happyz
/
llama.cpp
mirror of https://github.com/ggerganov/llama.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

mulmat type conversion update

This commit is contained in:
Cavus Mustafa 2025-07-29 18:17:14 -07:00 committed by Mustafa Cavus
parent e2fdc1b988
commit 93b2d09a2d
1 changed files with 9 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
ggml/src/ggml-openvino/openvino/op/mulmat.cpp
View File

@ -30,10 +30,13 @@ OutputVector translate_mulmat(const NodeContext& context) {
ov::Output<Node> res; ov::Output<Node> res;
ov::Output<ov::Node> B = context.get_input(0); ov::Output<ov::Node> B = context.get_input(0);
ov::Output<ov::Node> A = context.get_input(1); ov::Output<ov::Node> A = context.get_input(1);
bool convert_out_type = false;
if (ov::op::util::is_constant(B.get_node()) && context.get_input_type(0) != context.get_input_type(1)) { if (ov::op::util::is_constant(B.get_node()) && context.get_input_type(0) != context.get_input_type(1)) {
B = std::make_shared<ov::op::v0::Convert>(context.get_input(0), context.get_input_type(1)); B = std::make_shared<ov::op::v0::Convert>(context.get_input(0), context.get_input_type(1));
} else if (context.get_input_type(0) != context.get_input_type(1)) { } else if (context.get_input_type(0) != context.get_input_type(1)) {
A = std::make_shared<ov::op::v0::Convert>(context.get_input(1), context.get_input_type(0)); A = std::make_shared<ov::op::v0::Convert>(context.get_input(1), context.get_input_type(0));
convert_out_type = true;
} }
auto B_shape = context.get_input_shape(0).to_shape(); auto B_shape = context.get_input_shape(0).to_shape();
@ -68,7 +71,12 @@ OutputVector translate_mulmat(const NodeContext& context) {
A = Z; A = Z;
} }
res = std::make_shared<ov::op::v0::MatMul>(A, B, false, true); if (convert_out_type) {
auto result_lp = std::make_shared<ov::op::v0::MatMul>(A, B, false, true);
res = std::make_shared<ov::op::v0::Convert>(result_lp, context.get_output_type(0));
} else {
res = std::make_shared<ov::op::v0::MatMul>(A, B, false, true);
}
return rename_outputs_with_suffix({res}, context.get_name()); return rename_outputs_with_suffix({res}, context.get_name());
} }