API名称 |
是否支持 |
限制与说明 |
|---|---|---|
torch.backends.cuda.is_built |
否 |
|
torch.backends.cuda.matmul.allow_tf32 |
否 |
|
torch.backends.cuda.matmul.allow_fp16_reduced_precision_reduction |
否 |
|
torch.backends.cuda.clear |
否 |
|
torch.backends.cuda.preferred_linalg_library |
否 |
|
torch.backends.cuda.SDPBackend |
否 |
|
torch.backends.cuda.flash_sdp_enabled |
否 |
|
torch.backends.cuda.enable_mem_efficient_sdp |
否 |
|
torch.backends.cuda.mem_efficient_sdp_enabled |
否 |
|
torch.backends.cuda.enable_flash_sdp |
否 |
|
torch.backends.cuda.math_sdp_enabled |
否 |
|
torch.backends.cuda.enable_math_sdp |
否 |
|
torch.backends.cuda.sdp_kernel |
否 |
|
torch.backends.cudnn.version |
否 |
|
torch.backends.cudnn.is_available |
否 |
|
torch.backends.cudnn.enabled |
否 |
|
torch.backends.cudnn.allow_tf32 |
否 |
|
torch.backends.cudnn.deterministic |
否 |
|
torch.backends.cudnn.benchmark |
否 |
|
torch.backends.mps.is_available |
否 |
|
torch.backends.mps.is_built |
否 |
|
torch.backends.mkl.is_available |
否 |
|
torch.backends.mkl.verbose |
否 |
|
torch.backends.mkldnn.is_available |
否 |
|
torch.backends.mkldnn.verbose |
否 |
|
torch.backends.openmp.is_available |
否 |
|
torch.backends.opt_einsum.is_available |
否 |
|
torch.backends.opt_einsum.get_opt_einsum |
否 |