from __future__ import annotations

from .base import DeviceBackend


class CudaBackend(DeviceBackend):
    name = "cuda"
    visibility_env_keys = ("CUDA_VISIBLE_DEVICES",)
    torch_namespace = "cuda"
    flaggems_vendor = "nvidia"
    supports_cuda_graph = True
    supports_mm_cluster_fix = True
    supports_triton_hooks = True
    list_command = (("nvidia-smi", "--list-gpus"), "GPU ")
    # Unchanged from when this list lived in _cudagraph_plugin; CUDA numbers do
    # not move because of the classification refactor.
    fatal_error_markers = (
        "illegal memory access",
        "cudaerrorillegaladdress",
        "out of memory",
        "device-side assert triggered",
        "an illegal instruction",
        "misaligned address",
        "uncorrectable ecc",
    )
    # No CUDA wording is known to poison the context without also matching one
    # of the fatal markers above, so nothing is listed here rather than guessed.
    poisoning_error_markers = ()