diff options
Diffstat (limited to 'src')
5 files changed, 276 insertions, 8 deletions
diff --git a/src/cpu/kernels/CpuElementwiseUnaryKernel.cpp b/src/cpu/kernels/CpuElementwiseUnaryKernel.cpp index 0adf28af63..4b61ee3a1e 100644 --- a/src/cpu/kernels/CpuElementwiseUnaryKernel.cpp +++ b/src/cpu/kernels/CpuElementwiseUnaryKernel.cpp @@ -51,18 +51,18 @@ std::unique_ptr<uint8_t[]> q8_prepare_lut(ElementWiseUnary op, const ITensorInfo ARM_COMPUTE_ERROR_ON(!is_data_type_quantized(src->data_type())); ARM_COMPUTE_ERROR_ON(src->element_size() != 1); - auto lut = std::unique_ptr<uint8_t[]>(new uint8_t[256]); + auto lut = std::unique_ptr<uint8_t[]>(new uint8_t[256]); const auto is_signed = src->data_type() == DataType::QASYMM8_SIGNED; - const auto src_qi = src->quantization_info().uniform(); - const auto dst_qi = dst->quantization_info().uniform(); + const auto src_qi = src->quantization_info().uniform(); + const auto dst_qi = dst->quantization_info().uniform(); const auto dst_min_fp = (((is_signed) ? -128 : 0) - dst_qi.offset) * dst_qi.scale; const auto dst_max_fp = (((is_signed) ? 127 : 255) - dst_qi.offset) * dst_qi.scale; for(int i = 0; i < 256; ++i) { - const auto in = (is_signed) ? dequantize_qasymm8_signed(static_cast<int8_t>(i), src_qi) : dequantize_qasymm8(i, src_qi); - float result = 0; + const auto in = (is_signed) ? dequantize_qasymm8_signed(static_cast<int8_t>(i), src_qi) : dequantize_qasymm8(i, src_qi); + float result = 0; switch(op) { @@ -101,7 +101,7 @@ std::unique_ptr<uint8_t[]> q8_prepare_lut(ElementWiseUnary op, const ITensorInfo result = utility::clamp(result, dst_min_fp, dst_max_fp); const auto out = (is_signed) ? static_cast<uint8_t>(quantize_qasymm8_signed(result, dst_qi)) : quantize_qasymm8(result, dst_qi); - lut[i] = out; + lut[i] = out; } return lut; @@ -174,7 +174,7 @@ static const std::vector<CpuElementwiseUnaryKernel::ElementwiseUnaryKernel> avai }, REGISTER_QASYMM8_SVE(sve_q8_elementwise_unary), &q8_prepare_lut, - }, + }, { "neon_q8_elementwise_unary", [](const DataTypeISASelectorData & data) @@ -184,6 +184,25 @@ static const std::vector<CpuElementwiseUnaryKernel::ElementwiseUnaryKernel> avai REGISTER_QASYMM8_NEON(neon_q8_elementwise_unary), &q8_prepare_lut, }, +#else // __aarch64__ + { + "neon_qasymm8_signed_elementwise_unary", + [](const DataTypeISASelectorData & data) + { + return data.dt == DataType::QASYMM8_SIGNED; + }, + REGISTER_QASYMM8_NEON(neon_qasymm8_signed_elementwise_unary), + nullptr, + }, + { + "neon_qasymm8_elementwise_unary", + [](const DataTypeISASelectorData & data) + { + return data.dt == DataType::QASYMM8; + }, + REGISTER_QASYMM8_NEON(neon_qasymm8_elementwise_unary), + nullptr, + }, #endif // __aarch64__ }; diff --git a/src/cpu/kernels/elementwise_unary/generic/neon/impl.cpp b/src/cpu/kernels/elementwise_unary/generic/neon/impl.cpp index 30caa4ebeb..2b23e46f40 100644 --- a/src/cpu/kernels/elementwise_unary/generic/neon/impl.cpp +++ b/src/cpu/kernels/elementwise_unary/generic/neon/impl.cpp @@ -1,5 +1,5 @@ /* - * Copyright (c) 2018-2022 Arm Limited. + * Copyright (c) 2018-2023 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -22,6 +22,8 @@ * SOFTWARE. */ #include "src/cpu/kernels/elementwise_unary/generic/neon/impl.h" +#include "arm_compute/core/Helpers.h" +#include "src/core/NEON/NEAsymm.h" namespace arm_compute { @@ -111,5 +113,166 @@ template void elementwise_op<__fp16>(const ITensor *in, ITensor *out, const Wind template void elementwise_op<float>(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op); template void elementwise_op<int32_t>(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op); +template <> +void elementwise_op<int8_t>(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op) +{ + 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 qi_in = in->info()->quantization_info().uniform(); + const UniformQuantizationInfo qi_out = out->info()->quantization_info().uniform(); + const auto min_clamped_value = vdupq_n_f32((-128 - qi_out.offset) * qi_out.scale); + const auto max_clamped_value = vdupq_n_f32((127 - qi_out.offset) * qi_out.scale); + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input(in, win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + int8x16_t vout; + auto output_ptr = reinterpret_cast<int8_t *>(output.ptr()); + const auto input_ptr = reinterpret_cast<const int8_t *>(input.ptr()); + const auto vconst_0_f32 = vdupq_n_f32(0); + auto clamped_value = (op == ElementWiseUnary::LOG) ? min_clamped_value : max_clamped_value; + + int x = window_start_x; + for(; x <= window_end_x - window_step_x; x += window_step_x) + { + const auto vin = wrapper::vloadq(input_ptr + x); + + // De-quantize + const auto vin_deq = vdequantize(vin, qi_in); + + // Perform activation + float32x4x4_t vtmp_deq = + { + { + elementwise_op_imp<float>(op, vin_deq.val[0]), + elementwise_op_imp<float>(op, vin_deq.val[1]), + elementwise_op_imp<float>(op, vin_deq.val[2]), + elementwise_op_imp<float>(op, vin_deq.val[3]), + } + }; + + if((op == ElementWiseUnary::LOG) || (op == ElementWiseUnary::RSQRT)) + { + vtmp_deq.val[0] = vbslq_f32(vcleq_f32(vin_deq.val[0], vconst_0_f32), clamped_value, vtmp_deq.val[0]); + vtmp_deq.val[1] = vbslq_f32(vcleq_f32(vin_deq.val[1], vconst_0_f32), clamped_value, vtmp_deq.val[1]); + vtmp_deq.val[2] = vbslq_f32(vcleq_f32(vin_deq.val[2], vconst_0_f32), clamped_value, vtmp_deq.val[2]); + vtmp_deq.val[3] = vbslq_f32(vcleq_f32(vin_deq.val[3], vconst_0_f32), clamped_value, vtmp_deq.val[3]); + } + + // Re-quantize to new output space + vout = vquantize_signed(vtmp_deq, qi_out); + wrapper::vstore(output_ptr + x, vout); + } + for(; x < window_end_x; ++x) + { + qasymm8_signed_t in = *(reinterpret_cast<const qasymm8_signed_t *>(input_ptr + x)); + qasymm8_signed_t tmp = 0; + float tmp_f = dequantize_qasymm8_signed(in, qi_in); + if(tmp_f <= 0.0) + { + if(op == ElementWiseUnary::LOG) + { + tmp_f = (-128 - qi_out.offset) * qi_out.scale; + } + else if(op == ElementWiseUnary::RSQRT) + { + tmp_f = (127 - qi_out.offset) * qi_out.scale; + } + } + else + { + tmp_f = elementwise_op_scalar_imp<float>(op, tmp_f); + } + tmp = quantize_qasymm8_signed(tmp_f, qi_out, RoundingPolicy::TO_ZERO); // Set rounding policy TO_ZERO to be compatible with vquantize_signed() used above that follow same policy for armv7a. + // For aarch64 LUT is used and rounding to nearest is used + *(output_ptr + x) = tmp; + } + }, + input, output); +} +template <> +void elementwise_op<uint8_t>(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op) +{ + 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 qi_in = in->info()->quantization_info().uniform(); + const UniformQuantizationInfo qi_out = out->info()->quantization_info().uniform(); + const auto vconst_0_f32 = vdupq_n_f32(0); + const auto min_clamped_value = vdupq_n_f32((0 - qi_out.offset) * qi_out.scale); + const auto max_clamped_value = vdupq_n_f32((255 - qi_out.offset) * qi_out.scale); + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input(in, win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + uint8x16_t vout; + auto clamped_value = (op == ElementWiseUnary::LOG) ? min_clamped_value : max_clamped_value; + auto output_ptr = reinterpret_cast<uint8_t *>(output.ptr()); + const auto input_ptr = reinterpret_cast<const uint8_t *>(input.ptr()); + int x = window_start_x; + for(; x <= window_end_x - window_step_x; x += window_step_x) + { + const auto vin = wrapper::vloadq(input_ptr + x); + + // De-quantize + const auto vin_deq = vdequantize(vin, qi_in); + + // Perform activation + float32x4x4_t vtmp_deq = + { + { + elementwise_op_imp<float>(op, vin_deq.val[0]), + elementwise_op_imp<float>(op, vin_deq.val[1]), + elementwise_op_imp<float>(op, vin_deq.val[2]), + elementwise_op_imp<float>(op, vin_deq.val[3]), + } + }; + if((op == ElementWiseUnary::LOG) || (op == ElementWiseUnary::RSQRT)) + { + vtmp_deq.val[0] = vbslq_f32(vcleq_f32(vin_deq.val[0], vconst_0_f32), clamped_value, vtmp_deq.val[0]); + vtmp_deq.val[1] = vbslq_f32(vcleq_f32(vin_deq.val[1], vconst_0_f32), clamped_value, vtmp_deq.val[1]); + vtmp_deq.val[2] = vbslq_f32(vcleq_f32(vin_deq.val[2], vconst_0_f32), clamped_value, vtmp_deq.val[2]); + vtmp_deq.val[3] = vbslq_f32(vcleq_f32(vin_deq.val[3], vconst_0_f32), clamped_value, vtmp_deq.val[3]); + } + + // Re-quantize to new output space + vout = vquantize(vtmp_deq, qi_out); + wrapper::vstore(output_ptr + x, vout); + } + for(; x < window_end_x; ++x) + { + qasymm8_t in = *(reinterpret_cast<const qasymm8_t *>(input_ptr + x)); + qasymm8_t tmp = 0; + float tmp_f = dequantize_qasymm8(in, qi_in); + if(tmp_f <= 0.0) + { + if(op == ElementWiseUnary::LOG) + { + tmp_f = (0 - qi_out.offset) * qi_out.scale; + } + else if(op == ElementWiseUnary::RSQRT) + { + tmp_f = (255 - qi_out.offset) * qi_out.scale; + } + } + else + { + tmp_f = elementwise_op_scalar_imp<float>(op, tmp_f); + } + tmp = quantize_qasymm8(tmp_f, qi_out, RoundingPolicy::TO_ZERO); + *(output_ptr + x) = tmp; + } + }, + input, output); +} } // namespace cpu } // namespace arm_compute diff --git a/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8.cpp b/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8.cpp new file mode 100644 index 0000000000..d987f7747b --- /dev/null +++ b/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8.cpp @@ -0,0 +1,42 @@ +/* + * Copyright (c) 2023 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 "arm_compute/core/Window.h" +#include "src/cpu/kernels/elementwise_unary/generic/neon/impl.h" + +namespace arm_compute +{ +namespace cpu +{ +#ifndef __aarch64__ +// Fallback function to be used for armv7a, for aarch64 LUT is used +void neon_qasymm8_elementwise_unary(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op, const uint8_t *lut) +{ + ARM_COMPUTE_UNUSED(lut); + return elementwise_op<uint8_t>(in, out, window, op); +} +#endif // #ifndef __aarch64__ + +} // namespace cpu +} // namespace arm_compute diff --git a/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8_signed.cpp b/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8_signed.cpp new file mode 100644 index 0000000000..e00970a1e0 --- /dev/null +++ b/src/cpu/kernels/elementwise_unary/generic/neon/qasymm8_signed.cpp @@ -0,0 +1,42 @@ +/* + * Copyright (c) 2023 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 "arm_compute/core/Window.h" +#include "src/cpu/kernels/elementwise_unary/generic/neon/impl.h" + +namespace arm_compute +{ +namespace cpu +{ +#ifndef __aarch64__ +// Fallback function to be used for armv7a, for aarch64 LUT is used +void neon_qasymm8_signed_elementwise_unary(const ITensor *in, ITensor *out, const Window &window, ElementWiseUnary op, const uint8_t *lut) +{ + ARM_COMPUTE_UNUSED(lut); + return elementwise_op<int8_t>(in, out, window, op); +} +#endif // #ifndef __aarch64__ + +} // namespace cpu +} // namespace arm_compute diff --git a/src/cpu/kernels/elementwise_unary/list.h b/src/cpu/kernels/elementwise_unary/list.h index 04c3bb6bcb..c1cfbb8a3a 100644 --- a/src/cpu/kernels/elementwise_unary/list.h +++ b/src/cpu/kernels/elementwise_unary/list.h @@ -42,6 +42,8 @@ DECLARE_ELEMETWISE_UNARY_KERNEL(neon_fp32_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_fp16_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_s32_elementwise_unary); DECLARE_ELEMETWISE_UNARY_KERNEL(neon_q8_elementwise_unary); +DECLARE_ELEMETWISE_UNARY_KERNEL(neon_qasymm8_signed_elementwise_unary); +DECLARE_ELEMETWISE_UNARY_KERNEL(neon_qasymm8_elementwise_unary); #undef DECLARE_ELEMETWISE_UNARY_KERNEL |