Algorithms with core thread performance analysis. More...

Detailed Description

PerformanceAnalysisAlgorithm::FastMedian

PerformanceAnalysisAlgorithm::FastMAD

PerformanceAnalysisAlgorithm::NoiseMRS

PerformanceAnalysisAlgorithm::Sampling

PerformanceAnalysisAlgorithm::MinMax

PerformanceAnalysisAlgorithm::Sum

PerformanceAnalysisAlgorithm::BiweightMidvariance

PerformanceAnalysisAlgorithm::ColorSpaceConversion

PerformanceAnalysisAlgorithm::GetLightness

PerformanceAnalysisAlgorithm::GetIntensity

PerformanceAnalysisAlgorithm::HistogramGeneration

PerformanceAnalysisAlgorithm::HistogramTransformation

PerformanceAnalysisAlgorithm::PSFFit

PerformanceAnalysisAlgorithm::Convolution

PerformanceAnalysisAlgorithm::MorphologicalMedian

PerformanceAnalysisAlgorithm::SeparableConvolution_Rows

PerformanceAnalysisAlgorithm::SeparableConvolution_Cols

PerformanceAnalysisAlgorithm::FFT2D

PerformanceAnalysisAlgorithm::FFT2D_Real

PerformanceAnalysisAlgorithm::Resample

PerformanceAnalysisAlgorithm::Rotation

PerformanceAnalysisAlgorithm::AstrometricReprojection

PerformanceAnalysisAlgorithm::Render

PerformanceAnalysisAlgorithm::FastRender

PerformanceAnalysisAlgorithm::SeparableConvolutionFasterThanNonseparable

PerformanceAnalysisAlgorithm::FFTConvolutionFasterThanNonseparable

Since version 1.9.3 Lockhart, PixInsight uses adaptive, machine-specific optimal thread execution data for a set of critical image processing and numerical algorithms. These data, generated by specialized microbenchmarks, are used internally by PCL and core routines to dynamically determine the optimal number of parallel execution threads as a function of data types, block sizes, and image dimensions, among other factors.

The thread performance optimization feature can significantly improve the performance of critical image processing tasks, especially on machines with 16 or more logical processors.