I can think of parallel quicksort and mergesort, any other parallel sorting algorithms being implemented? And, what's the performance advantage/tradeoff of implementing them in a parallel way? It seems to me that creating loads of pthreads doing small amount of jobs is a waste and incurs huge overhead, there must be a balance between using serial sorts and parallel sorts. Would be great to see if someone could point me to an implementation of such parallel sorts being efficiently implemented/used. thanks a lot.

http://code.google.com/p/back40computin ... dixSorting

AnimationsNo help I know, but nice pictures.

there's an example of a bitonic sort in gpu gems; thrust is an excellent STL-like CUDA template library with a few examples too

QuoteOriginally posted by: rmaxAnimationsNo help I know, but nice pictures.Amazing

QuoteOriginally posted by: ww250I can think of parallel quicksort and mergesort, any other parallel sorting algorithms being implemented? And, what's the performance advantage/tradeoff of implementing them in a parallel way? It seems to me that creating loads of pthreads doing small amount of jobs is a waste and incurs huge overhead, there must be a balance between using serial sorts and parallel sorts. Would be great to see if someone could point me to an implementation of such parallel sorts being efficiently implemented/used. thanks a lot.Anyone could point me to an efficient quicksort or mergesort parallel implementation?

what platform? Thrust (CUDA) has a merge sort, or you could check this out -> https://wiki.cc.gatech.edu/multicore/in ... Algorithms