COMPMID-1413 - Improve the performance of GEMMLowp with 8 bit dot product on OpenCL
COMPMID-1424 - Add dot product support for CLDepthwise QASYMM8 3x3 NHWC non-unit stride
With this patch we are able to improve the performance of MobileNet v1-qasymm8 by 37 %
Tried to use the dot product instruction in CLDepthwise QASYMM8 3x3 NHWC non-unit stride
but I have not seen any benefit (maybe because we have few arithemtic operation and we
do not have more load instructions). However Depthwise convolution has been improved by
30%
Change-Id: Id768a99c2e53a04276707e427af5d0ec93419ada
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/155082
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
diff --git a/src/core/Utils.cpp b/src/core/Utils.cpp
index 41fc87e..78c3dba 100644
--- a/src/core/Utils.cpp
+++ b/src/core/Utils.cpp
@@ -252,6 +252,19 @@
return pool_type_map[type];
}
+const std::string &arm_compute::string_from_gemmlowp_output_stage(GEMMLowpOutputStageType output_stage)
+{
+ static std::map<GEMMLowpOutputStageType, const std::string> output_stage_map =
+ {
+ { GEMMLowpOutputStageType::NONE, "" },
+ { GEMMLowpOutputStageType::QUANTIZE_DOWN, "quantize_down" },
+ { GEMMLowpOutputStageType::QUANTIZE_DOWN_FIXEDPOINT, "quantize_down_fixedpoint" },
+ { GEMMLowpOutputStageType::QUANTIZE_DOWN_FLOAT, "quantize_down_float" }
+ };
+
+ return output_stage_map[output_stage];
+}
+
std::string arm_compute::string_from_pixel_value(const PixelValue &value, const DataType data_type)
{
std::stringstream ss;