projects / platform / upstream / armnn.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b98bbcf) | patch
IVGCVSW-2957 MergerLayer subtensor optimization now backend agnostic
authorDerek Lamberti <derek.lamberti@arm.com>
Mon, 15 Apr 2019 17:37:35 +0000 (18:37 +0100)
committerderek.lamberti <derek.lamberti@arm.com>
Tue, 16 Apr 2019 13:50:11 +0000 (13:50 +0000)
commit0790dcea1056298d63f97dec904c8ade5d21f439
treed75967e2eabe39ec08dc928fa77a7d4a51d85c5dtree | snapshot
parentb98bbcfa2a809c4ad025883c059ae49c82b37cbdcommit | diff
IVGCVSW-2957 MergerLayer subtensor optimization now backend agnostic

+ Update clframework pin
+ Cl and Neon Merger workloads updated to use MemoryLayout agnostic API
+ Workloads only use sub-tensor optimization if ALL input tensors are sub-tensors
+ Refactor LayerSupportCommon code to be a bit more succinct

Change-Id: Ib61ad4ccbd767e924dff07e61022e0cda4069828
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
13 files changed:
include/armnn/Tensor.hpp diff | blob | history
scripts/get_compute_library.sh diff | blob | history
src/armnn/LayerSupportCommon.hpp diff | blob | history
src/armnn/Tensor.cpp diff | blob | history
src/armnn/layers/MergerLayer.cpp diff | blob | history
src/backends/cl/ClLayerSupport.cpp diff | blob | history
src/backends/cl/ClWorkloadFactory.cpp diff | blob | history
src/backends/cl/workloads/ClMergerWorkload.cpp diff | blob | history
src/backends/cl/workloads/ClMergerWorkload.hpp diff | blob | history
src/backends/neon/NeonLayerSupport.cpp diff | blob | history
src/backends/neon/NeonWorkloadFactory.cpp diff | blob | history
src/backends/neon/workloads/NeonMergerWorkload.cpp diff | blob | history
src/backends/neon/workloads/NeonMergerWorkload.hpp diff | blob | history
Domain: Machine Learning / ML Framework;