diff options
author | Michalis Spyrou <michalis.spyrou@arm.com> | 2018-11-20 18:38:29 +0000 |
---|---|---|
committer | Georgios Pinitas <georgios.pinitas@arm.com> | 2018-11-22 13:22:24 +0000 |
commit | 2897e61e8fe04aaf95540f4525c3dd3f7f46ebfa (patch) | |
tree | a887648225c485ddf06363170d6111290b6111eb /arm_compute/core/NEON/wrapper | |
parent | 303f0dbebf631b3db00d9d64e71018abbbe9d4fe (diff) | |
download | ComputeLibrary-2897e61e8fe04aaf95540f4525c3dd3f7f46ebfa.tar.gz |
COMPMID-1645 NEL2Normalization for FP32/FP16 & NHWC
Change-Id: I29e35024e29781a6b943b568abec9c73649215e6
Diffstat (limited to 'arm_compute/core/NEON/wrapper')
-rw-r--r-- | arm_compute/core/NEON/wrapper/intrinsics/intrinsics.h | 1 | ||||
-rw-r--r-- | arm_compute/core/NEON/wrapper/intrinsics/invsqrt.h | 53 |
2 files changed, 54 insertions, 0 deletions
diff --git a/arm_compute/core/NEON/wrapper/intrinsics/intrinsics.h b/arm_compute/core/NEON/wrapper/intrinsics/intrinsics.h index 2e6fd75005..7ea0aba565 100644 --- a/arm_compute/core/NEON/wrapper/intrinsics/intrinsics.h +++ b/arm_compute/core/NEON/wrapper/intrinsics/intrinsics.h @@ -32,6 +32,7 @@ #include "arm_compute/core/NEON/wrapper/intrinsics/getlane.h" #include "arm_compute/core/NEON/wrapper/intrinsics/getlow.h" #include "arm_compute/core/NEON/wrapper/intrinsics/inv.h" +#include "arm_compute/core/NEON/wrapper/intrinsics/invsqrt.h" #include "arm_compute/core/NEON/wrapper/intrinsics/load.h" #include "arm_compute/core/NEON/wrapper/intrinsics/max.h" #include "arm_compute/core/NEON/wrapper/intrinsics/min.h" diff --git a/arm_compute/core/NEON/wrapper/intrinsics/invsqrt.h b/arm_compute/core/NEON/wrapper/intrinsics/invsqrt.h new file mode 100644 index 0000000000..0bbf49b5c0 --- /dev/null +++ b/arm_compute/core/NEON/wrapper/intrinsics/invsqrt.h @@ -0,0 +1,53 @@ +/* + * Copyright (c) 2018 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. + */ +#ifndef __ARM_COMPUTE_WRAPPER_INVSQRT_H__ +#define __ARM_COMPUTE_WRAPPER_INVSQRT_H__ + +#include "arm_compute/core/NEON/NEMath.h" +#include <arm_neon.h> + +namespace arm_compute +{ +namespace wrapper +{ +#define VINVSQRT_IMPL(stype, vtype, prefix, postfix) \ + inline vtype vinvsqrt(const vtype &a) \ + { \ + return prefix##_##postfix(a); \ + } + +VINVSQRT_IMPL(float, float32x2_t, vinvsqrt, f32) +#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC +VINVSQRT_IMPL(float16_t, float16x4_t, vinvsqrt, f16) +#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC + +VINVSQRT_IMPL(float, float32x4_t, vinvsqrtq, f32) +#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC +VINVSQRT_IMPL(float16_t, float16x8_t, vinvsqrtq, f16) +#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC + +#undef VINVSQRT_IMPL +} // namespace wrapper +} // namespace arm_compute +#endif /* __ARM_COMPUTE_WRAPPER_INVSQRT_H__ */ |