# ==========================================
# ACTUAL TEST CASES
# ==========================================

test_cases:
  - name: "Qwen3.6-35B-A3B-in16k-out1k-0-128-32"
    model: "Eco-Tech/Qwen3.6-35B-A3B-w8a8"
    envs:
      HCCL_OP_EXPANSION_MODE: "AIV"
      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
      OMP_PROC_BIND: "false"
      OMP_NUM_THREADS: "1"
      TASK_QUEUE_ENABLE: "1"
      HCCL_BUFFSIZE: "1024"
      VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
      VLLM_ASCEND_BALANCE_SCHEDULING: "1"
      VLLM_ASCEND_ENABLE_FUSED_MC2: "1"
      SERVER_PORT: "DEFAULT_PORT"
    server_cmd:
      - "--quantization"
      - "ascend"
      - "--allowed-local-media-path"
      - "/"
      - "--trust-remote-code"
      - "--tensor-parallel-size"
      - "2"
      - "--enable-expert-parallel"
      - "--port"
      - "$SERVER_PORT"
      - "--max-num-seqs"
      - "36"
      - "--max-model-len"
      - "20000"
      - "--max-num-batched-tokens"
      - "8192"
      - "--gpu-memory-utilization"
      - "0.9"
      - "--seed"
      - "1024"
      - "--no-enable-prefix-caching"
      - "--async-scheduling"
      - "--compilation-config"
      - '{"cudagraph_capture_sizes":[1,4,8,12,16,20,24,28,40,48,64,76,80,88,96,100,104,108,112,116,120,136,140,144],"cudagraph_mode":"FULL_DECODE_ONLY"}'
      - "--speculative-config"
      - '{"method": "qwen3_5_mtp","num_speculative_tokens": 3}'
      - "--additional-config"
      - '{"enable_cpu_binding":true}'
      - "--profiler-config"
      - '{"profiler": "torch","torch_profiler_dir": "./profiling","torch_profiler_with_stack": false}'
      - "--mm-processor-cache-gb"
      - "0"
      - "--mm-encoder-tp-mode"
      - "data"
    benchmarks:
      perf:
        case_type: performance
        dataset_path: vllm-ascend/GSM8K_prefix0_in16384_bs200_qwen
        request_conf: vllm_api_stream_chat
        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
        num_prompts: 128
        max_out_len: 1024
        batch_size: 32
        request_rate: 0.6
        baseline: 548.2128
        threshold: 0.97
  - name: "Qwen3.6-35B-A3B-in984k-out1k-0-4-2"
    model: "Eco-Tech/Qwen3.6-35B-A3B-w8a8"
    envs:
      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
      HCCL_OP_EXPANSION_MODE: "AIV"
      HCCL_BUFFSIZE: "200"
      OMP_NUM_THREADS: "1"
      TASK_QUEUE_ENABLE: "1"
      VLLM_ASCEND_ENABLE_FUSED_MC2: "1"
      VLLM_ASCEND_GDN_FAST_PATH: "1"
      VLLM_ASCEND_GDN_MAX_PADDING_RATIO: "2.0"
      VLLM_ASCEND_GDN_MAX_H_OVERALLOC_RATIO: "2.0"
      SERVER_PORT: "DEFAULT_PORT"
    server_cmd:
      - "--hf-overrides"
      - '{"text_config":{"rope_parameters":{"rope_type":"yarn","factor":4.0,"original_max_position_embeddings":262144}}}'
      - "--port"
      - "$SERVER_PORT"
      - "--data-parallel-size"
      - "1"
      - "--tensor-parallel-size"
      - "2"
      - "--max-model-len"
      - "1010000"
      - "--max-num-batched-tokens"
      - "16384"
      - "--max-num-seqs"
      - "128"
      - "--gpu-memory-utilization"
      - "0.9"
      - "--no-enable-prefix-caching"
      - "--compilation-config"
      - '{"cudagraph_capture_sizes":[1,4,8,12,16,24,32,48,56,64,72,84,96,108,112,128,160,172,196,200,212,232,256,160,172,196,200,212,232,256,272,288,312,328,344,360,384,400,416,432,448,480,512],"cudagraph_mode":"FULL_DECODE_ONLY"}'
      - "--speculative-config"
      - '{"method": "qwen3_5_mtp", "num_speculative_tokens": 3, "enforce_eager": true}'
      - "--trust-remote-code"
      - "--async-scheduling"
      - "--quantization"
      - "ascend"
      - "--allowed-local-media-path"
      - "/"
      - "--mm-processor-cache-gb"
      - "0"
      - "--additional-config"
      - '{"enable_cpu_binding":true}'
    benchmarks:
      perf:
        case_type: performance
        dataset_path: vllm-ascend/GSM8K_in1007616_bs16_prefix0_qwen
        request_conf: vllm_api_stream_chat
        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
        num_prompts: 4
        max_out_len: 1024
        batch_size: 2
        request_rate: 0
        baseline: 2.2225
        threshold: 0.97
  - name: "Qwen3.6-35B-A3B-in128k-out1k-90-60-15"
    model: "Eco-Tech/Qwen3.6-35B-A3B-w8a8"
    envs:
      PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
      HCCL_OP_EXPANSION_MODE: "AIV"
      OMP_PROC_BIND: "false"
      OMP_NUM_THREADS: "1"
      TASK_QUEUE_ENABLE: "1"
      HCCL_BUFFSIZE: "1024"
      VLLM_ASCEND_ENABLE_FLASHCOMM1: "1"
      VLLM_ASCEND_BALANCE_SCHEDULING: "1"
      VLLM_DISABLE_COMPILE_CACHE: "1"
      VLLM_ASCEND_ENABLE_FUSED_MC2: "1"
      SERVER_PORT: "DEFAULT_PORT"
    server_cmd:
      - "--quantization"
      - "ascend"
      - "--allowed-local-media-path"
      - "/"
      - "--trust-remote-code"
      - "--tensor-parallel-size"
      - "2"
      - "--enable-expert-parallel"
      - "--port"
      - "$SERVER_PORT"
      - "--max-num-seqs"
      - "36"
      - "--max-model-len"
      - "140000"
      - "--max-num-batched-tokens"
      - "16384"
      - "--gpu-memory-utilization"
      - "0.9"
      - "--seed"
      - "1024"
      - "--enable-prefix-caching"
      - "--async-scheduling"
      - "--compilation-config"
      - '{"cudagraph_capture_sizes":[1,4,8,12,16,20,24,28,40,48,64,76,80,88,96,100,104,108,112,116,120,136,140,144],"cudagraph_mode":"FULL_DECODE_ONLY"}'
      - "--speculative-config"
      - '{"method": "qwen3_5_mtp","num_speculative_tokens": 3}'
      - "--additional-config"
      - '{"enable_cpu_binding":true}'
      - "--profiler-config"
      - '{"profiler": "torch","torch_profiler_dir": "./profiling","torch_profiler_with_stack": false}'
      - "--mm-processor-cache-gb"
      - "0"
      - "--mm-encoder-tp-mode"
      - "data"
    benchmarks:
      perf:
        case_type: performance
        dataset_path: vllm-ascend/GSM8K_prefix90_in131072_bs100_qwen
        request_conf: vllm_api_stream_chat
        dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_str_perf
        num_prompts: 60
        max_out_len: 1024
        batch_size: 15
        request_rate: 0
        baseline: 223.2484
        threshold: 0.97