This is avectorizedlibm.Elementaryfunctions liketrig, pow,asinh can beevaluated usingSIMDinstructions.It is portable,and currentlysupports SSE2,AVX, AVX2,FMA4, AVX512Fand AArch64Advanced SIMD.The library iscoded without aconditionalbranch. It isextensivelytested to checkif it handlescorner caseswell.
Vc is a freesoftwarelibrary to easeexplicitvectorizationof C++ code. Ithas anintuitive APIand providesportabilitybetweendifferentcompilers andcompilerversions aswell asportabilitybetweendifferentvectorinstructionsets. Thus anapplicationwritten with Vccan be compiledfor: * AVX andAVX2 * SSE2 upto SSE4.2 orSSE4a * Scalar* MIC
ConnectedComponentLabeling (CCL)is a well knowntechnique forassigning aunique label toeach ofconnectedcomponents in agiven binaryimage. In thisproject,libraries thatimplement aparallel CCLalgorithm thatis suitable forexecution onGPGPU areprovided. Inthe archive, animplementationin OpenCL and Cthat is usablein combinationwith OpenCVlibrary, asingle-threadedvectorized CPUimplementationthat utilizesAVX2instructions,and a Javaimplementationthat is forhelping unde
dispy is aPythonframework forparallelcomputation. Itdistributestasks acrossmultipleprocessors in asingle machine(SMP), othermachine nodesin a cluster,grid or cloud.It'ssuitable forlarge sets ofdata beingworked on inparallel(SIMD). It doesnot facilitatecommunicationor sharingamong executiontasks. Itutilizesasynchronoussockets,coroutines,epoll, kqueueor WindowsIOCP.