Ianna Osborne

Results 156 comments of Ianna Osborne

fix: prod cuda test

fixed in https://github.com/scikit-hep/awkward/pull/3235

Discrepancy in behavior of `ak.any()` for GPU backend

@kmohrman - thanks for looking into it and the report! Floating-point arithmetic isn’t perfectly precise. Even the same computation on CPU and GPU can differ in the last few bits...

Discrepancy in behavior of `ak.any()` for GPU backend

> The differences are pretty "high" though no? Even for floating point arithmetic. I see differences in the 2nd or 3rd decimal point. @ikrommyd - Are you looking at `np.isclose`?

Discrepancy in behavior of `ak.any()` for GPU backend

Hi @kmohrman . It looks like the issue is not in `ak.any`, but in `ak.combinations`. I'm not sure yet if it's a bug or the operation is not deterministic across...

Discrepancy in behavior of `ak.any()` for GPU backend

Thanks @fstrug! Your help with debugging is much appreciated!

`IndexError` in `ak.combinations()` with `cuda` backend

@kmohrman - thanks for reporting it! Indeed, scalars are one dimensional arrays in `cupy` (unlike in `numpy`).

`IndexError` in `ak.combinations()` with `cuda` backend

@kmohrman - Thanks for reporting it! It looks like this test should have raised a "not implemented error": the `cuda` combinations kernel supports only `n == 2` for now. I...

Issues with `min()` and `argmin()` with `cuda` backend

Thanks, @Moelf ! Could you please make a PR? Thanks!

feat: combinations kernels

> > Then it's ready to merge. Thanks! > > Once the macOS issue is solved I'll merge it. Thanks! I think, you can merge it now - MacOS is...

feat: combinations kernels

> The jobs that haven't succeeded haven't even started. I just re-triggered it, and they're still not starting. > > They're asking for MacOS 11, and I think that was...