Maxunpooling changes to enable fp16 in armv8a multi_isa builds

* Code guarded with __ARM_FEATURE_FP16_VECTOR_ARITHMETIC needs
  to be moved to an fp16.cpp file to allow compilation with
  -march=armv8.2-a+fp16

* fp16.cpp needs to use the template max_unpooling() which had to be moved from impl.cpp to impl.h

* Partially resolves MLCE-1102

Change-Id: Iabf9a9ba9d2441032f931f33aad97acc3e332575
Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10362
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>

6 files changed, 19 insertions, 66 deletions

diff --git a/Android.bp b/Android.bp
index 696942c866..ae0c79bddc 100644
--- a/Android.bp
+++ b/Android.bp
@@ -534,7 +534,6 @@ cc_library_static {
         "src/cpu/kernels/lut/generic/neon/u8.cpp",
         "src/cpu/kernels/maxunpool/generic/neon/fp16.cpp",
         "src/cpu/kernels/maxunpool/generic/neon/fp32.cpp",
-        "src/cpu/kernels/maxunpool/generic/neon/impl.cpp",
         "src/cpu/kernels/maxunpool/generic/neon/qasymm8.cpp",
         "src/cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp",
         "src/cpu/kernels/meanstddevnorm/generic/neon/fp16.cpp",
diff --git a/filelist.json b/filelist.json
index 215b363255..b74e2e25f1 100644
--- a/filelist.json
+++ b/filelist.json
@@ -1855,8 +1855,7 @@
             "fp32":["src/cpu/kernels/maxunpool/generic/neon/fp32.cpp"],
             "fp16":["src/cpu/kernels/maxunpool/generic/neon/fp16.cpp"],
             "qasymm8":["src/cpu/kernels/maxunpool/generic/neon/qasymm8.cpp"],
-            "qasymm8_signed":[ "src/cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp"],
-            "common":["src/cpu/kernels/maxunpool/generic/neon/impl.cpp"]
+            "qasymm8_signed":[ "src/cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp"]
           }
         }
       },
diff --git a/src/BUILD.bazel b/src/BUILD.bazel
index b989bd1963..f4fa950766 100644
--- a/src/BUILD.bazel
+++ b/src/BUILD.bazel
@@ -783,7 +783,6 @@ filegroup(
 	"cpu/kernels/lut/generic/neon/u8.cpp",
 	"cpu/kernels/maxunpool/generic/neon/fp16.cpp",
 	"cpu/kernels/maxunpool/generic/neon/fp32.cpp",
-	"cpu/kernels/maxunpool/generic/neon/impl.cpp",
 	"cpu/kernels/maxunpool/generic/neon/qasymm8.cpp",
 	"cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp",
 	"cpu/kernels/meanstddevnorm/generic/neon/fp16.cpp",
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index ee1ff476e6..eb17d51bbd 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -774,7 +774,6 @@ target_sources(
 	cpu/kernels/lut/generic/neon/u8.cpp
 	cpu/kernels/maxunpool/generic/neon/fp16.cpp
 	cpu/kernels/maxunpool/generic/neon/fp32.cpp
-	cpu/kernels/maxunpool/generic/neon/impl.cpp
 	cpu/kernels/maxunpool/generic/neon/qasymm8.cpp
 	cpu/kernels/maxunpool/generic/neon/qasymm8_signed.cpp
 	cpu/kernels/meanstddevnorm/generic/neon/fp16.cpp
diff --git a/src/cpu/kernels/maxunpool/generic/neon/impl.cpp b/src/cpu/kernels/maxunpool/generic/neon/impl.cpp
deleted file mode 100644
index 77e3b8594a..0000000000
--- a/src/cpu/kernels/maxunpool/generic/neon/impl.cpp
+++ /dev/null
@@ -1,54 +0,0 @@
-/*
- * Copyright (c) 2020-2022 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.
- */
-#include "src/cpu/kernels/maxunpool/generic/neon/impl.h"
-namespace arm_compute
-{
-class ITensor;
-class Window;
-namespace cpu
-{
-template <typename T>
-void max_unpooling(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window)
-{
-    Iterator  input_itr(input, window);
-    Iterator  indices_itr(indices, window);
-    auto      out_ptr      = reinterpret_cast<T *>(output->buffer());
-    const int out_stride_w = static_cast<int>(output->info()->strides_in_bytes()[3]);
-    execute_window_loop(window, [&](const Coordinates & id)
-    {
-        auto vindices                                         = reinterpret_cast<uint32_t *>(indices_itr.ptr());
-        auto vinput                                           = reinterpret_cast<T *>(input_itr.ptr());
-        out_ptr[id[3] * out_stride_w / sizeof(T) + *vindices] = *vinput;
-    },
-    input_itr, indices_itr);
-}
-template void max_unpooling<float>(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window);
-template void max_unpooling<int8_t>(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window);
-template void max_unpooling<uint8_t>(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window);
-
-#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS)
-template void max_unpooling<float16_t>(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window);
-#endif //defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS)
-} // namespace cpu
-} // namespace arm_compute
diff --git a/src/cpu/kernels/maxunpool/generic/neon/impl.h b/src/cpu/kernels/maxunpool/generic/neon/impl.h
index 3fea9cfcf3..5fe19c4707 100644
--- a/src/cpu/kernels/maxunpool/generic/neon/impl.h
+++ b/src/cpu/kernels/maxunpool/generic/neon/impl.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022 Arm Limited.
+ * Copyright (c) 2022-2023 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -21,19 +21,30 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef SRC_CORE_SVE_KERNELS_MAXUNPOOLING_IMPL_H
-#define SRC_CORE_SVE_KERNELS_MAXUNPOOLING_IMPL_H
+#ifndef ACL_SRC_CPU_KERNELS_MAXUNPOOL_GENERIC_NEON_IMPL_H
+#define ACL_SRC_CPU_KERNELS_MAXUNPOOL_GENERIC_NEON_IMPL_H
 #include "arm_compute/core/Helpers.h"
 #include "arm_compute/core/Window.h"
 #include "src/core/NEON/wrapper/wrapper.h"
 namespace arm_compute
 {
-class ITensor;
-class Window;
 namespace cpu
 {
 template <typename T>
-void max_unpooling(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window);
+void max_unpooling(const ITensor *input, const ITensor *indices, ITensor *output, const Window &window)
+{
+    Iterator  input_itr(input, window);
+    Iterator  indices_itr(indices, window);
+    auto      out_ptr      = reinterpret_cast<T *>(output->buffer());
+    const int out_stride_w = static_cast<int>(output->info()->strides_in_bytes()[3]);
+    execute_window_loop(window, [&](const Coordinates & id)
+    {
+        auto vindices                                         = reinterpret_cast<uint32_t *>(indices_itr.ptr());
+        auto vinput                                           = reinterpret_cast<T *>(input_itr.ptr());
+        out_ptr[id[3] * out_stride_w / sizeof(T) + *vindices] = *vinput;
+    },
+    input_itr, indices_itr);
+}
 } // namespace cpu
 } // namespace arm_compute
-#endif //define SRC_CORE_SVE_KERNELS_MAXUNPOOLING_IMPL_H
+#endif // ACL_SRC_CPU_KERNELS_MAXUNPOOL_GENERIC_NEON_IMPL_H