Simd v8.0.166

C++ image processing and machine learning library with using of SIMD: SSE, AVX, AVX-512, AMX for x86/x64, NEON, SVE for ARM, HVX for Hexagon

Release notes

<h4>Algorithms</h4>
<h5>New features</h5>
<ul>
 <li>C++ wrapper Simd::SynetScale16b.</li>
 <li>C++ wrapper Simd::CopyFrame.</li>
 <li>Base implementation, SSE4.1, AVX2, AVX-512BW, AVX-512VNNI, AMX-INT8, NEON, SVE2 optimizations of class SynetQuantizedConvolutionNchwGemm.</li>
</ul>
<h5>Improving</h5>
<ul>
 <li>AMX-BF16 optimizations of class SynetQuantizedConvolutionNhwcSpecV1.</li>
 <li>SSE4.1 optimizations of class ImagePngLoader.</li>
 <li>SSE4.1, AVX2, AVX-512BW, NEON, SVE2 optimizations of method SynetConvolution32fNhwcDirect::ReorderWeight.</li>
 <li>SSE4.1, AVX2, AVX-512BW, NEON, SVE2 optimizations of methods SynetMergedConvolution32fCdc::ReorderFirstWeight, SynetMergedConvolution32fCdc::ReorderSecondWeight, SynetMergedConvolution32fCdc::ReorderThirdWeight.</li>
 <li>SSE4.1, AVX2, AVX-512BW, NEON, SVE2 optimizations of methods SynetMergedConvolution32fCd::ReorderFirstWeight, SynetMergedConvolution32fCd::ReorderSecondWeight.</li>
 <li>SSE4.1, AVX2, AVX-512BW, NEON, SVE2 optimizations of methods SynetMergedConvolution32fDc::ReorderFirstWeight, SynetMergedConvolution32fDc::ReorderSecondWeight.</li>
</ul>
<h5>Bug fixing</h5>
<ul>
 <li>Error in SSE4.1, AVX2, AVX-512BW, NEON, SVE2 optimizations of class SynetMergedConvolution32f.</li>
 <li>Error in AVX-512BW optimizations of class SynetConvolution32fNhwcDepthwise (microkernel Convolution32fNhwcDepthwise_k7p3d1s1w4).</li>
 <li>Error in AMX-BF16 optimizations of class SynetMergedConvolution16bCdc (microkernel DepthwiseConvolution_k7p3d1s1w4).</li>
</ul>
<h5>Removing</h5>
<ul>
 <li>Function SimdCopyFrame.</li>
 <li>Function SimdFill.</li>
 <li>Function SimdFillBgr.</li>
 <li>Function SimdFillBgra.</li>
 <li>Function SimdHogDirectionHistograms.</li>
 <li>C++ wrapper Simd::HogDirectionHistograms.</li>
 <li>Function SimdHogExtractFeatures.</li>
 <li>C++ wrapper Simd::HogExtractFeatures.</li>
 <li>Function SimdHogDeinterleave.</li>
 <li>Function SimdHogFilterSeparable.</li>
 <li>Types and functions of namespace Simd::Neural.</li>
 <li>Function SimdNeuralConvert.</li>
 <li>C++ wrapper Simd::NeuralConvert.</li>
 <li>Function SimdNeuralProductSum.</li>
 <li>Function SimdNeuralAddVector.</li>
 <li>Function SimdNeuralAddValue.</li>
 <li>Function SimdNeuralPow.</li>
 <li>Function SimdNeuralDerivativeSigmoid.</li>
 <li>Function SimdNeuralDerivativeTanh.</li>
 <li>Function SimdNeuralDerivativeRelu.</li>
 <li>Function SimdNeuralUpdateWeights.</li>
 <li>Function SimdNeuralAdaptiveGradientUpdate.</li>
 <li>Function SimdNeuralPooling1x1Max3x3.</li>
 <li>Function SimdNeuralPooling2x2Max2x2.</li>
 <li>Function SimdNeuralPooling2x2Max3x3.</li>
 <li>Function SimdNeuralAddConvolution2x2Forward.</li>
 <li>Function SimdNeuralAddConvolution3x3Forward.</li>
 <li>Function SimdNeuralAddConvolution4x4Forward.</li>
 <li>Function SimdNeuralAddConvolution5x5Forward.</li>
 <li>Function SimdNeuralAddConvolution2x2Backward.</li>
 <li>Function SimdNeuralAddConvolution3x3Backward.</li>
 <li>Function SimdNeuralAddConvolution4x4Backward.</li>
 <li>Function SimdNeuralAddConvolution5x5Backward.</li>
 <li>Function SimdNeuralAddConvolution2x2Sum.</li>
 <li>Function SimdNeuralAddConvolution3x3Sum.</li>
 <li>Function SimdNeuralAddConvolution4x4Sum.</li>
 <li>Function SimdNeuralAddConvolution5x5Sum.</li>
 <li>Function SimdNeuralConvolutionForward.</li>
 <li>Function SimdSynetAdd8i.</li>
 <li>Function SimdSynetInnerProduct8i.</li>
 <li>Function SimdSynetScale8iInit.</li>
 <li>Function SimdSynetScale8iInternalBufferSize.</li>
 <li>Function SimdSynetScale8iSetParams.</li>
 <li>Function SimdSynetScale8iForward.</li>
 <li>C++ wrapper Simd::SynetScale8i.</li>
 <li>Function SimdSynetMergedConvolution8iInit.</li>
 <li>Function SimdSynetMergedConvolution8iExternalBufferSize.</li>
 <li>Function SimdSynetMergedConvolution8iInternalBufferSize.</li>
 <li>Function SimdSynetMergedConvolution8iInfo.</li>
 <li>Function SimdSynetMergedConvolution8iSetParams.</li>
 <li…

Share this resource