Subject: Re: [boost] [review] [sort] Sort library review manager results
From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2014-11-28 12:45:26
On 28/11/14 13:55, Steven Ross wrote:
> Here are some potential standard
> algorithms that it might be nice to add in the future:
> LSD radix sort, for fixed-length data where stability matters
> Timsort, a stable comparison sort that is fast for mostly-sorted data and
> is used by standard Java implementations.
> A sorting network based comparison sort for small and fixed-size datasets.
> I've found these to be about twice as fast as Insertionsort.
> k-way merge parallel sort
AFAIK, in the parallel world, merge sort and radix sort are the only two
Without the special parallel optimizations though, they tend to
underperform compared to smarter algorithms.
There is room for different algorithms depending on the size of the data
set, the architecture, and the type/size of the input itself.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk