COMPMID-1451:Fuse RELU,LU_BOUNDED_RELU with requantization in NEGEMMConvolutionLayer.
Change-Id: Iea5f2c5bcac8051c4c7655a6eabb2c43772eb31f
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/154104
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
diff --git a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
index 55b70ff..fb6d4a1 100644
--- a/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
+++ b/src/runtime/NEON/functions/NEGEMMConvolutionLayer.cpp
@@ -32,6 +32,7 @@
#include "support/ToolchainSupport.h"
#include <cmath>
+#include <set>
#include <tuple>
using namespace arm_compute;
@@ -190,13 +191,14 @@
const unsigned int kernel_width = weights->info()->dimension(idx_width);
const unsigned int kernel_height = weights->info()->dimension(idx_height);
- _is_prepared = weights_info.retain_internal_weights();
- _original_weights = weights;
- _is_quantized = is_data_type_quantized_asymmetric(input->info()->data_type());
- _data_layout = data_layout;
- _skip_im2col = (data_layout == DataLayout::NHWC && kernel_width == 1 && kernel_height == 1 && conv_info.stride().first == 1 && conv_info.stride().second == 1);
- _skip_col2im = data_layout == DataLayout::NHWC;
- _append_bias = (biases != nullptr) && (!_is_quantized);
+ _is_prepared = weights_info.retain_internal_weights();
+ _original_weights = weights;
+ _is_quantized = is_data_type_quantized_asymmetric(input->info()->data_type());
+ _data_layout = data_layout;
+ _skip_im2col = (data_layout == DataLayout::NHWC && kernel_width == 1 && kernel_height == 1 && conv_info.stride().first == 1 && conv_info.stride().second == 1);
+ _skip_col2im = data_layout == DataLayout::NHWC;
+ _append_bias = (biases != nullptr) && (!_is_quantized);
+ _is_activationlayer_enabled = act_info.enabled();
const ITensor *gemm_input_to_use = input;
ITensor *gemm_output_to_use = output;
@@ -285,9 +287,10 @@
if(_is_quantized)
{
const bool skip_reshape = data_layout == DataLayout::NHWC;
- const QuantizationInfo output_quant_info = (output->info()->total_size() == 0) ? input->info()->quantization_info() : output->info()->quantization_info();
+ const QuantizationInfo input_quant_info = input->info()->quantization_info();
+ const QuantizationInfo output_quant_info = (output->info()->total_size() == 0) ? input_quant_info : output->info()->quantization_info();
- float multiplier = input->info()->quantization_info().scale * weights->info()->quantization_info().scale / output_quant_info.scale;
+ float multiplier = input_quant_info.scale * weights->info()->quantization_info().scale / output_quant_info.scale;
int output_multiplier, output_shift;
quantization::calculate_quantized_multiplier_less_than_one(multiplier, &output_multiplier, &output_shift);
@@ -297,7 +300,29 @@
gemm_output_staged_to_use = &_tmp_output;
}
- _gemmlowp_output_stage.configure(gemm_output_to_use, biases, gemm_output_staged_to_use, output_multiplier, output_shift, output_quant_info.offset, 0, 0, skip_reshape ? conv_h : 1);
+ // Merge activation with output stage
+ uint8_t min = 0;
+ uint8_t max = 0;
+ const std::set<ActivationLayerInfo::ActivationFunction> supported_acts = { ActivationLayerInfo::ActivationFunction::RELU,
+ ActivationLayerInfo::ActivationFunction::BOUNDED_RELU,
+ ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU
+ };
+ if(_is_activationlayer_enabled && supported_acts.count(act_info.activation()) != 0)
+ {
+ min = sqcvt_qasymm8_f32(act_info.b(), input_quant_info.scale, input_quant_info.offset);
+ max = sqcvt_qasymm8_f32(act_info.a(), input_quant_info.scale, input_quant_info.offset);
+ if(act_info.activation() != ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU)
+ {
+ min = sqcvt_qasymm8_f32(0.f, input_quant_info.scale, input_quant_info.offset);
+ }
+ if(act_info.activation() == ActivationLayerInfo::ActivationFunction::RELU)
+ {
+ max = 255;
+ }
+ _is_activationlayer_enabled = false;
+ }
+
+ _gemmlowp_output_stage.configure(gemm_output_to_use, biases, gemm_output_staged_to_use, output_multiplier, output_shift, output_quant_info.offset, min, max, skip_reshape ? conv_h : 1);
}
if(!_skip_col2im && _data_layout == DataLayout::NCHW)
@@ -319,9 +344,7 @@
ARM_COMPUTE_ERROR_ON_MSG((output->info()->dimension(idx_width) != conv_w) || (output->info()->dimension(idx_height) != conv_h),
"Output shape does not match the expected one");
- //Configure Activation Layer
- _is_activationlayer_enabled = act_info.enabled();
-
+ // Configure Activation Layer
if(_is_activationlayer_enabled)
{
_activationlayer_function.configure(output, nullptr, act_info);
@@ -356,10 +379,11 @@
const ITensorInfo *gemm_output_staged_to_use = output;
const ITensorInfo *weights_to_use = weights;
- const bool is_quantized = is_data_type_quantized_asymmetric(data_type);
- const bool append_bias = (biases != nullptr) && (!is_quantized);
- bool skip_im2col = (data_layout == DataLayout::NHWC && kernel_width == 1 && kernel_height == 1 && conv_info.stride().first == 1 && conv_info.stride().second == 1);
- bool skip_col2im = data_layout == DataLayout::NHWC;
+ const bool is_quantized = is_data_type_quantized_asymmetric(data_type);
+ const bool append_bias = (biases != nullptr) && (!is_quantized);
+ bool skip_im2col = (data_layout == DataLayout::NHWC && kernel_width == 1 && kernel_height == 1 && conv_info.stride().first == 1 && conv_info.stride().second == 1);
+ bool skip_col2im = data_layout == DataLayout::NHWC;
+ bool is_activation_enabled = act_info.enabled();
// Get convolved dimensions
unsigned int conv_w = 0;
@@ -457,9 +481,11 @@
if(is_quantized)
{
- const bool skip_reshape = data_layout == DataLayout::NHWC;
- const float multiplier = input->quantization_info().scale * weights_to_use->quantization_info().scale / output->quantization_info().scale;
- int output_multiplier, output_shift;
+ const bool skip_reshape = data_layout == DataLayout::NHWC;
+ const QuantizationInfo input_quant_info = input->quantization_info();
+ const QuantizationInfo output_quant_info = (output->total_size() == 0) ? input_quant_info : output->quantization_info();
+ const float multiplier = input_quant_info.scale * weights_to_use->quantization_info().scale / output_quant_info.scale;
+ int output_multiplier, output_shift;
quantization::calculate_quantized_multiplier_less_than_one(multiplier, &output_multiplier, &output_shift);
if(!skip_reshape)
@@ -469,8 +495,30 @@
gemm_output_staged_to_use = &tmp_info;
}
+ // Merge activation with output stage
+ uint8_t min = 0;
+ uint8_t max = 0;
+ const std::set<ActivationLayerInfo::ActivationFunction> supported_acts = { ActivationLayerInfo::ActivationFunction::RELU,
+ ActivationLayerInfo::ActivationFunction::BOUNDED_RELU,
+ ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU
+ };
+ if(is_activation_enabled && supported_acts.count(act_info.activation()) != 0)
+ {
+ min = sqcvt_qasymm8_f32(act_info.b(), input_quant_info.scale, input_quant_info.offset);
+ max = sqcvt_qasymm8_f32(act_info.a(), input_quant_info.scale, input_quant_info.offset);
+ if(act_info.activation() != ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU)
+ {
+ min = sqcvt_qasymm8_f32(0.f, input_quant_info.scale, input_quant_info.offset);
+ }
+ if(act_info.activation() == ActivationLayerInfo::ActivationFunction::RELU)
+ {
+ max = 255;
+ }
+ is_activation_enabled = false;
+ }
+
// Validate output stage for quantized case
- NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint::validate(gemm_output_to_use, biases, gemm_output_staged_to_use, 0, 0, skip_reshape ? conv_h : 1);
+ NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint::validate(gemm_output_to_use, biases, gemm_output_staged_to_use, min, max, skip_reshape ? conv_h : 1);
}
// Validate Col2Im/ReshapeLayer
@@ -482,7 +530,7 @@
}
//Validate Activation Layer
- if(act_info.enabled())
+ if(is_activation_enabled)
{
ARM_COMPUTE_RETURN_ON_ERROR(NEActivationLayer::validate(output, nullptr, act_info));
}