Files
enginex-ascend-910-vllm/tests/e2e/weekly/single_node/configs/Qwen3.5-122B-A10B-W8A8-A2.yaml
Sun Ruoxi 7f8a1b1f7a init v0.23.0
Signed-off-by: Sun Ruoxi <sunruoxi@4paradigm.com>
2026-08-27 15:11:51 +08:00

75 lines
2.3 KiB
YAML

# ==========================================
# Shared Configurations
# ==========================================
#workflow暂缺A2yaml
# - name: Qwen3.5-122B-A10B-W8A8-single-A2
# os: linux-aarch64-a2b3-8
# config_file_path: Qwen3.5-122B-A10B-W8A8-A2.yaml
_envs: &envs
TASK_QUEUE_ENABLE: "1"
HCCL_OP_EXPANSION_MODE: "AIV"
VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
SERVER_PORT: "DEFAULT_PORT"
PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
HCCL_BUFFSIZE: "1024"
OMP_NUM_THREADS: "1"
VLLM_ASCEND_ENABLE_FUSED_MC2: "0"
_server_cmd: &server_cmd
- "--quantization"
- "ascend"
- "--no-enable-prefix-caching"
- "--tensor-parallel-size"
- "8"
- "--port"
- "$SERVER_PORT"
- "--max-model-len"
- "262144"
- "--max-num-batched-tokens"
- "16384"
- "--max-num-seqs"
- "128"
- "--trust-remote-code"
- "--gpu-memory-utilization"
- "0.9"
- "--allowed-local-media-path"
- "/"
- "--mm-processor-cache-gb"
- "0"
- "--data-parallel-size"
- "1"
- "--additional-config"
- '{"enable_cpu_binding":true}'
- "--enable-expert-parallel"
- "--profiler-config"
- '{"profiler": "torch", "torch_profiler_dir": "./profiling", "torch_profiler_with_stack": false}'
- "--compilation-config"
- '{"cudagraph_capture_sizes":[1,4,8,12,16,20,24,28,32,36,40,44,48,52,56,60,64,68,72,76,80,84,88,92,96,100,104,108,112,116,120,124,128,132,136,140,144,148,152,156,160,164,168,172,176,180,184,188,192,196,200,204,208,212,216,220,224,228,232,236,240,244,248,252,256,260,264,268,272,276,280,284,286,292,296,300,304,308,312,316,320,324,328,332,336,340], "cudagraph_mode":"FULL_DECODE_ONLY"}'
- "--speculative_config"
- '{"method": "qwen3_5_mtp", "num_speculative_tokens": 3, "enforce_eager": true}'
_benchmarks: &benchmarks
perf:
case_type: performance
dataset_path: vllm-ascend/GSM8K-in3500-bs8000-qwen3
request_conf: vllm_api_stream_chat
dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
num_prompts: 256
max_out_len: 1500
batch_size: 64
baseline: 1
threshold: 0.97
# ==========================================
# ACTUAL TEST CASES
# ==========================================
test_cases:
- name: "Qwen3.5-122B-A10B-W8A8-single-A2"
model: "Eco-Tech/Qwen3.5-122B-A10B-w8a8-mtp"
envs:
<<: *envs
server_cmd: *server_cmd
benchmarks:
<<: *benchmarks