[environment]
gcp_gpu_attached = true
gcp_image_family = "tracel-ci-ubuntu-2404-amd64-nvidia"
# https://cloud.google.com/compute/docs/accelerator-optimized-machines
# put the faster machine on first place for possibly faster 'Benchmarks Started' feedback in PRs
gcp_machine_types = [
  "a2-highgpu-1g", # 1 A100 40GB (listed as a2 standard)
  "g2-standard-4", # 1 L4 24GB
]
# define the available zones for each machine type
# be sure to check what machine types are available in each region
# https://cloud.google.com/compute/docs/gpus/gpu-regions-zones#view-using-table
gcp_zones = [
  # a2-highgpu-1g
  [
  "asia-northeast1-a",
  "asia-northeast1-c",
  "asia-northeast3-b",
  "asia-southeast1-b",
  "asia-southeast1-c",
  "europe-west4-a",
  "europe-west4-b",
  "us-central1-a",
  "us-central1-b",
  "us-central1-c",
  "us-central1-f",
  "us-east1-b",
  "us-west1-b",
  "us-west3-b",
  "us-west4-b"
  ],
  # g2-standard-4
  [
  "northamerica-northeast2-a",
  "northamerica-northeast2-b",
  "us-central1-a",
  "us-central1-b",
  "us-central1-c",
  "us-east1-b",
  "us-east1-c",
  "us-east1-d",
  "us-east4-a",
  "us-east4-c",
  "us-west1-a",
  "us-west1-b",
  "us-west1-c",
  "us-west4-a",
  "us-west4-c"
  ],
]
repo_full = "tracel-ai/burn"
rust_toolchain = "stable"
rust_version = "stable"

[burn-bench]
github_organization = "tracel-ai"
github_repository = "burn-bench"
github_branch = "main"
github_workflow = "benchmarks.yml"
# vulkan autotune seems to take ages, disabling it for now
# backends = ["cuda-fusion", "vulkan-fusion", "wgpu-fusion"]
backends = ["cuda-fusion", "cuda"]
benches = ["autodiff",
  "binary",
  "bool_select",
  "conv-transpose2d",
  "conv-transpose3d",
  "conv2d",
  "conv3d",
  "custom-gelu",
  "data",
  "load-record",
  "matmul-fused",
  "matmul",
  "max-pool2d",
  "random",
  "reduce",
  "softmax",
  "transformer-encoder",
  "unary"
]
dtypes = ["f16"]