pykokkos icon indicating copy to clipboard operation
pykokkos copied to clipboard

Performance portable parallel programming in Python.

Results 90 pykokkos issues
Sort by recently updated
recently updated
newest added

Full test suite result vs. Kokkos `develop` branch (OpenMP only): ` = 66 failed, 31 passed, 9 skipped, 9 xfailed, 2 warnings in 148.22s (0:02:28) ==` Most of the failures...

bug

I've finally managed to avoid various CUDA driver/runtime errors, but now when I run `pytest` I see all tests fail in a manner similar to this: ``` ____________________________________________________________________________________________________________________________________________________ TestViews.test_v3d ____________________________________________________________________________________________________________________________________________________...

I was able to install the pykokkos, but when I run the 'runtests.py' lot of tests are failing. I am adding the terminal output as an attachment. [test_report.txt](https://github.com/user-attachments/files/15934272/test_report.txt)

I am running into an issue that when I am calling a pk.parallel_for loop sequentially on a gpu more than 5 time in a row, the code hangs. Here is...

This checks if torch is installed and uses `torch.is_tensor()` to check if the passed argument is a pytorch tensor.

When I try to pass a PyTorch tensor into a workunit as I would a CuPy array, as in the following script, ```python import torch import pykokkos as pk @pk.workunit...

In a `pk.workunit`, my code fails when I try to access a cupy array as `x[a, b]` instead of `x[a][b]`. For example, the script ```python import cupy as cp import...

This adds [TeamThreadMDRange](https://kokkos.org/kokkos-core-wiki/API/core/policies/TeamThreadMDRange.html?highlight=teamthreadmdrange#_CPPv4I00E17TeamThreadMDRange) from Kokkos

Our current implementation does not support float32 scalars for `add`, `multiply`, and `divide`. This PR introduces a simple fix in the logic to allow computation with `float32` scalars. Currently Pykokkos...

Kokkos supports calling multiple levels of scratch memory as in the code below ```cpp policy.set_scratch_size(0, PerTeam(bytes0)) .set_scratch_size(1, PerTeam(bytes1)); ``` but pykokkos throws an error when I call ```python pk.TeamPolicy(league_size, team_size).set_scratch_size(0,...