Skip to content

Npu use cpu argwhere #10640

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
ShawnXuan wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Npu use cpu argwhere #10640

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3152ef4
update
ShawnXuan May 14, 2025
01044a8
refine
Flowingsun007 May 15, 2025
57b93ba
Merge branch 'master' into npu_use_cpu_argwhere
MagicalMogu May 29, 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
20 changes: 20 additions & 0 deletions oneflow/core/functional/impl/array_functor.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -588,7 +588,27 @@ class ArgWhereFunctor {
const Symbol<DType>& dtype) const {
auto& attrs = THREAD_CACHED_MUTABLE_ATTR_MAP("dtype");
attrs.SetAllAttrs(dtype->data_type());
#ifdef WITH_NPU
auto device_type = DeviceType::kCPU;
if (x->is_global()) {
device_type = JUST(x->parallel_desc())->device_type();
} else {
device_type = JUST(x->device())->enum_type();
}
if (device_type == DeviceType::kNPU) {
// NOTE: use cpu argwhere when device="npu"
auto cpu_tensor = JUST(one::functional::To(x, "cpu"));
auto result = JUST(OpInterpUtil::Dispatch<TensorTuple>(*op_, {cpu_tensor}, attrs));
for (int i = 0; i < result->size(); ++i) {
(*result)[i] = JUST(one::functional::To((*result)[i], "npu"));
}
return result;
} else {
return OpInterpUtil::Dispatch<TensorTuple>(*op_, {x}, attrs);
}
#else
return OpInterpUtil::Dispatch<TensorTuple>(*op_, {x}, attrs);
#endif // WITH_NPU
}

private:
Expand Down
Loading