CUDA Tester

Mix.install(
  [
    {:nx, "~> 0.6.1"},
    {:exla, "~> 0.6.1"}
  ],
  config: [
    nx: [
      default_backend: EXLA.Backend,
      default_defn_options: [compiler: EXLA]
    ],
    exla: [
      default_client: :cuda,
      clients: [
        host: [platform: :host],
        cuda: [platform: :cuda]
      ]
    ]
  ],
  system_env: [
    XLA_TARGET: "cuda120"
  ]
)

Section

The next segment should contain output about my GPU ID, cuDNN, and allocating memory, and the tensor should show EXLA.Backend.

Nx.with_default_backend({EXLA.Backend, client: :cuda}, fn ->
  Nx.iota({10, 10})
  |> Nx.add(10)
end)

08:05:24.476 [info] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355

08:05:24.478 [info] XLA service 0x7f09f844a8b0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:

08:05:24.478 [info]   StreamExecutor device (0): NVIDIA GeForce RTX 3060, Compute Capability 8.6

08:05:24.478 [info] Using BFC allocator.

08:05:24.478 [info] XLA backend allocating 11366557286 bytes on device 0 for BFCAllocator.

08:05:24.535 [info] Loaded cuDNN version 8901

08:05:24.581 [info] Using nvlink for parallel linking

#Nx.Tensor<
  s64[10][10]
  EXLA.Backend
  [
    [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
    [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
    [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
    [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
    [50, 51, 52, 53, 54, 55, 56, 57, 58, 59],
    ...
  ]
>