Skip to content

Support @view for DestructuredArgs in SymbolicUtils.Code for CuArray Compatibility #698

@chooron

Description

@chooron

Hello, I would like to use the functions generated by SymbolicUtils.Code for CuArray computations, but due to a limitation in CUDA.jl, scalar index operations are not allowed. However, @view can be used. In the functions generated by SymbolicUtils.Code, the DestructuredArgs type is used to accept function arguments and assign values. During this assignment process, the data is directly extracted using indexing rather than with @view, which makes it incompatible with CuArray computations. Therefore, I would like to add an option to extract the values of DestructuredArgs using @view in order to make it work with CuArray.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions