COMPMID-1451: Perform CLOutputStage using floats.

Change-Id: Ic8312a5b6790aa7cd4468d42f08d557ad40e9441
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/154570
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>

1 files changed, 16 insertions, 0 deletions

diff --git a/src/runtime/CL/functions/CLGEMMLowpOutputStage.cpp b/src/runtime/CL/functions/CLGEMMLowpOutputStage.cpp
index b18d23fac9..f5dc655776 100644
--- a/src/runtime/CL/functions/CLGEMMLowpOutputStage.cpp
+++ b/src/runtime/CL/functions/CLGEMMLowpOutputStage.cpp
@@ -25,6 +25,7 @@
 
 #include "arm_compute/core/CL/ICLTensor.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel.h"
+#include "arm_compute/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel.h"
 #include "arm_compute/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleKernel.h"
 #include "support/ToolchainSupport.h"
 
@@ -56,4 +57,19 @@ Status CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint::validate(const ITens
 {
     return CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel::validate(input, bias, output, min, max, output_3d_depth);
 }
+
+void CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloat::configure(const ICLTensor *input, const ICLTensor *bias, ICLTensor *output,
+                                                               float multiplier, int offset,
+                                                               int min, int max, unsigned int output_3d_depth)
+{
+    auto k = arm_compute::support::cpp14::make_unique<CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel>();
+    k->configure(input, bias, output, multiplier, offset, min, max, output_3d_depth);
+    _kernel = std::move(k);
+}
+
+Status CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloat::validate(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output,
+                                                                int min, int max, unsigned int output_3d_depth)
+{
+    return CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel::validate(input, bias, output, min, max, output_3d_depth);
+}
 } // namespace arm_compute
\ No newline at end of file