4c4a8458a5
### What this PR does / why we need it? Refactor the multi-machine CI use case. The purpose of this PR is to increase the ease of adding multi-machine CI use cases, allowing developers to add multi-machine cluster model testing use cases (including PD separation) by simply adding a new YAML configuration file. ### Does this PR introduce _any_ user-facing change? ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: wangli <wangli858794774@gmail.com>
77 lines
2.0 KiB
YAML
77 lines
2.0 KiB
YAML
test_name: "test Qwen3-235B-A22B multi-dp"
|
|
model: "Qwen/Qwen3-235B-A22B"
|
|
num_nodes: 2
|
|
npu_per_node: 16
|
|
env_common:
|
|
VLLM_USE_MODELSCOPE: true
|
|
OMP_PROC_BIND: false
|
|
OMP_NUM_THREADS: 100
|
|
HCCL_BUFFSIZE: 1024
|
|
SERVER_PORT: 8080
|
|
|
|
deployment:
|
|
-
|
|
local_index: 0
|
|
master_index: 0
|
|
headless: false
|
|
env_extend:
|
|
server_cmd: >
|
|
vllm serve "Qwen/Qwen3-235B-A22B"
|
|
--host 0.0.0.0
|
|
--port $SERVER_PORT
|
|
--data-parallel-size 4
|
|
--data-parallel-size-local 2
|
|
--data-parallel-address $LOCAL_IP
|
|
--data-parallel-rpc-port 13389
|
|
--tensor-parallel-size 8
|
|
--seed 1024
|
|
--enable-expert-parallel
|
|
--max-num-seqs 16
|
|
--max-model-len 8192
|
|
--max-num-batched-tokens 8192
|
|
--trust-remote-code
|
|
--no-enable-prefix-caching
|
|
--gpu-memory-utilization 0.9
|
|
-
|
|
local_index: 1
|
|
master_index: 0
|
|
headless: true
|
|
env_extend:
|
|
server_cmd: >
|
|
vllm serve "Qwen/Qwen3-235B-A22B"
|
|
--headless
|
|
--data-parallel-size 4
|
|
--data-parallel-size-local 2
|
|
--data-parallel-start-rank 2
|
|
--data-parallel-address $MASTER_IP
|
|
--data-parallel-rpc-port 13389
|
|
--tensor-parallel-size 8
|
|
--seed 1024
|
|
--max-num-seqs 16
|
|
--max-model-len 8192
|
|
--max-num-batched-tokens 8192
|
|
--enable-expert-parallel
|
|
--trust-remote-code
|
|
--no-enable-prefix-caching
|
|
--gpu-memory-utilization 0.9
|
|
benchmarks:
|
|
perf:
|
|
case_type: performance
|
|
dataset_path: vllm-ascend/GSM8K-in3500-bs400
|
|
request_conf: vllm_api_stream_chat
|
|
dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
|
|
num_prompts: 1
|
|
max_out_len: 2
|
|
batch_size: 1
|
|
baseline: 5
|
|
threshold: 0.97
|
|
acc:
|
|
case_type: accuracy
|
|
dataset_path: vllm-ascend/AIME2024
|
|
request_conf: vllm_api_general_chat
|
|
dataset_conf: aime2024/aime2024_gen_0_shot_chat_prompt
|
|
max_out_len: 10
|
|
batch_size: 32
|
|
baseline: 1
|
|
threshold: 1
|