| commit | f2bd261353979db33503f5645202e7375eae8d99 | [log] [tgz] |
|---|---|---|
| author | Gian Marco Iodice <gianmarco.iodice@arm.com> | Tue Aug 07 17:22:24 2018 +0100 |
| committer | Anthony Barbier <anthony.barbier@arm.com> | Fri Nov 02 16:54:54 2018 +0000 |
| tree | a429b9212cfea49593dacc7a4e1944553b694132 | |
| parent | d51ea0afb0917f78f0c72e51c4947873f72ad537 [diff] [blame] |
COMPMID-1188 - Removed the multiplication by 4 in NEGEMMInterleavedWrapper Change-Id: Iaf8519bc483b947876a9b6ba83b4eb43b45b83a1 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/143135 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
diff --git a/src/runtime/NEON/functions/assembly/NEGEMMInterleavedWrapper.cpp b/src/runtime/NEON/functions/assembly/NEGEMMInterleavedWrapper.cpp index 434723c..b52ce66 100644 --- a/src/runtime/NEON/functions/assembly/NEGEMMInterleavedWrapper.cpp +++ b/src/runtime/NEON/functions/assembly/NEGEMMInterleavedWrapper.cpp
@@ -64,7 +64,7 @@ //Maximum number of workloads to create: const unsigned int num_threads = NEScheduler::get().num_threads(); - const unsigned int max_iterations = num_threads == 1 ? 1 : num_threads * 4; + const unsigned int max_iterations = num_threads == 1 ? 1 : num_threads; //Maximum number of iterations the parameters allow: const unsigned int num_iterations = _batch_window.num_iterations_total(); // Keep the smallest of the two: