# ==========================================
# Shared Configurations
# ==========================================

_envs: &envs
  OMP_NUM_THREADS: "10"
  OMP_PROC_BIND: "false"
  HCCL_BUFFSIZE: "1024"
  PYTORCH_NPU_ALLOC_CONF: "expandable_segments:True"
  SERVER_PORT: "DEFAULT_PORT"

_server_cmd: &server_cmd
  - "--quantization"
  - "ascend"
  - "--data-parallel-size"
  - "2"
  - "--tensor-parallel-size"
  - "8"
  - "--enable-expert-parallel"
  - "--port"
  - "$SERVER_PORT"
  - "--seed"
  - "1024"
  - "--max-model-len"
  - "36864"
  - "--max-num-batched-tokens"
  - "4096"
  - "--max-num-seqs"
  - "16"
  - "--trust-remote-code"
  - "--gpu-memory-utilization"
  - "0.9"
  - "--speculative-config"
  - '{"num_speculative_tokens": 1, "method": "mtp"}'
  - "--additional-config"
  - '{"enable_weight_nz_layout": true}'

_benchmarks_acc: &benchmarks_acc
  acc:
    case_type: accuracy
    dataset_path: vllm-ascend/gsm8k-lite
    request_conf: vllm_api_general_chat
    dataset_conf: gsm8k/gsm8k_gen_0_shot_cot_chat_prompt
    max_out_len: 32768
    batch_size: 32
    baseline: 95
    threshold: 5

_benchmarks_perf: &benchmarks_perf
  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: 400
    max_out_len: 1500
    batch_size: 1000
    baseline: 1
    threshold: 0.97

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

test_cases:
  - name: "DeepSeek-R1-0528-W8A8-single"
    model: "vllm-ascend/DeepSeek-R1-0528-W8A8"
    envs:
      <<: *envs
    server_cmd: *server_cmd
    server_cmd_extra:
      - "--enforce-eager"
    benchmarks:

  - name: "DeepSeek-R1-0528-W8A8-aclgraph"
    model: "vllm-ascend/DeepSeek-R1-0528-W8A8"
    envs:
      <<: *envs
    server_cmd: *server_cmd
    benchmarks:
      <<: *benchmarks_acc
      <<: *benchmarks_perf

  - name: "DeepSeek-R1-0528-W8A8-EPLB"
    model: "vllm-ascend/DeepSeek-R1-0528-W8A8"
    envs:
      <<: *envs
      DYNAMIC_EPLB: "true"
    server_cmd: *server_cmd
    server_cmd_extra:
      - "--additional-config"
      - '{"enable_weight_nz_layout": true, "eplb_config": {"dynamic_eplb": "true", "expert_heat_collection_interval": 1000, "algorithm_execution_interval": 50, "eplb_policy_type": 3}}'
    benchmarks:
      <<: *benchmarks_acc