Skip to content

Add condition for Matmul and Select in ClampFP16OutputSoftmaxMatcher #31798

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

Merged
merged 2 commits into from
Aug 20, 2025
Merged

Add condition for Matmul and Select in ClampFP16OutputSoftmaxMatcher #31798

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
b6f37e5
Add condition for Matmul and Select in ClampFP16OutputSoftmaxMatcher
byungilm Aug 19, 2025
25945b6
Merge branch 'master' into fix_inf_out_matmul
byungilm Aug 20, 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
7 changes: 5 additions & 2 deletions src/plugins/intel_gpu/src/plugin/transformations/clamp_fp16_output.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@
#include "openvino/op/multiply.hpp"
#include "openvino/op/subtract.hpp"
#include "openvino/op/divide.hpp"
#include "openvino/op/select.hpp"
#include "openvino/core/rt_info.hpp"
#include "openvino/pass/pattern/op/pattern.hpp"
#include "openvino/pass/pattern/op/wrap_type.hpp"
Expand All @@ -38,10 +39,12 @@ ClampFP16OutputSoftmaxMatcher::ClampFP16OutputSoftmaxMatcher() {
auto in1 = any_input(class_other_than<v0::Constant>());
auto matmul_m = wrap_type<v0::MatMul>({in0, in1}, type_matches(ov::element::f16) && consumers_count(1));
auto reshape_m = wrap_type<v1::Reshape>({matmul_m, any_input()}, type_matches(ov::element::f16) && consumers_count(1));
auto add_m = wrap_type<v1::Add>({matmul_m, any_input()}, type_matches(ov::element::f16) && consumers_count(1));
auto eltwise_m = wrap_type<v1::Divide, v1::Add, v1::Multiply, v1::Subtract>({matmul_m, any_input()},
type_matches(ov::element::f16) && consumers_count(1));
auto softmax_input_m = std::make_shared<Or>(ov::OutputVector{eltwise_m, reshape_m, matmul_m});
auto input_m = std::make_shared<Or>(ov::OutputVector{matmul_m, eltwise_m, reshape_m});
auto select_then_m = wrap_type<v1::Select>({any_input(), input_m, any_input()}, type_matches(ov::element::f16) && consumers_count(1));
auto select_else_m = wrap_type<v1::Select>({any_input(), any_input(), input_m}, type_matches(ov::element::f16) && consumers_count(1));
auto softmax_input_m = std::make_shared<Or>(ov::OutputVector{select_then_m, select_else_m, input_m});
auto softmax_m = wrap_type<v8::Softmax>({softmax_input_m}, type_matches(ov::element::f16));

ov::matcher_pass_callback callback = [OV_CAPTURE_CPY_AND_THIS](ov::pass::pattern::Matcher& m) {
Expand Down
31 changes: 31 additions & 0 deletions src/plugins/intel_gpu/tests/unit/transformations/clamp_fp16_output_test.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,7 @@
#include "openvino/op/maximum.hpp"
#include "openvino/op/matmul.hpp"
#include "openvino/op/softmax.hpp"
#include "openvino/op/select.hpp"
#include <plugin/transformations/clamp_fp16_output.hpp>
#include <transformations/init_node_info.hpp>
#include <transformations/utils/utils.hpp>
Expand Down Expand Up @@ -159,6 +160,36 @@ TEST_F(TransformationTestsF, ClampFp16OutputTest6) {
comparator.enable(FunctionsComparator::CmpValues::ATTRIBUTES);
}

TEST_F(TransformationTestsF, ClampFp16OutputTest7) {
{
auto input1 = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 3, 2, 2 });
auto input2 = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 1, 2, 2 });
auto matmul = std::make_shared<ov::op::v0::MatMul>(input1, input2, true, false);
auto data = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 3, 2, 2 });
auto cond = std::make_shared<ov::op::v0::Parameter>(ov::element::boolean, ov::Shape{ 3, 2, 2 });
auto select = std::make_shared<ov::op::v1::Select>(cond, data, matmul);
auto softmax = std::make_shared<ov::op::v8::Softmax>(select, 1);

model = std::make_shared<ov::Model>(ov::OutputVector{softmax}, ov::ParameterVector{input1, input2, data, cond});
manager.register_pass<ClampFP16Output>();
}
{
auto input1 = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 3, 2, 2 });
auto input2 = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 1, 2, 2 });
auto matmul = std::make_shared<ov::op::v0::MatMul>(input1, input2, true, false);
auto data = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::Shape{ 3, 2, 2 });
auto cond = std::make_shared<ov::op::v0::Parameter>(ov::element::boolean, ov::Shape{ 3, 2, 2 });
auto select = std::make_shared<ov::op::v1::Select>(cond, data, matmul);
auto min = static_cast<double>(std::numeric_limits<ov::float16>::lowest());
auto max = static_cast<double>(std::numeric_limits<ov::float16>::max());
auto clamp = std::make_shared<ov::op::v0::Clamp>(select, min, max);
auto softmax = std::make_shared<ov::op::v8::Softmax>(clamp, 1);

model_ref = std::make_shared<ov::Model>(ov::OutputVector{softmax}, ov::ParameterVector{input1, input2, data, cond});
}
comparator.enable(FunctionsComparator::CmpValues::ATTRIBUTES);
}

TEST_F(TransformationTestsF, ClampFp16OutputRMS) {
{
auto input1 = std::make_shared<ov::op::v0::Parameter>(ov::element::f16, ov::PartialShape{ -1, -1, 2560 });
Expand Down
Loading