Enable FP16 in multi_isa+v8a

* Enable FP16 kernels in

    NEROIAlignLayerKernel
    NEComputeAllAnchorsKernel
    NEBoundingBoxTransformKernel
    NEInstanceNormalizationLayerKernel
    NEBatchNormalizationLayerKernel

* The FP16 kernels were disabled due to the use of __ARM_FEATURE_FP16_VECTOR_ARITHMETIC

* Resolves MLCE-1305

Change-Id: Ib8dd3cad631667018b25db4ba76007dbfb4bf5a5
Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11677
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

5 files changed, 15 insertions, 15 deletions

diff --git a/src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp b/src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp
index 717fd11485..153c36052a 100644
--- a/src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp
+++ b/src/core/NEON/kernels/NEBatchNormalizationLayerKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2021, 2023 Arm Limited.
+ * Copyright (c) 2017-2021, 2023-2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -78,11 +78,11 @@ static const BatchNormalizationKernel available_kernels[] = {
      REGISTER_FP32_SVE(arm_compute::cpu::fp32_sve_batch_normalization)},
 #endif /* !defined(ARM_COMPUTE_ENABLE_SVE) */
 #if defined(ARM_COMPUTE_ENABLE_NEON)
-#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC)
+#if ARM_COMPUTE_ENABLE_FP16
     {"neon_fp16_batch_normalization",
      [](const BatchNormalizationSelectorData &data) { return data.dt == DataType::F16; },
      REGISTER_FP16_NEON(arm_compute::cpu::fp16_neon_batch_normalization)},
-#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
     {"neon_fp32_batch_normalization",
      [](const BatchNormalizationSelectorData &data) { return data.dt == DataType::F32; },
      REGISTER_FP32_NEON(arm_compute::cpu::fp32_neon_batch_normalization)},
diff --git a/src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp b/src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp
index cb869838e2..694def1a3a 100644
--- a/src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp
+++ b/src/core/NEON/kernels/NEBoundingBoxTransformKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022 Arm Limited.
+ * Copyright (c) 2019-2022, 2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -63,11 +63,11 @@ static const BoundingBoxTransformKernel available_kernels[] = {
     {"fp32_neon_boundingboxtransform",
      [](const BoundingBoxTransformSelectorData &data) { return data.dt == DataType::F32; },
      REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_boundingboxtransform)},
-#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#ifdef ARM_COMPUTE_ENABLE_FP16
     {"fp16_neon_boundingboxtransform",
      [](const BoundingBoxTransformSelectorData &data) { return data.dt == DataType::F16; },
      REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_boundingboxtransform)},
-#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#endif // ARM_COMPUTE_ENABLE_FP16
 #if defined(ARM_COMPUTE_ENABLE_NEON)
     {"qu16_neon_boundingboxtransform",
      [](const BoundingBoxTransformSelectorData &data) { return data.dt == DataType::QASYMM16; },
diff --git a/src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp b/src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp
index 549319e49f..e23e3d020f 100644
--- a/src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp
+++ b/src/core/NEON/kernels/NEGenerateProposalsLayerKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022 Arm Limited.
+ * Copyright (c) 2019-2022, 2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -61,10 +61,10 @@ static const ComputeAllAnchorsKernel available_kernels[] = {
     {"neon_qu16_computeallanchors", [](const ComputeAllAnchorsData &data) { return data.dt == DataType::QSYMM16; },
      REGISTER_QSYMM16_NEON(arm_compute::cpu::neon_qu16_computeallanchors)},
 #endif //defined(ARM_COMPUTE_ENABLE_NEON)
-#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#ifdef ARM_COMPUTE_ENABLE_FP16
     {"neon_fp16_computeallanchors", [](const ComputeAllAnchorsData &data) { return data.dt == DataType::F16; },
      REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_computeallanchors)},
-#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#endif // ARM_COMPUTE_ENABLE_FP16
     {"neon_fp32_computeallanchors", [](const ComputeAllAnchorsData &data) { return data.dt == DataType::F32; },
      REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_computeallanchors)},
 };
diff --git a/src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp b/src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp
index 0a1780f6ee..5883731088 100644
--- a/src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp
+++ b/src/core/NEON/kernels/NEInstanceNormalizationLayerKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022 Arm Limited.
+ * Copyright (c) 2019-2022, 2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -70,10 +70,10 @@ struct InstanceNormKernel
 static const InstanceNormKernel available_kernels[] = {
     {"fp32_neon_instancenorm", [](const InstanceNormSelectorData &data) { return data.dt == DataType::F32; },
      REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_instancenorm)},
-#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#ifdef ARM_COMPUTE_ENABLE_FP16
     {"fp16_neon_instancenorm", [](const InstanceNormSelectorData &data) { return data.dt == DataType::F16; },
      REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_instancenorm)},
-#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#endif // ARM_COMPUTE_ENABLE_FP16
 };
 
 /** Micro-kernel selector
diff --git a/src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp b/src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp
index 451031d696..cfe4ac9a4c 100644
--- a/src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp
+++ b/src/core/NEON/kernels/NEMeanStdDevNormalizationKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2022 Arm Limited.
+ * Copyright (c) 2019-2022, 2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -60,10 +60,10 @@ struct MeanStdDevNormKernel
 static const std::vector<MeanStdDevNormKernel> available_kernels = {
     {"fp32_neon_meanstddevnorm", [](const MeanStdDevNormSelectorData &data) { return data.dt == DataType::F32; },
      REGISTER_FP32_NEON(arm_compute::cpu::neon_fp32_meanstddevnorm)},
-#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#ifdef ARM_COMPUTE_ENABLE_FP16
     {"fp16_neon_meanstddevnorm", [](const MeanStdDevNormSelectorData &data) { return data.dt == DataType::F16; },
      REGISTER_FP16_NEON(arm_compute::cpu::neon_fp16_meanstddevnorm)},
-#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+#endif // ARM_COMPUTE_ENABLE_FP16
     {"qasymm8_neon_meanstddevnorm", [](const MeanStdDevNormSelectorData &data) { return data.dt == DataType::QASYMM8; },
      REGISTER_QASYMM8_NEON(arm_compute::cpu::neon_qasymm8_meanstddevnorm)},
 };