Simd Library Release Notes (2015).

Home | Release Notes | Download | Documentation | Forum | SourceForge | GitHub

2017 | 2016 | 2015 | 2014 | 2013

December 1, 2015 (version 3.0.36.559)

Algorithms

New features
  • Extended functionality of function OperationBinary16i (added Subtraction operation).
  • Base implementation, SSE2, SSSE3, AVX2 and VMX optimizations of function MeanFilter3x3.
  • Macro for ARM NEON architecture.
  • Runtime ARM NEON detection.
  • NEON optimization of function OperationBinary16i.
  • NEON optimization of function OperationBinary8u.
  • NEON optimization of function VectorProduct.
  • NEON optimization of function ReduceGray2x2.
  • NEON optimization of function Reorder16bit.
  • NEON optimization of function Reorder32bit.
  • NEON optimization of function Reorder64bit.

Test framework

New features
  • Tests for verifying functionality of function MeanFilter3x3.
  • Performance report generation for ARM NEON.
  • Tests for verifying functionality of NEON version of function OperationBinary16i.
  • Tests for verifying functionality of NEON version of function OperationBinary8u.
  • Tests for verifying functionality of NEON version of function VectorProduct.
  • Tests for verifying functionality of NEON version of function ReduceGray2x2.
  • Tests for verifying functionality of NEON version of function Reorder16bit.
  • Tests for verifying functionality of NEON version of function Reorder32bit.
  • Tests for verifying functionality of NEON version of function Reorder64bit.

Infrastructure

New features
  • Added to CMakeLists.txt a section for building of the library for ARM(32-bit/64-bit) CPU architecture.
  • Neon project for ARM NEON extension (CMake and Visual Studio).
Home

November 2, 2015 (version 2.3.35.542)

Algorithms

New features
  • Pyramid structure.

Test framework

Bug fixing
  • Using uninitialized test data in tests for verifying functionality of function ConditionalFill.

Documentation

New features
  • A description of the library building.
  • A description of test application using.

Infrastructure

Removing
  • Project files for Visual Studio 2012.
Home

October 1, 2015 (version 2.3.34.537)

Algorithms

New features
  • Frame structure.
  • Copy function (to copy frames).
  • Convert function (to convert frames).
  • Convert function (to convert one plane images of Gray8, Bgr24 and Bgra32 formats).
  • Base implementation, SSE2, AVX2 and VMX optimizations of function InterleaveUv.
  • Base implementation, SSE2, AVX2 and VMX optimizations of function ConditionalFill.

Test framework

New features
  • Tests for verifying functionality of function InterleaveUv.
  • Tests for verifying functionality of function ConditionalFill.
Home

September 4, 2015 (version 2.3.33.529)

Algorithms

New features
  • Extended functionality of function OperationBinary8u (added Minimum operation).
Bug fixing
  • Compiler errors for Visual Studio 2015.
Home

August 3, 2015 (version 2.3.32.526)

Algorithms

Bug fixing
  • A bug in algorithm of tilted sum estimation.
  • Loading of unaligned address as aligned in SSE2, SSSE3 and AVX2 optimizations of function SquaredDifferenceSumMasked.
  • Loading of unaligned address as aligned in SSE2 and AVX2 optimizations of function Bgr48pToBgra32.
  • A crash in function ResizeBilinear (resizing of 8-bit gray image from 642x300 to 200x200).
Home

July 1, 2015 (version 2.3.31.522)

Algorithms

New features
  • SSE2, AVX2 and VMX optimizations of function HistogramMasked.
  • SSSE3 optimization of function ReduceGray4x4.
  • SSSE3 optimization of function Laplace.
  • SSSE3 optimization of function SobelDx.
  • SSSE3 optimization of function SobelDy.
  • SSSE3 optimization of function TextureBoostedSaturatedGradient.
  • SSSE3 optimization of function SquaredDifferenceSum.
  • SSSE3 optimization of function SquaredDifferenceSumMasked.
  • Base implementation of function NormalizeHistogram.
  • Base implementation, SSE, AVX and VSX optimizations of function SquaredDifferenceKahanSum32f.
  • SSSE3 optimization of function GaussianBlur3x3.
Bug fixing
  • Loading of unaligned address as aligned in SSE2 and AVX2 optimizations of function AbsDifferenceSumMasked.
  • A bug in Base implementation of function FillBgr.
Improving
  • Improving of VMX optimization of function ReduceGray2x2.
  • Loop unrolling of VMX optimization of function AbsDifferenceSums3x3.
  • Loop unrolling of VMX optimization of function AbsGradientSaturatedSum.
  • Improving of AVX2 and VMX optimizations of function ReduceGray4x4.
  • Improving of AVX2 and VMX optimizations of function TextureBoostedSaturatedGradient.
  • Improving of AVX2 optimization of function Laplace.
  • Improving of AVX2 optimization of function SobelDx.
  • Improving of AVX2 optimization of function SobelDy.
  • Improving of SSSE3 and AVX2 optimizations of function LaplaceAbs.
  • Improving of SSSE3 and AVX2 optimizations of function LaplaceAbsSum.
  • Improving of SSSE3 and AVX2 optimizations of function SobelDxAbs.
  • Improving of SSSE3 and AVX2 optimizations of function SobelDxAbsSum.
  • Improving of SSSE3 and AVX2 optimizations of function SobelDyAbs.
  • Improving of SSSE3 and AVX2 optimizations of function SobelDyAbsSum.
  • Improving of SSSE3 and AVX2 optimizations of function ContourMetrics.
  • Improving of SSSE3 and AVX2 optimizations of function ContourMetricsMasked.
  • Improving of AVX2 optimization of function SquaredDifferenceSum.
  • Improving of AVX2 optimization of function SquaredDifferenceSumMasked.
  • Improving of SSSE3 and AVX2 optimizations of function ResizeBilinear for 8-bit gray images.
  • Improving of AVX2 optimization of function GaussianBlur3x3.

Test framework

New features
  • Tests for verifying functionality of function NormalizeHistogram.
  • Tests for verifying functionality of function SquaredDifferenceKahanSum32f.

Infrastructure

New features
  • Project files for Visual Studio 2015.
Home

June 1, 2015 (version 2.3.30.503)

Algorithms

New features
  • Base implementation, SSE2, AVX2 and VMX optimizations of function Laplace.
  • Base implementation, SSSE3, AVX2 and VMX optimizations of function LaplaceAbs.
  • Base implementation, SSSE3, AVX2 and VMX optimizations of function LaplaceAbsSum.
Bug fixing
  • A GCC compiler error in function Simd::Sse2::ResizeBilinear.
Improving
  • Loop unrolling of SSE, AVX and VSX optimizations of function SquaredDifferenceSum32f.
  • Loop unrolling of VSX optimization of function SvmSumLinear.
  • Loop unrolling of VMX optimization of function SquaredDifferenceSum.
  • Loop unrolling of VMX optimization of function SquaredDifferenceSumMasked.
  • Loop unrolling of VMX optimization of function AbsDifferenceSum.
  • Loop unrolling of VMX optimization of function AbsDifferenceSumMasked.
  • Loop unrolling of VMX optimization of function ValueSum.
  • Loop unrolling of VMX optimization of function SquareSum.
  • Loop unrolling of VMX optimization of function CorrelationSum.
  • Loop unrolling of VMX optimization of function ConditionalCount8u.
  • Loop unrolling of VMX optimization of function ConditionalCount16i.
  • Loop unrolling of VMX optimization of function ConditionalSum.
  • Loop unrolling of VMX optimization of function ConditionalSquareSum.
  • Loop unrolling of VMX optimization of function ConditionalSquareGradientSum.
  • Loop unrolling of VMX optimization of function GetStatistic.
  • Loop unrolling of VMX optimization of function GetRowSums.
  • Loop unrolling of VMX optimization of function GetAbsDyRowSums.
  • Loop unrolling of VMX optimization of function OperationBinary8u.
  • Loop unrolling of VMX optimization of function VectorProduct.
  • Loop unrolling of VMX optimization of function AddFeatureDifference.
  • Loop unrolling of SSE2, AVX2 and VMX optimizations of function AbsSecondDerivativeHistogram.

Test framework

New features
  • Common performance measurement.
  • Multi threading performance testing.
  • Tests for verifying functionality of function Laplace.
  • Tests for verifying functionality of function LaplaceAbs.
  • Tests for verifying functionality of function LaplaceAbsSum.
Bug fixing
  • A GCC compiler warning in function Test::Data::CreatePath.

Documentation

Bug fixing
  • The wrong classification of the function GaussianBlur3x3.
  • A description of function Laplace.
  • A description of function LaplaceAbs.
  • A description of function LaplaceAbsSum.
Home

May 4, 2015 (version 2.3.29.469)

Algorithms

New features
  • Base implementation, SSE, AVX and VSX optimizations of function SvmSumLinear.
  • SSE2, SSSE3, AVX2 and VMX optimizations of function ResizeBilinear for Uv16, Bgr24 and Bgra32 images.
  • Base implementation, SSSE3, AVX2 and VMX optimizations of function Yuv422pToBgr.
  • Base implementation, SSE2, AVX2 and VMX optimizations of function Yuv422pToBgra.
  • Runtime detection of VMX and VSX extensions for PowerPC.
  • Base implementation, SSE2, SSSE3, AVX2 and VMX optimizations of function BgraToYuv422p.
  • Base implementation, SSSE3, AVX2 and VMX optimizations of function BgrToYuv422p.
Bug fixing
  • A potential program crash in function SimdSquaredDifferenceSum32f.
  • Removing of redundant inclusion of header files.

Test framework

New features
  • Tests for verifying functionality of function SvmSumLinear.
  • Tests for verifying functionality of function Yuv422pToBgr.
  • Tests for verifying functionality of function Yuv422pToBgra.
  • Tests for verifying functionality of function BgraToYuv422p.
  • Tests for verifying functionality of function BgrToYuv422p.
  • Joint declaration and execution of auto- and data-tests.

Documentation

New features
  • A description of function SvmSumLinear.
  • A description of function Yuv422pToBgr.
  • A description of function Yuv422pToBgra.
  • A description of function BgraToYuv422p.
  • A description of function BgrToYuv422p.
  • Expanding of the library overview.
Bug fixing
  • A small inaccuracy in description of SimdShiftBilinear.
  • Incorrect insert of pseudocode in function's descriptions.

Infrastructure

New features
  • Vmx project for PowerPC VMX(AltiVec) extension.
Home

April 2, 2015 (version 2.2.28.451)

Algorithms

New features
  • Base implementation, SSE2, AVX2 and VSX optimization of function CorrelationSum.
  • Functions Avx::Square and Avx::Combine.
Bug fixing
  • A program crash in function SimdShiftBilinear.

Test framework

New features
  • Tests for verifying functionality of function CorrelationSum.

Documentation

New features
  • A description of function CorrelationSum.
  • A reference to the project at GitHub.

Infrastructure

New features
  • This project was added to GitHub.
Changes
  • A support of non SVN repositories was added to GetVersion.cmd and GetVersion.sh scripts.
Bug fixing
  • An error for x86/x64 in CMakeLists.txt.
Home

March 2, 2015 (version 2.2.27.443)

Algorithms

New features
  • An operator of division of rectangle on scalar value.
  • SIMD_ALLOCATE_ASSERT macro.
  • Base implementation, SSSE3, AVX2 and VSX optimization of function SobelDxAbsSum.
  • Base implementation, SSSE3, AVX2 and VSX optimization of function SobelDyAbsSum.
  • Base implementation, SSE2, AVX2 and VSX optimization of function HogDirectionHistograms.
Bug fixing
  • Function SimdAllocate() returns NULL if align is less then pointer size.

Test framework

New features
  • Tests for verifying functionality of function SobelDxAbsSum.
  • Tests for verifying functionality of function SobelDyAbsSum.
  • Tests for verifying functionality of function HogDirectionHistograms.

Documentation

New features
  • A description of function SobelDxAbsSum.
  • A description of function SobelDyAbsSum.
  • A description of function HogDirectionHistograms.
Home

February 4, 2015 (version 2.2.26.436)

Algorithms

New features
  • C API functions for memory management: SimdAllocate, SimdFree, SimdAlign and SimdAlignment.
  • Methods Allocator::Align and Allocator::Alignment.
Changes
  • An extension of files with C++ API classes was changed from *.h to *.hpp.
  • SimdUtils.h was renamed to SimdLib.hpp.
  • A C++ API was made independent from internal implementation of Simd Library.
  • A content of SimdTypes.h was moved to SimdLib.h.
Bug fixing
  • Wrong using of "bool" type in C API.
  • Error in function Allocate for PowerPC.

Test framework

New features
  • A file TestCheckC.c for checking of C API.
  • A file TestCheckCpp.c for checking of C++ API.

Documentation

New features
  • An on-line and off-line documentation generated by Doxygen.
  • A separate file SimdHelp.h with the service information required to generate a reference documentation.
  • A description of Allocator structure.
  • A description of Point structure and related functions.
  • A description of Rectangle structure and related functions.
  • A description of View structure and related functions.
  • A description of Pixel::Bgr24 structure.
  • A description of Pixel::Bgra32 structure.
  • A description of Pixel::Hsv24 structure.
  • A description of Pixel::Hsl24 structure.
  • A description for all C++ wrapper functions.
Home

January 5, 2015 (version 2.1.25.419)

Algorithms

New features
  • Pixel::Bgr24 structure.
  • Pixel::Bgra32 structure.
  • Pixel::Hsv24 structure.
  • Pixel::Hsl24 structure.
  • Base implementation of function BgrToHsv.
  • Base implementation of function BgrToHsl.
  • Base implementation of function Yuv444pToHsv.
  • Base implementation of function Yuv444pToHsl.
Bug fixing
  • Wrong assert condition in function SegmentationShrinkRegion.

Test framework

New features
  • Tests for verifying functionality of function BgrToHsv.
  • Tests for verifying functionality of function BgrToHsl.
  • Tests for verifying functionality of function Yuv444pToHsv.
  • Tests for verifying functionality of function Yuv444pToHsl.

Infrastructure

Changes
  • A scripts' location.
Home
2017 | 2016 | 2015 | 2014 | 2013