From 5f39091e502b0805f292d79a2a7da66d485f70ac Mon Sep 17 00:00:00 2001 From: Michalis Spyrou Date: Wed, 13 May 2020 00:12:08 +0100 Subject: COMPMID-3176: Remove padding from NEArithmeticSubtractionKernel COMPMID-3487: Refactor NEArithmeticSubtractionKernel Refactored code in order to remove paddings. This resulted in a big increase in libary size so after some rework the total size dropped by 4Kb. Change-Id: I4e3014c2ae49c29c6090b195ea16620afcf6c09f Signed-off-by: Michalis Spyrou Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3206 Comments-Addressed: Arm Jenkins Reviewed-by: Michele Di Giorgio Tested-by: Arm Jenkins --- arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h') diff --git a/arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h b/arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h index 919c685886..f75c6bfb98 100644 --- a/arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h +++ b/arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h @@ -52,7 +52,7 @@ public: /** Default destructor */ ~NEArithmeticSubtractionKernel() = default; - /** Initialise the kernel's input, output and border mode. + /** Initialise the kernel's input and output. * * Valid configurations (Input1,Input2) -> Output : * @@ -87,7 +87,6 @@ public: // Inherited methods overridden: void run(const Window &window, const ThreadInfo &info) override; - BorderSize border_size() const override; private: /** Common signature for all the specialised sub functions @@ -96,13 +95,15 @@ private: * @param[in] input2 An input tensor. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/QSYMM16/S16/F16/F32 * @param[out] output The output tensor. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/QSYMM16/S16/F16/F32. * @param[in] window Region on which to execute the kernel. + * @param[in] is_sat Flag to indicate if the policy is SATURATE. */ - using SubFunction = void(const ITensor *input1, const ITensor *input2, ITensor *output, const Window &window); + using SubFunction = void(const ITensor *input1, const ITensor *input2, ITensor *output, const Window &window, bool is_sat); /** Sub function to use for the particular tensor types passed to configure() */ SubFunction *_func; const ITensor *_input1; const ITensor *_input2; ITensor *_output; + ConvertPolicy _policy; }; } // namespace arm_compute #endif /* ARM_COMPUTE_NEARITHMETICSUBTRACTIONKERNEL_H */ -- cgit v1.2.1