projects / platform / upstream / armnn.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: efdbca6) | patch
In CopyTensorContentsGeneric coalesce inner dimensions where possible
authorMatthew Bentham <matthew.bentham@arm.com>
Sat, 14 Sep 2019 23:06:05 +0000 (00:06 +0100)
committerMatthew Bentham <matthew.bentham@arm.com>
Mon, 16 Sep 2019 09:07:13 +0000 (09:07 +0000)
commit019c4b1cd87849194cd30cb5bfd0e5d68089df08
tree68d7907a31a23a4e6fc47827ba49538e7b0688c2tree | snapshot
parentefdbca6b1a25dbd491be7f308dd729ff2255bb28commit | diff
In CopyTensorContentsGeneric coalesce inner dimensions where possible

This reduces the number of function calls in the inner loop, and
allows for optimised implementations of memcpy to improve bandwidth

Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I7458b45c075c87805242e92e54448b9dd762227f
src/backends/backendsCommon/WorkloadUtils.hpp diff | blob | history
Domain: Machine Learning / ML Framework;