Closed drewrobson closed 1 year ago
Base: 87.74% // Head: 87.78% // Increases project coverage by +0.03%
:tada:
Coverage data is based on head (
f21ad6b
) compared to base (f432a7b
). Patch coverage: 100.00% of modified lines in pull request are covered.
:umbrella: View full report at Codecov.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
Thanks. Changing a mutable struct to immutable might be a breaking change though. I'm not sure why it is mutable to begin with.
This will require a minor version change at least.
I ran into an error when trying to use
FilledExtrapolation
on the GPU. Turns outsrc/gpu_support.jl
was missing support forFilledExtrapolation
. This PR fixes it. For example, the following now works:using Interpolations, CUDA, Adapt A = rand(Float32, 10) eitp = extrapolate(interpolate(A, BSpline(Linear())), 0.0f0) d_eitp = adapt(CuArray{Float32}, eitp) @assert eitp.(-5:5) == Array(d_eitp.(-5:5))
Hello, I tried this code on Julia 1.8.5 on Linux Ubuntu 22.04 with a Nvidia GPU but it does not work and raises this error.
ERROR: MethodError: objects of type CuArray{Float32, 1, CUDA.Mem.DeviceBuffer} are not callable Use square brackets [] for indexing an Array.
I think the Interpolations.extrapolation
on the GPU is not working properly.
Have you tried the master branch of this package? This code has not been released yet.
@mkitti Thanks for your answer.
Yes, I have tried the master branch. Do I have to try this (https://github.com/drewrobson/Interpolations.jl) branch?
I tried Drew Robson's branch (https://github.com/drewrobson/Interpolations.jl), and it ran successfully.
Is it possible to merge his branch with the master branch? @mkitti
Thanks.
This pull request is merged. Can you test the master branch here?
I see we're in a conversational loop. I'm confused why the master branch here does not work since the changes should be in https://github.com/JuliaMath/Interpolations.jl/commit/85d3ccd515c397b4f37778d8826cc028cf13133c
How did you try the master branch here?
Thanks for your answer.
I tried the master branch of the Interpolations package and it is running now. (https://github.com/JuliaMath/Interpolations.jl)
It is interesting that if I try to install Interpolations.jl
from the Julia package registry (] add Interpolations
), it will not work. (I do not know why)
] add Interpolations
will install the latest released version, which was six months ago. ] add Interpolations#master
will grab the development branch with the commit of interest. So perhaps your actual request is for us to make a new release of Interpolations.jl.
@mkitti Thank you so much for the answer. Sorry, I was at the conference last week.
Yes, the ] add Interpolations#master
is working. Have you considered releasing the new version of the Interpolations
in the near future?
I ran into an error when trying to use
FilledExtrapolation
on the GPU. Turns outsrc/gpu_support.jl
was missing support forFilledExtrapolation
. This PR fixes it. For example, the following now works: