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

[tests] enable testing for xpu #344

Closed
wants to merge 3 commits into from
Closed

[tests] enable testing for xpu #344

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
35fa8e1
add xpu support for testing
faaany Nov 1, 2024
6ca6beb
update version
faaany Nov 11, 2024
76c2817
update int_mm minimal pytorch version
faaany Nov 12, 2024
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
2 changes: 1 addition & 1 deletion optimum/quanto/library/qbytes_mm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,7 +92,7 @@ def qbytes_mm_impl_cuda(activations: torch.Tensor, weights: torch.Tensor, output
def qbytes_mm_impl_cpu(activations: torch.Tensor, weights: torch.Tensor, output_scales: torch.Tensor) -> torch.Tensor:
if (
# FIXME: accuracy issues with 2.4.x
version.parse(torch.__version__).release > version.parse("2.5.0").release
version.parse(torch.__version__).release >= version.parse("2.6.0").release
and activations.dtype == torch.int8
and weights.dtype == torch.int8
):
Expand Down
2 changes: 2 additions & 0 deletions test/conftest.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,8 @@
devices += ["cuda"]
elif torch.backends.mps.is_available():
devices += ["mps"]
elif torch.xpu.is_available():
devices += ["xpu"]


@pytest.fixture(scope="module", params=devices)
Expand Down
3 changes: 3 additions & 0 deletions test/helpers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -103,6 +103,9 @@ def get_device_memory(device):
elif device.type == "mps":
torch.mps.empty_cache()
return torch.mps.current_allocated_memory()
elif device.type == "xpu":
torch.xpu.empty_cache()
return torch.xpu.memory_allocated()
return None


Expand Down
2 changes: 1 addition & 1 deletion test/tensor/weights/weight_helpers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ def check_weight_qtensor_linear(qweight, batch_size, tokens, use_bias, rel_max_e
max_err = (out - qout).abs().max()
rel_max_err = max_err / mean_val
# These values were evaluated empirically without any optimized kernels.
rtol = {"cpu": 1e-2, "cuda": 2e-2, "mps": 1e-2}[device.type]
rtol = {"cpu": 1e-2, "cuda": 2e-2, "mps": 1e-2, "xpu": 2e-2}[device.type]
assert (
rel_max_err < rtol
), f"Maximum error {max_err:.2f} is too high for input of mean value {mean_val:.2f} ({rel_max_err*100:.2f} %)"
Loading