IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[IE CLDNN] Fix reshape for yxfb layout (#1632)

Browse Source 
In one of the network it was the following pipeline:
```
FullyConnected -> Reshape -> FullyConnected
```
And the output of Reshape wasn't in the same order as input for this
layer. I found that the problem was connected with format of the layers.
During optimization passes this pipeline was transformed to the
following:
```
FullyConnected -> Reorder -> Reshape -> Reorder -> FullyConnected
```
Both `FullyConnected` layers works with `yxfb` format.  This is why
Reorder layer after the Reshape has output layout with format `yxfb` and
`reshape_in_layout.format` returns `yxfb` format. But in this case we
have to convert Reshape to `bfyx` format because in this case we won't
change the order of elements.
I replaced `reshape_in_layout.format` (which returns `yxfb`) and
explicitly set `bfyx` format.

JIRA: 35288
This commit is contained in:
Egor Churaev
2020-08-11 14:52:04 +03:00
committed by GitHub
parent 129376f609
commit 2caca604ca
2 changed files with 8 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
inference-engine/thirdparty/clDNN/src/graph_optimizer/handle_reshape.cpp vendored
View File

@@ -125,10 +125,16 @@ void handle_reshape::run(program_impl& p) {
for (const auto& reorder_node : reorder_node_to_split) {
auto& reorder_reshape_node = reorder_reshape_nodes[reshape_reorder_id];
auto reshape_in_layout = reorder_node->get_output_layout();
auto dims = cldnn::format::dimension(reshape_in_layout.format);
auto format = cldnn::format::bfyx;
if (dims == 5)
format = cldnn::format::bfzyx;
else if (dims == 6)
format = cldnn::format::bfwzyx;
auto reshape_input = std::make_shared<reorder>(
"reorder:_reshape_input_" + reorder_node->id() + "_" + reorder_reshape_node->id(),
input_node.id(),
reshape_in_layout.format,
format,
reshape_in_layout.data_type);
auto& reshape_input_node = p.get_or_create(reshape_input);
p.add_intermediate(reshape_input_node,

2
inference-engine/thirdparty/clDNN/tests/test_cases/reshape_gpu_test.cpp vendored
View File

@@ -465,7 +465,7 @@ TEST(reshape_gpu_f32, multiple_users_with_reorder) {
topology.add(activation("relu2", "reshape", activation_func::relu));
std::vector<float> input_vec = {-1.f, 2.f, -3.f, 4.f};
std::vector<float> out1 = {0.f, 0.f, 2.f, 4.0f};
std::vector<float> out1 = {0.f, 2.f, 0.f, 4.0f};
std::vector<float> out2 = {0.f, 2.f, 0.f, 4.0f};
set_values(input, input_vec);