[PyTorch] Utilize preswizzled inputs in GroupedGemm #2464

Closed

Labels

opened

Enable and optimize the use of preswizzled input tensors in GroupedGemm operations in PyTorch, reducing input transform overhead and maximizing device throughput.

Metadata

Assignees

No one assigned

Labels

Type

No type

Fields

No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests