", + "test_name": "serving_llama8B_bf16_tp4_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -96,7 +96,7 @@ } }, { - "test_name": "serving_llama8B_tp1_random_128_128", + "test_name": "serving_llama8B_bf16_tp1_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -131,7 +131,7 @@ } }, { - "test_name": "serving_llama8B_tp2_random_128_128", + "test_name": "serving_llama8B_bf16_tp2_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -166,7 +166,7 @@ } }, { - "test_name": "serving_llama8B_tp4_random_128_128", + "test_name": "serving_llama8B_bf16_tp4_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -198,5 +198,413 @@ "random-output-len": 128, "num_prompts": 1000 } + }, + { + "test_name": "serving_llama8B_int8_tp1_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_tp2_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_tp4_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 4, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_tp1_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int8_tp2_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int8_tp4_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 4, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_tp1_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_tp2_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_tp4_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 4, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_tp1_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_tp2_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_tp4_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 4, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } } ] diff --git a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json index 823abbaa99f86..ce396d6e54f27 100644 --- a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json +++ b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json @@ -1,6 +1,6 @@ [ { - "test_name": "serving_llama8B_pp1_sharegpt", + "test_name": "serving_llama8B_bf16_pp1_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -32,7 +32,39 @@ } }, { - "test_name": "serving_llama8B_pp3_sharegpt", + "test_name": "serving_llama8B_bf16_tp2_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "meta-llama/Llama-3.1-8B-Instruct", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "meta-llama/Llama-3.1-8B-Instruct", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_bf16_pp3_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -64,7 +96,7 @@ } }, { - "test_name": "serving_llama8B_tp2pp3_sharegpt", + "test_name": "serving_llama8B_bf16_tp2pp3_sharegpt", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], "server_environment_variables": { @@ -97,7 +129,7 @@ } }, { - "test_name": "serving_llama8B_pp1_random_128_128", + "test_name": "serving_llama8B_bf16_pp1_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -132,7 +164,42 @@ } }, { - "test_name": "serving_llama8B_pp3_random_128_128", + "test_name": "serving_llama8B_bf16_tp2_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "meta-llama/Llama-3.1-8B-Instruct", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "meta-llama/Llama-3.1-8B-Instruct", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_bf16_pp3_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -167,7 +234,7 @@ } }, { - "test_name": "serving_llama8B_tp2pp3_random_128_128", + "test_name": "serving_llama8B_bf16_tp2pp3_random_128_128", "qps_list": ["inf"], "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], "server_environment_variables": { @@ -201,5 +268,553 @@ "ignore-eos": "", "num_prompts": 1000 } + }, + { + "test_name": "serving_llama8B_int8_pp1_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "pipeline_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_tp2_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_pp3_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_tp2pp3_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int8_pp1_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "pipeline_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int8_tp2_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int8_pp3_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int8_tp2pp3_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "tensor_parallel_size": 2, + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_pp1_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "pipeline_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_tp2_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_pp3_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_tp2pp3_sharegpt", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "sharegpt", + "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json", + "num_prompts": 200 + } + }, + { + "test_name": "serving_llama8B_int4_pp1_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "pipeline_parallel_size": 1, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_tp2_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_pp3_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } + }, + { + "test_name": "serving_llama8B_int4_tp2pp3_random_128_128", + "qps_list": ["inf"], + "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000], + "server_environment_variables": { + "VLLM_RPC_TIMEOUT": 100000, + "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1, + "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120, + "VLLM_CPU_SGL_KERNEL": 1, + "VLLM_CPU_KVCACHE_SPACE": 40 + }, + "server_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "quantization": "awq", + "tensor_parallel_size": 2, + "pipeline_parallel_size": 3, + "dtype": "bfloat16", + "distributed_executor_backend": "mp", + "block_size": 128, + "trust_remote_code": "", + "enable_chunked_prefill": "", + "disable_log_stats": "", + "enforce_eager": "", + "max_num_batched_tokens": 2048, + "max_num_seqs": 256, + "load_format": "dummy" + }, + "client_parameters": { + "model": "hugging-quants/Meta-Llama-3.1-8B-Instruct-AWQ-INT4", + "backend": "vllm", + "dataset_name": "random", + "random-input-len": 128, + "random-output-len": 128, + "ignore-eos": "", + "num_prompts": 1000 + } } ] diff --git a/docs/contributing/benchmarks.md b/docs/contributing/benchmarks.md index 2bbed778f3c6a..25c2d2955ff2f 100644 --- a/docs/contributing/benchmarks.md +++ b/docs/contributing/benchmarks.md @@ -11,9 +11,39 @@ vLLM contains two sets of benchmarks: The performance benchmarks are used for development to confirm whether new changes improve performance under various workloads. They are triggered on every commit with both the `perf-benchmarks` and `ready` labels, and when a PR is merged into vLLM. +### Manually Trigger the benchmark + +Use [vllm-ci-test-repo images](https://gallery.ecr.aws/q9t5s3a7/vllm-ci-test-repo) with vLLM benchmark suite. +For CPU environment, please use the image with "-cpu" postfix. + +Here is an example for docker run command for CPU. + +```bash +docker run -it --entrypoint /bin/bash -v /data/huggingface:/root/.cache/huggingface -e HF_TOKEN='' --shm-size=16g --name vllm-cpu-ci public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:1da94e673c257373280026f75ceb4effac80e892-cpu +``` + +Then, run below command inside the docker instance. + +```bash +bash .buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh +``` + +When run, benchmark script generates results under **benchmark/results** folder, along with the benchmark_results.md and benchmark_results.json. + +#### Runtime environment variables + +- `ON_CPU`: set the value to '1' on Intel® Xeon® Processors. Default value is 0. +- `SERVING_JSON`: JSON file to use for the serving tests. Default value is empty string (use default file). +- `LATENCY_JSON`: JSON file to use for the latency tests. Default value is empty string (use default file). +- `THROUGHPUT_JSON`: JSON file to use for the throughout tests. Default value is empty string (use default file). +- `REMOTE_HOST`: IP for the remote vLLM service to benchmark. Default value is empty string. +- `REMOTE_PORT`: Port for the remote vLLM service to benchmark. Default value is empty string. + +For more results visualization, check the [visualizing the results](https://github.com/intel-ai-tce/vllm/blob/more_cpu_models/.buildkite/nightly-benchmarks/README.md#visualizing-the-results). + The latest performance results are hosted on the public [vLLM Performance Dashboard](https://hud.pytorch.org/benchmark/llms?repoName=vllm-project%2Fvllm). -More information on the performance benchmarks and their parameters can be found [here](gh-file:.buildkite/nightly-benchmarks/performance-benchmarks-descriptions.md). +More information on the performance benchmarks and their parameters can be found in [Benchmark README](https://github.com/intel-ai-tce/vllm/blob/more_cpu_models/.buildkite/nightly-benchmarks/README.md) and [performance benchmark description](gh-file:.buildkite/nightly-benchmarks/performance-benchmarks-descriptions.md). [](){ #nightly-benchmarks }