diff options
Diffstat (limited to 'src/core/NEON/kernels/NEElementwiseOperationKernel.cpp')
| -rw-r--r-- | src/core/NEON/kernels/NEElementwiseOperationKernel.cpp | 85 |
1 files changed, 85 insertions, 0 deletions
diff --git a/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp b/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp index b8e6f0cc69..444ee8e0d3 100644 --- a/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp +++ b/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp @@ -534,6 +534,24 @@ inline int elementwise_comp_op_quantized_loop(int window_start_x, int window_end return x; } +template <ComparisonOperation op> +inline int elementwise_comp_op_quantized_signed_loop(int window_start_x, int window_end_x, int window_step_x, + const int8_t *input1_ptr, const int8_t *input2_ptr, uint8_t *output_ptr, + int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, + float32x4_t voffseto, float32x4_t invvscaleo) +{ + ARM_COMPUTE_UNUSED(voffseto, invvscaleo); + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized_signed(input1_ptr + x, voffset1, vscale1); + const float32x4x4_t bf = load_quantized_signed(input2_ptr + x, voffset2, vscale2); + const uint32x4x4_t rf = elementwise_comp_op<op>(af, bf); + store_quantized(output_ptr + x, rf); + } + return x; +} + template <ComparisonOperation op, typename InputScalarType, typename InputVectorType> inline int elementwise_comp_op_broadcast_16_loop(int window_start_x, int window_end_x, int window_step_x, const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) @@ -772,6 +790,66 @@ void elementwise_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *o } } +void elementwise_comp_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, + uint8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), + int (*neon_func)(int, int, int, const int8_t *, const int8_t *, uint8_t *, + int32x4_t, int32x4_t, float32x4_t, float32x4_t, + float32x4_t, float32x4_t)) +{ + // Create input windows + Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); + Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); + + // Clear X Dimension on execution window as we handle manually + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = 16; + const auto window_start_x = static_cast<int>(window.x().start()); + const auto window_end_x = static_cast<int>(window.x().end()); + const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); + + const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset); + const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); + { + const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); + const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); + + // Input1 quantization info + const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); + const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); + + // Input2 quantization info + const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); + const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input1(in1, input1_win); + Iterator input2(in2, input2_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto input1_ptr = reinterpret_cast<const int8_t *>(input1.ptr()); + const auto input2_ptr = reinterpret_cast<const int8_t *>(input2.ptr()); + const auto output_ptr = reinterpret_cast<uint8_t *>(output.ptr()); + + int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, + vscale1, vscale2, voffseto, invvscaleo); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8_signed(*(input1_ptr + x), input1_qinfo); + const float bfs = dequantize_qasymm8_signed(*(input2_ptr + x), input2_qinfo); + *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); + } + }, + input1, input2, output); + } +} + void elementwise_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, int8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), int (*broadcast_func)(int, int, int, const int8_t *, float32x4x4_t, int8_t *, int32x4_t, float32x4_t, @@ -931,6 +1009,12 @@ void elementwise_comp_op_quantized(const ITensor *in1, const ITensor *in2, ITens &elementwise_comp_op_quantized_loop<op>); } +template <ComparisonOperation op> +void elementwise_comp_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_comp_quantized_signed(in1, in2, out, window, &elementwise_comp_op_quantized_scalar<op>, &elementwise_comp_op_quantized_signed_loop<op>); +} + std::function<void(const ITensor *, const ITensor *, ITensor *, const Window &)> configure_func(const ITensor *input1, const ITensor *input2, ITensor *output, std::map<std::string, NEElementwiseOperationKernel::ElementwiseFunction *> map_function) @@ -981,6 +1065,7 @@ configure_comp_func(const ITensor *input1, const ITensor *input2, ITensor *outpu { "op_F32_F32_U8", &elementwise_comp_op_32<op, float, float32x4_t> }, { "op_S16_S16_U8", &elementwise_comp_op_16<op, int16_t, int16x8_t> }, { "op_S32_S32_U8", &elementwise_comp_op_32<op, int32_t, int32x4_t> }, + { "op_QASYMM8_SIGNED_QASYMM8_SIGNED_U8", &elementwise_comp_op_quantized_signed<op> }, { "op_QASYMM8_QASYMM8_U8", &elementwise_comp_op_quantized<op> } }; #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC |