# ==========================================
# ACTUAL TEST CASES
# ==========================================
test_cases:
  - name: "Qwen3.5-397B-A17B-w8a8-3.5k-1.5k-0-50-single"
    model: "Eco-Tech/Qwen3.5-397B-A17B-w8a8-mtp"
    envs:
      VLLM_USE_MODELSCOPE: "true"
      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
      HCCL_BUFFSIZE: "1024"
      OMP_NUM_THREADS: "1"
      TASK_QUEUE_ENABLE: "1"
      VLLM_ASCEND_ENABLE_FUSED_MC2: "0"
      VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
      HCCL_OP_EXPANSION_MODE: "AIV"
      SERVER_PORT: "DEFAULT_PORT"
    server_cmd:
      - "--port"
      - "$SERVER_PORT"
      - --data-parallel-size
      - "1"
      - --tensor-parallel-size
      - "16"
      - --enable-expert-parallel
      - --max-model-len
      - "6144"
      - --max-num-batched-tokens
      - "16384"
      - --max-num-seqs
      - "128"
      - --gpu-memory-utilization
      - "0.9"
      - --compilation-config
      - '{"cudagraph_capture_sizes":[1,4,8,12,16,24,32,48,56,64,72,84,96,108,112,128,160,172,196,200,212,232,256,272,288,312,328,344,360,384,400,416,432,448,480,512], "cudagraph_mode":"FULL_DECODE_ONLY"}'
      - --speculative_config
      - '{"method": "qwen3_5_mtp", "num_speculative_tokens": 3, "enforce_eager": true}'
      - --trust-remote-code
      - --async-scheduling
      - --allowed-local-media-path
      - "/"
      - --quantization
      - "ascend"
      - --mm-processor-cache-gb
      - "0"
      - --additional-config
      - '{"enable_cpu_binding":true}'

    benchmarks:
      perf_50:
        case_type: performance
        dataset_path: vllm-ascend/GSM8K-in3500-bs2800
        request_conf: vllm_api_stream_chat
        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
        num_prompts: 640
        max_out_len: 1500
        batch_size: 160
        request_rate: 5
        baseline: 2663.3728
        threshold: 0.97
      perf_20:
        case_type: performance
        dataset_path: vllm-ascend/GSM8K-in3500-bs2800
        request_conf: vllm_api_stream_chat
        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
        num_prompts: 128
        max_out_len: 1500
        batch_size: 32
        request_rate: 0
        baseline: 1421.89
        threshold: 0.97