Graham Markall

It seems that clang transformed the inline asm, so I will need to try something more like this: ```asm .Ltmp0: nop .section .note.stapsdt,"",@note .p2align 2 .long .Ltmp1-.Ltmp2 .long .Ltmp3-.Ltmp4 .long...

The root cause here is that NumPy uses `calloc` to allocate memory for `np.zeros`, and therefore doesn't need to zero-initialize the memory - pages will be actually allocated and zero-initialized...

@esc @stuartarchibald It seems like a summary of this issue is: - The pip-installed wheel puts libsvml somewhere that Numba doesn't expect it - The workarounds include patching Numba to...

I'll add this to the agenda for the next triage meeting after the break. A couple of thoughts I wanted to mention there: - Is NumPy using Python's allocator? I...

I've removed the question label, so it shouldn't go stale again (and I think is not purely a question at this point).

gpuci run tests

@oren-hecht It looks like there has been no follow-up from the original submitter, so it's not planned at the moment. I think it could continue if someone else were to...

Thanks for the report. I can reproduce the behaviour you mention but there is too much complexity here to look into the issue, and it needs reducing - it could...

Thanks for the report. I can reproduce similar behaviour with: ```python from numba import njit, vectorize import math import numpy as np np.seterr(all='raise') @njit def finite(x): return math.isfinite(x) finite(1.0) finite(math.nan)...

The implementation if `is_finite` (in numba/cpython/mathimpl.py) might be setting an error flag: ``` 56 def is_finite(builder, val): 57 """ 58 Return a condition testing whether *val* is a finite. 59...