Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Resolve numpy 1.25 issues #973

Merged
merged 7 commits into from
Jun 23, 2023
Merged

Resolve numpy 1.25 issues #973

Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
bb36aa9
fix bare ._cunumeric usage
bryevdv Jun 21, 2023
1320682
fix deprecated .product
bryevdv Jun 21, 2023
14b03d0
Fix (most) find_common_type instances
bryevdv Jun 21, 2023
30c4b26
pass scalar values, not types
bryevdv Jun 22, 2023
d2eabd8
fix ufunc usage
bryevdv Jun 22, 2023
774f0ea
Merge branch 'branch-23.07' into bv/numpy-1.25-hotfix
bryevdv Jun 22, 2023
7cd8aca
re-enable scalar binop type tests
bryevdv Jun 22, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 2 additions & 20 deletions cunumeric/_ufunc/ufunc.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -557,33 +557,15 @@ def _find_common_type(
if len(unique_dtypes) == 1 and all_ndarray:
return arrs[0].dtype

# FIXME: The following is a miserable attempt to implement type
bryevdv marked this conversation as resolved.
Show resolved Hide resolved
# coercion rules that try to match NumPy's rules for a subset of cases;
# for the others, cuNumeric computes a type different from what
# NumPy produces for the same operands. Type coercion rules shouldn't
# be this difficult to imitate...

all_scalars = all(arr.ndim == 0 for arr in arrs)
all_arrays = all(arr.ndim > 0 for arr in arrs)
kinds = set(arr.dtype.kind for arr in arrs)
lossy_conversion = ("i" in kinds or "u" in kinds) and (
"f" in kinds or "c" in kinds
)
use_min_scalar = not (all_scalars or all_arrays or lossy_conversion)

scalar_types = []
array_types = []
for arr, orig_arg in zip(arrs, orig_args):
if arr.ndim == 0:
scalar_types.append(
np.dtype(np.min_scalar_type(orig_arg))
if use_min_scalar
else arr.dtype
)
scalar_types.append(orig_arg)
else:
array_types.append(arr.dtype)

return np.find_common_type(array_types, scalar_types) # type: ignore
return np.result_type(*array_types, *scalar_types)

def _resolve_dtype(
self,
Expand Down