-
Notifications
You must be signed in to change notification settings - Fork 81
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Compatibility with FillArrays #381
Comments
Could you provide an MWE of what doesn't work, and what you would want to work? |
MWE: using FillArrays, GPUArrays, CUDA, Test
A = CUDA.rand(3, 3)
@test A isa AbstractGPUArray
@test A + Fill(1.2, 3, 3) isa AbstractGPUArray # works as expected
@test_broken A + Diagonal(Fill(1.2, 3)) isa AbstractGPUArray # would expect this to work, too |
I just became aware of JuliaArrays/FillArrays.jl#165, Update: I just tested JuliaArrays/FillArrays.jl#165 but it doesn't affect the MWE above (second test remains broken). |
Sorry for the slow response.
That's because you nest the Fill in a Diagonal; we can't have dispatch for every combination of nested wrappers. See JuliaGPU/Adapt.jl#21, this needs support in Base (e.g. JuliaLang/julia#31563) for better support. |
We are trying to provide GPU support for AbstractGPs.jl. In AbstractGPs we make use of FillArrays.jl, which unfortunately results in CPU Matrix results when combined with e.g. a CuArray. Would be great if GPUArrays.jl could provide support for the FillArray types!
The text was updated successfully, but these errors were encountered: