Skip to content

[BUILD] CUDA 13.3 requires LLVM 22.1.2 or newer #2129

Description

@jfpanisset

This is less of a build "problem", more of a FYI that might be added to the documentation: when trying to build current OSL 1.15.5.0 against latest CUDA 13.3.0, you need to use LLVM 22.1.2 or newer in order to get the following fix:

Artem Belevich: [clang][CUDA] Define _NV_RSQRT_SPECIFIER for glibc-2.42/cuda-13.2 compatibility (#185701) (c1c833734cf0)

as documented in the LLVM release notes:

https://discourse.llvm.org/t/llvm-22-1-2-released/90308

Otherwise OSL code which calls CUDA will fail to compile, complaining about _NV_RSQRT_SPECIFIER being undefined.

I ran across this while working on VFX Platform 2027 container images. The fix does not appear to have been backported to LLVM 21.1.8, but I was able to compile the CUDA sections with LLVM 22.1.7.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    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