Gregory Lee
Gregory Lee
> Is there anything left to do on this one? Or is it just waiting for a new CuPy release with the referenced fixes? I still need to make one...
sorry, I forgot that I opened #746 before going on vacation. Everything on our end should be addressed then!
Thanks for making the Cupy 13.3.0 dev packages @seberg. I tried in a Python 3.10 environment with Numpy 2.0.1 and all tests were passing. I did not use any `env.yaml`...
I found a single test failure on cuCIM with current CuPy 13.3 dev and NumPy 2.0.1. It occurs in kernel fusion code and a simplified pure CuPy reproducer is this:...
We can pass a device scalar for the constant in cuCIM to work around the issue in that single kernel, but I wanted to report it here for completeness.
Hi @Ian990466, Was your issue also related to symbolic links as discussed here? https://github.com/rapidsai/cucim/issues/454#issuecomment-3001600887 We unfortunately have not had time yet to add a windows wheel for pip
/ok to test 0de560e
Thanks @martinResearch, is the primary goal of this to avoid having to compile/cache a new kernel for each shape? I agree that this change will do that at cost of...