cupy
cupy copied to clipboard
NumPy & SciPy for GPU
https://github.com/cupy/cupy/pull/8457#issuecomment-2656568499
### Description The `cp.inner` function raises a `ZeroDivisionError` when called with two empty arrays, while `np.inner` returns` 0.0`. From a mathematical perspective, the inner product of two empty vectors should...
### Description In `cuml` we have a need for an API similar to `numpy.logaddexp.reduce`. For now I'm getting by with a custom `cupy.ReductionKernel`, but it would be nice if this...
e.g., Use `cusolverDnXgetrf` instead of `cusolverDnDgetrf` to support 64-bit indexing.
### Description I want to be able to use thread block cluster sizes of 16 on Hopper cards with CuPy, but as far as I can tell this isn't currently...
### Description Hello, I'd like to request the addition of a batched convolve2d function - one that takes two lists of arrays (or perhaps implemented as two 3D arrays) and...
In a license review for packaging cupy for fedora here https://bugzilla.redhat.com/show_bug.cgi?id=2395004 An issue was found that the some LICENSE files list in this repo should also be pypi source. Specifically...
### Description `cupy.histogramdd` will trigger `ValueError: zero-size array to reduction operation CUPY_CUB_MAX which has no identity`, when using an empty input array. ### To Reproduce ```py import numpy as np...
### Description `cupyx.scipy.signal.gauss_spline` returns incorrect result when input `x` is an integer (e.g., `[0]`), returning `[0]` instead of the expected Gaussian value `~0.7979`. When the input is a float (e.g.,...
### Description When a 1-dimensional array is passed as the `sos` parameter to `cupyx.scipy.signal.sosfilt`, the function does not validate the input shape and instead produces **inconsistent, incorrect, and non-reproducible outputs**....