commit 05288a2b871ef99f544771621c3bba409b2f70df
author Gian Marco <gianmarco.iodice@arm.com> Tue Nov 21 10:57:50 2017 +0000
committer Anthony Barbier <anthony.barbier@arm.com> Fri Nov 02 16:41:17 2018 +0000
tree 21e3d2a9927ef31f6d5bcdd5523c4c8e933047a6
parent c82799003fbfdc5bb9526ff944e41eaae23e3f03

COMPMID-697 - Rework GEMMLowp interface on OpenCL

Reworked the interface of GemmLowp in order to make easy the integration
in Android NN

- Added support for different output stage
- Added validation for both matrix multiplication and output stage
- Added bounded relu support in the output stage
- Added in32_t bias support
- Added optimized path for vector by matrix case

This rework is required for:
- Convolution quantized
- Fully connected quantized

Change-Id: I512283d406099cf8c614dd89d0a97ed411143afc
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/110625
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: BSG Visual Compute Jenkins server to access repositories on http://mpd-gerrit.cambridge.arm.com <bsgcomp@arm.com>

arm_compute/runtime/CL/CLFunctions.h

diff --git a/arm_compute/runtime/CL/CLFunctions.h b/arm_compute/runtime/CL/CLFunctions.h
index 360372d..9a20769 100644
--- a/arm_compute/runtime/CL/CLFunctions.h
+++ b/arm_compute/runtime/CL/CLFunctions.h
@@ -59,7 +59,8 @@
 #include "arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h"
 #include "arm_compute/runtime/CL/functions/CLGEMM.h"
 #include "arm_compute/runtime/CL/functions/CLGEMMInterleave4x4.h"
-#include "arm_compute/runtime/CL/functions/CLGEMMLowp.h"
+#include "arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h"
+#include "arm_compute/runtime/CL/functions/CLGEMMLowpOutputStage.h"
 #include "arm_compute/runtime/CL/functions/CLGEMMTranspose1xW.h"
 #include "arm_compute/runtime/CL/functions/CLGaussian3x3.h"
 #include "arm_compute/runtime/CL/functions/CLGaussian5x5.h"