diff options
| author | giuros01 <giuseppe.rossini@arm.com> | 2019-09-30 16:36:17 +0100 |
|---|---|---|
| committer | Gian Marco Iodice <gianmarco.iodice@arm.com> | 2019-10-01 09:11:56 +0000 |
| commit | 04ea4e88af1bdf9cf34436f7302c73e15f7bd4a5 (patch) | |
| tree | bf57a6580cbddfaab4a6f314776bb3a05741a889 | |
| parent | ae0fc8612dba6faebf58c3ebbfae8d6e639d432d (diff) | |
| download | ComputeLibrary-04ea4e88af1bdf9cf34436f7302c73e15f7bd4a5.tar.gz | |
COMPMID-2581: Fuse batch normalization with convolution and depthwise convolution at graph level for OpenCL - FP16
Change-Id: I757e27f45b5c25a5f98dacf2a6d519450d366760
Signed-off-by: giuros01 <giuseppe.rossini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2007
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
| -rw-r--r-- | src/graph/mutators/NodeFusionMutator.cpp | 3 |
1 files changed, 1 insertions, 2 deletions
diff --git a/src/graph/mutators/NodeFusionMutator.cpp b/src/graph/mutators/NodeFusionMutator.cpp index 4c3a905598..61d9479fca 100644 --- a/src/graph/mutators/NodeFusionMutator.cpp +++ b/src/graph/mutators/NodeFusionMutator.cpp @@ -318,8 +318,7 @@ void NodeFusionMutator::mutate(Graph &g) // Currently fuse batch normalization brings performance uplift only on OpenCL with FP32 data type // TODO (COMPMID-2524): Fuse batch normalization with convolution and depthwise convolution at graph level for NEON - FP32 - // TODO (COMPMID-2581): Fuse batch normalization with convolution and depthwise convolution at graph level for OpenCL - FP16 - if(target == Target::CL && (g.nodes()[0].get()->output(0)->desc().data_type == DataType::F32)) + if(target == Target::CL) { //Depthwise Convolution and Batch Normalization Fusion active only for CL detail::fuse_layer<ConvolutionLayerNode, BatchNormalizationLayerNode>(g, empty_prec, detail::fuse_convolution_with_batch_normalization); |