# SuperBench Config
#
# Server:
#   - Product: HPE Apollo 6500

version: v0.7
superbench:
  enable: null
  var:
    default_local_mode: &default_local_mode
      enable: true
      modes:
        - name: local
          proc_num: 8
          prefix: HIP_VISIBLE_DEVICES={proc_rank}
          parallel: yes
    default_pytorch_mode: &default_pytorch_mode
      enable: true
      modes:
        - name: torch.distributed
          proc_num: 8
          node_num: 1
      frameworks:
        - pytorch
    common_model_config: &common_model_config
      duration: 0
      num_warmup: 64
      num_steps: 2048
      sample_count: 8192
      batch_size: 32
      precision:
        - float32
        - float16
      model_action:
        - train
      pin_memory: yes
  benchmarks:
    kernel-launch:
      <<: *default_local_mode
    rccl-bw:
      enable: true
      modes:
        - name: local
          proc_num: 1
          parallel: no
      parameters:
        maxbytes: 8G
        ngpus: 8
        operation: allreduce
    cpu-memory-bw-latency:
      enable: false
      modes:
        - name: local
          proc_num: 1
          parallel: no
      parameters:
        tests:
          - bandwidth_matrix
          - latency_matrix
          - max_bandwidth
    mem-bw:
      enable: true
      modes:
        - name: local
          proc_num: 8
          prefix: HIP_VISIBLE_DEVICES={proc_rank} numactl -N $(({proc_rank}/2))
          parallel: no
    gemm-flops:
      <<: *default_local_mode
      parameters:
        m: 7680
        n: 8192
        k: 8192
    ib-loopback:
      enable: true
      modes:
        - name: local
          proc_num: 4
          prefix: PROC_RANK={proc_rank} IB_DEVICES=0,1,2,3
          parallel: no
    disk-benchmark:
      enable: false
      modes:
        - name: local
          proc_num: 1
          parallel: no
      parameters:
        block_devices: []
    gpu-copy-bw:correctness:
      enable: true
      modes:
        - name: local
          parallel: no
      parameters:
        mem_type:
          - htod
          - dtoh
          - dtod
        copy_type:
          - sm
          - dma
        size: 4096
        num_warm_up: 0
        num_loops: 1
        check_data: true
    gpu-copy-bw:perf:
      enable: true
      modes:
        - name: local
          parallel: no
      parameters:
        mem_type:
          - htod
          - dtoh
          - dtod
        copy_type:
          - sm
          - dma
    ib-traffic:
      enable: false
      modes:
        - name: mpi
          proc_num: 1
          mca:
            btl: tcp,self
            pml: ob1
            btl_tcp_if_include: ens17f0
    gpcnet-network-test:
      enable: false
      modes:
        - name: mpi
          proc_num: 1
          mca:
            pml: ucx
            btl: ^uct
            btl_tcp_if_include: ens17f0
    tcp-connectivity:
      enable: false
      modes:
        - name: local
          parallel: no
      parameters:
        port: 22
    ort-models:
      enable: false
      modes:
        - name: local
      frameworks:
        - onnxruntime
    gpt_models:
      <<: *default_pytorch_mode
      models:
        - gpt2-small
        - gpt2-large
      parameters:
        <<: *common_model_config
        batch_size: 8
        seq_len: 224
    bert_models:
      <<: *default_pytorch_mode
      models:
        - bert-base
        - bert-large
      parameters:
        <<: *common_model_config
        seq_len: 224
    lstm_models:
      <<: *default_pytorch_mode
      models:
        - lstm
      parameters:
        <<: *common_model_config
        batch_size: 224
        input_size: 224
        hidden_size: 1000
        seq_len: 32
        pin_memory: no
    resnet_models:
      <<: *default_pytorch_mode
      models:
        - resnet50
        - resnet101
        - resnet152
      parameters:
        <<: *common_model_config
        pin_memory: no
    densenet_models:
      <<: *default_pytorch_mode
      models:
        - densenet169
        - densenet201
      parameters:
        <<: *common_model_config
        pin_memory: no
    vgg_models:
      <<: *default_pytorch_mode
      models:
        - vgg11
        - vgg13
        - vgg16
        - vgg19
      parameters:
        <<: *common_model_config
        pin_memory: no