Skip to content

fix: check full device (incl. index) in XPU fallback consistency check #3713

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
laifenxiawucha wants to merge 2 commits into
base: main
Choose a base branch
from
+5 −6
Open

fix: check full device (incl. index) in XPU fallback consistency check #3713

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9eb8964
fix: check full device (incl. index) in XPU fallback consistency check
laifenxiawucha May 18, 2026
4cb4e3e
fix: remove stale skipCPUIf import in sparse CSR XPU test
Copilot May 19, 2026
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
9 changes: 5 additions & 4 deletions src/ATen/native/xpu/XPUFallback.template
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,20 +6,21 @@ namespace at {
static bool DEBUG_XPU_FALLBACK = false;

/*
* Verifies that all input tensors reside on the same device type.
* Verifies that all input tensors reside on the same device
* (including device index, e.g., xpu:0 vs xpu:1).
* The first defined tensor establishes the reference device;
* all subsequent tensors must match it.
* Raises an error if a mismatch (e.g., CPU vs. XPU) is detected.
* Raises an error if a mismatch (e.g., CPU vs. XPU, or xpu:0 vs. xpu:1) is detected.
*/
static void check_device_consistency(
const torch::jit::Stack* stack,
const c10::OperatorHandle& op) {
std::optional<c10::DeviceType> reference_device;
std::optional<c10::Device> reference_device;

auto check_tensor = [&](const at::Tensor& t) {
if (!t.defined()) return;

auto cur_device = t.device().type();
auto cur_device = t.device();
if (!reference_device) {
reference_device = cur_device;
return;
Expand Down
2 changes: 0 additions & 2 deletions test/xpu/test_sparse_csr_xpu.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,6 @@
OpDTypes,
ops,
precisionOverride,
skipCPUIf,
skipCPUIfNoMklSparse,
skipCUDAIf,
skipCUDAIfNoSparseGeneric,
Expand Down Expand Up @@ -1301,7 +1300,6 @@ def values(lst, device=device):
subtest("sparse_compressed_tensor_no_size"),
],
)
@skipCPUIf(True, "https://github.com/intel/torch-xpu-ops/issues/2229")
def test_invalid_input(self, layout, device, target):
for (
label,
Expand Down
Loading