Skip to content

Version guard for AccumulateGrad::add #1768

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

Closed
wants to merge 1 commit into from
+14 −0
Closed

Version guard for AccumulateGrad::add #1768

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
14 changes: 14 additions & 0 deletions python/monarch/gradient/_gradient_generator.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,12 @@
#include <torch/csrc/autograd/python_function.h> // @manual=//caffe2:torch_extension
#include <torch/extension.h> // @manual=//caffe2:torch_extension

#define TORCH_VERSION_NEWER_THAN(major, minor, patch) \
((TORCH_VERSION_MAJOR > (major)) || \
(TORCH_VERSION_MAJOR == (major) && TORCH_VERSION_MINOR > (minor)) || \
(TORCH_VERSION_MAJOR == (major) && TORCH_VERSION_MINOR == (minor) && \
TORCH_VERSION_PATCH > (patch)))

using torch::autograd::Edge;
using torch::autograd::InputBuffer;
using torch::autograd::Node;
Expand Down Expand Up @@ -420,12 +426,20 @@ struct GradientGenerator {
DEBUG_PRINT(
"// add: " << node->node->name()
<< ", input_nr=" << static_cast<int>(input_nr) << "\n");
#if TORCH_VERSION_NEWER_THAN(2, 8, 0)
Copy link
Contributor

@jithunnair-amd jithunnair-amd Nov 25, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@pzhan9 Shouldn't this condition have been for 2.9.1? Even the release/2.9 branch has a 4-argument version: https://github.com/pytorch/pytorch/blob/d38164a545b4a4e4e0cf73ce67173f70574890b6/torch/csrc/autograd/input_buffer.h#L31

Copy link
Contributor

@jithunnair-amd jithunnair-amd Nov 25, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Addressed by #2000

realInputBuffer(node).add(
input_nr,
check_and_reduce(node->node, input_nr, std::move(t)),
std::nullopt,
std::nullopt,
node->node);
#else
realInputBuffer(node).add(
input_nr,
check_and_reduce(node->node, input_nr, std::move(t)),
std::nullopt,
std::nullopt);
#endif
}

InputBuffer& realInputBuffer(NodeState* state) {
Expand Down