COMPMID-2577: Fuse bias addition and activation in gemm assembly kernels
Change-Id: I7f52112d2d05b1ea3d3f3d4b19b8eafab05d6c44
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2141
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/bias_adder.hpp b/src/core/NEON/kernels/arm_gemm/bias_adder.hpp
new file mode 100644
index 0000000..745d005
--- /dev/null
+++ b/src/core/NEON/kernels/arm_gemm/bias_adder.hpp
@@ -0,0 +1,75 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#pragma once
+
+namespace arm_gemm
+{
+
+// Fallback routine to add bias to a block
+template<typename T>
+inline void bias_adder(T *out, unsigned int stride, const T *bias, unsigned int rows, unsigned int cols) {
+ for (unsigned int row=0; row<rows; row++) {
+ for (unsigned int col=0; col<cols; col++) {
+ out[row * stride + col] += bias[col];
+ }
+ }
+}
+
+template<bool DoBias, typename T>
+inline void activator(T *out, unsigned int stride, const T *bias, Activation act, unsigned int rows, unsigned int cols) {
+ if (act.type == Activation::Type::None) {
+ if (DoBias) {
+ bias_adder(out, stride, bias, rows, cols);
+ }
+ return;
+ }
+
+ if (act.type == Activation::Type::ReLU) {
+ for (unsigned int row=0; row<rows; row++) {
+ for (unsigned int col=0; col<cols; col++) {
+ T &v = out[row * stride + col];
+ if (DoBias) {
+ v += bias[col];
+ }
+ v = std::max(static_cast<T>(0), v);
+ }
+ }
+ }
+
+ if (act.type == Activation::Type::BoundedReLU) {
+ const T max = static_cast<T>(act.param1);
+
+ for (unsigned int row=0; row<rows; row++) {
+ for (unsigned int col=0; col<cols; col++) {
+ T &v = out[row * stride + col];
+ if (DoBias) {
+ v += bias[col];
+ }
+ v = std::max(static_cast<T>(0), std::min(v, max));
+ }
+ }
+ }
+}
+
+} // namespace arm_gemm