Skip to content

[GPU] Asr fixes #25215

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
p-durandin merged 9 commits into from
Jul 1, 2024
+15 −10
Merged

[GPU] Asr fixes #25215

Show file tree
Hide file tree
Changes from 7 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
7de4c8f
fix for loop op with int32 parameter
michal-miotk Jun 24, 2024
cfb685f
compile by comment
michal-miotk Jun 24, 2024
b4d646c
Revert "compile by comment"
michal-miotk Jun 25, 2024
0b17b67
moved get_output_layouts to avoid stackoverflow
michal-miotk Jun 25, 2024
1948b38
Merge branch 'master' into asr_fixes
p-durandin Jun 26, 2024
571ff56
changed input type in test
michal-miotk Jun 26, 2024
a5fe1c0
Merge branch 'master' into asr_fixes
michal-miotk Jun 27, 2024
2618ca0
fix coding style
michal-miotk Jun 27, 2024
213e3a9
fixed types in test
michal-miotk Jun 28, 2024
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: 4 additions & 5 deletions src/plugins/intel_gpu/src/graph/program.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -517,7 +517,10 @@ void program::init_graph() {
apply_opt_pass<graph_initializations>();

apply_opt_pass<mark_nodes>();

for (auto& node : processing_order) {
if (!node->is_type<data>())
node->get_output_layouts();
}
// Perform initial shape_of subgraphs markup
apply_opt_pass<mark_shape_of_subgraphs>();
}
Expand All @@ -533,10 +536,6 @@ void program::pre_optimize_graph(bool is_internal) {
processing_order.calculate_BFS_processing_order(); // this method makes sense only for OOOQ (out of order execution queue)

bool output_size_handling_enabled = analyze_output_size_handling_need();
for (auto& node : processing_order) {
if (!node->is_type<data>())
node->get_output_layouts();
}

bool optimize_data = _config.get_property(ov::intel_gpu::optimize_data);
if (optimize_data) {
Expand Down
10 changes: 8 additions & 2 deletions src/plugins/intel_gpu/src/plugin/ops/loop.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,8 +30,14 @@ namespace intel_gpu {
template<class DATA_TYPE>
static DATA_TYPE CreateScalarData(ProgramBuilder &p, const cldnn::primitive_id& id, ov::Shape& shape, cldnn::data_types dtype, int64_t num, int64_t rank) {
auto mem = p.get_engine().allocate_memory({ shape, dtype, cldnn::format::get_default_format(rank) });
cldnn::mem_lock<int64_t> ptr{mem, p.get_engine().get_service_stream()};
*ptr.begin() = num;
if ( dtype == cldnn::data_types::i32 ) {
cldnn::mem_lock<int32_t> ptr{mem, p.get_engine().get_service_stream()};
*ptr.begin() = static_cast<int32_t>(num);
} else {
cldnn::mem_lock<int64_t> ptr{mem, p.get_engine().get_service_stream()};
*ptr.begin() = num;
}

return {id, mem};
}

Expand Down
4 changes: 2 additions & 2 deletions src/plugins/intel_gpu/tests/unit/passes/test_module_fusing_reorder.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -297,9 +297,9 @@ TEST_P(test_can_fuse_reorder_onednn, reorder_for_firstconv_onednn)

INSTANTIATE_TEST_SUITE_P(testing_can_fuse_reorder_first_conv, test_can_fuse_reorder_onednn,
::testing::ValuesIn(std::vector<reorder_test_param>{
reorder_test_param{format::bs_fs_yx_bsv8_fsv4, format::b_fs_yx_fsv32, data_types::f32, data_types::u8, {1, 3, 8, 8}, {1, 32, 8, 8}, {1, 3, 1, 1},
reorder_test_param{format::b_fs_yx_fsv16, format::b_fs_yx_fsv32, data_types::f32, data_types::u8, {1, 3, 8, 8}, {1, 32, 8, 8}, {1, 3, 1, 1},
tensor{1}, tensor{0}, data_types::u8, format::goiyx, true},
reorder_test_param{format::bs_fs_yx_bsv8_fsv2, format::b_fs_yx_fsv16, data_types::f32, data_types::f16, {1, 3, 8, 8}, {1, 32, 8, 8}, {1, 3, 1, 1},
reorder_test_param{format::b_fs_yx_fsv4, format::b_fs_yx_fsv16, data_types::f32, data_types::f16, {1, 3, 8, 8}, {1, 32, 8, 8}, {1, 3, 1, 1},
tensor{1}, tensor{0}, data_types::f16, format::goiyx, true},
}));

Expand Down