Skip to content

Update comments with MPS issues #43

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

Open
wants to merge 3 commits into
base: gh/manuelcandales/25/base
Choose a base branch
from
Open

Update comments with MPS issues #43

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
ff22572
Update
manuelcandales Aug 18, 2025
18745d1
Update
manuelcandales Aug 18, 2025
49697de
Update
manuelcandales Aug 18, 2025
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
5 changes: 5 additions & 0 deletions test/specdb/test_specdb_mps.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,19 +33,22 @@ class TestSpecDBOperationsMPS(BaseSpecDBTest):
# not implemented
"_pdist_forward.default",
# impl: clamp tensor number of dims must not be greater than that of input tensor
# https://github.com/pytorch/pytorch/issues/160734
"clamp.Tensor",
# crash: https://github.com/pytorch/pytorch/issues/154881
"cumsum.default",
# sparse_grad not supported in MPS yet
"gather.default",
# Dimension specified as -1 but tensor has no dimensions
# https://github.com/pytorch/pytorch/issues/160737
"index_select.default",
# crash: https://github.com/pytorch/pytorch/issues/154882
"max_pool2d_with_indices.default",
# On-going issue on MPSGraph topk when ndims() - axis > 4, see issue #154890
# https://github.com/pytorch/pytorch/issues/154890
"topk.default",
# var_mps: reduction dim must be in the range of input shape
# https://github.com/pytorch/pytorch/issues/160738
"var.correction",
"var.dim",
]
Expand All @@ -59,6 +62,7 @@ def test_all_ops_mps(self):
def test_all_ops_mps_half(self):
skip_ops = self.SKIP_OPS.copy()
# ConvTranspose 3D with BF16 or FP16 types is not supported on MPS
# https://github.com/pytorch/pytorch/issues/160739
skip_ops += ["convolution.default"]

config = TensorConfig(
Expand All @@ -70,6 +74,7 @@ def test_all_ops_mps_transposed(self):
skip_ops = self.SKIP_OPS.copy()
# argmax.default ['ArgType.Tensor torch.float32 (8, 8)', 'ArgType.DimOpt None', 'ArgType.Bool True']
# Exception occurred: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.
# https://github.com/pytorch/pytorch/issues/160740
skip_ops += ["argmax.default", "argmin.default"]

config = TensorConfig(
Expand Down