projects / platform / upstream / dldt.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: bff2fe0) | patch
[IE][VPU]: Introduces ENABLE_EARLY_ELTWISE_RELU_FUSION option (#3343)
authorGladilov, Gleb <gleb.gladilov@intel.com>
Thu, 26 Nov 2020 10:27:53 +0000 (13:27 +0300)
committerGitHub <noreply@github.com>
Thu, 26 Nov 2020 10:27:53 +0000 (13:27 +0300)
commitc067ed1b9a2f2a026ba423c350693587af8abf21
treee0e745b254e84fbe339a9000b1c15bdaf8a21edftree | snapshot
parentbff2fe0b2ff7ccd5357b1411132f8c299eb3683bcommit | diff
[IE][VPU]: Introduces ENABLE_EARLY_ELTWISE_RELU_FUSION option (#3343)

Unconditional Eltwise+ReLU merge makes some models take a bit more memory,
so they could not be inferred on Myriad-X. With this change used can
optionally disable this optimization and use Myriad-X.
inference-engine/src/vpu/graph_transformer/include/vpu/graph_transformer.hpp diff | blob | history
inference-engine/src/vpu/graph_transformer/include/vpu/private_plugin_config.hpp diff | blob | history
inference-engine/src/vpu/graph_transformer/src/middleend/pass_manager.cpp diff | blob | history
inference-engine/src/vpu/graph_transformer/src/middleend/passes/merge_eltwise_and_relu.cpp diff | blob | history
inference-engine/src/vpu/graph_transformer/src/parsed_config.cpp diff | blob | history
Domain: Machine Learning / ML Framework;