[MPS] Enable angle and atan2 for torch.long
#149017
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/149017
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 30 PendingAs of commit f20efa3 with merge base 6057641 ( UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
c2c5deb to
d248dd8
Compare
torch.long
| @@ -397,7 +397,7 @@ static void add_sub_lerp_template(const Tensor& self, | |||
| }); | |||
| } | |||
| TORCH_IMPL_FUNC(atan2_out_mps)(const Tensor& self, const Tensor& other, const Tensor& output) { | |||
| TORCH_CHECK(self.scalar_type() != ScalarType::Long, "MPS does not support atan2 op with int64 input"); | |||
| // TORCH_CHECK(self.scalar_type() != ScalarType::Long, "MPS does not support atan2 op with int64 input"); | |||
There was a problem hiding this comment.
Instead of commenting this out, can you just remove the line?
| @@ -272,7 +272,7 @@ static void unary_op(const Tensor& self, | |||
| } | |||
|
|
|||
| Tensor& angle_out_mps(const Tensor& self, Tensor& output) { | |||
| TORCH_CHECK(self.scalar_type() != ScalarType::Long, "MPS does not support angle op with int64 input"); | |||
| // TORCH_CHECK(self.scalar_type() != ScalarType::Long, "MPS does not support angle op with int64 input"); | |||
There was a problem hiding this comment.
I kept them around, thinking that may be something will fail on MacOS-13, but it did not, so now I can just remove this check
test/test_mps.py
Outdated
| @@ -710,8 +710,8 @@ def mps_ops_modifier(ops): | |||
| 'index_add': [torch.int64], | |||
| 'log1p': [torch.int64], | |||
| 'sigmoid': [torch.int64], | |||
| 'atan2': [torch.int64], | |||
| 'angle': [torch.int64], | |||
| # 'atan2': [torch.int64], | |||
May be those ops were unsupported on MacOS-12, but they work fine in MacOS-15
d248dd8 to
f20efa3
Compare
|
@pytorchbot merge -f "Lint + MPS are green" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
This check was added by #85817, that introduced no unit-tests and its content seems to be totally unrelated to title/subject of that PR. Anyway, right now it seems to be working fine on MacOS-13+