Skip to content
This repository was archived by the owner on Feb 24, 2026. It is now read-only.
This repository was archived by the owner on Feb 24, 2026. It is now read-only.

Potential Bitblas compat with cuda 12.5 #120

@Qubitium

Description

@Qubitium

While doing ci testing for GPTQModel we found that with cuda 12.5, bitblas is generating broken compiled codes via apache/tvm. There are no errors. The end result is no runtime error but cache model gpu code (in .cached folder) generates non-sense using backend=BACKEND.BITBLAS: failed our PPL sanity test. We copied over .cache/generated files from < cuda 12.5 and it works which isolates the issue to cuda 12.5/apache tvm.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions