-
Notifications
You must be signed in to change notification settings - Fork 33
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
GPU CI Pipeline #645
base: jl/fourier_cuda
Are you sure you want to change the base?
GPU CI Pipeline #645
Conversation
Ok, the pipeline works. It could be merged. @jackleland This sets up the GPU CI. The PR points towards your branch, feel free to merge it into your branch at any time you like. The tests are running, but there's also a failed test already in there. |
Tests failures look like ComplexF64[-0.3076805147278332 + 0.0im; 1.1085994123834553 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.0016467605892351855 + 0.0im; 1.0662793043766845 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.05971729033643114 + 0.0im; -0.04644323104982592 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; … ;;; 0.3453888679777678 + 0.0im; -0.1789364090484484 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; -0.023388671420225438 + 0.0im; 0.04181378235779623 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.09117647818657071 + 0.0im; 0.06556419506787445 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im]
≈
ComplexF64[-0.3076805507070919 + 0.0im; 1.1085994632838272 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.0016466971488262594 + 0.0im; 1.066279369053504 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.05971722919486555 + 0.0im; -0.0464432261222856 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; … ;;; 0.345388861730916 + 0.0im; -0.1789363932662944 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; -0.023388667412709563 + 0.0im; 0.04181376509700334 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im;;; 0.0911764715403362 + 0.0im; 0.0655641721511797 + 0.0im; … ; 0.0 + 0.0im; 0.0 + 0.0im] So might be just something about error tolerances. Most tests that involve a transform require a |
Can we fix the failing tests on buildkite sooner? Buildkite is already active and makes other pull requests fail ... see #579 for example |
Ah huh, I didn't expect it to be active there. But the failure in #579 isn't the test above failing, it's checking if there is a a pipeline there. I'll make another PR with a dummy pipeline.yml file. |
Fixed the tolerance on the tests now, still looking into an issue with the CPU FFTW transform though |
This merges into
jl/fourier_cuda
This is currently a "draft", but marked as a regular PR to hopefully trigger the pipeline.