George Solodovnikov <sgp.m...@gmail.com> wrote on 03/18/2015 03:11:23 PM:
> >
> > Hi,
> > In X10 2.5.2 in Random found no more setSeed(Long). Do you have any
> > info, please,
> > or I miss something since 2.5.0?
>
> >Hi,
> >
> > We switched the implementation of x10.util.Random to use the new
> >SplittableRandom algorithm by Steele, Lea, and Flood from OOPSLA'14. As
> >part of that, we got rid of the setSeed method because unlike the prior
> >MersenseTwister implementation, it is now very cheap to allocate a new
> >Random instance if you want to use a different seed.
> >
> > Have you observed that allocating a new Random instance with the new
> >implementation actually has a measurable performance cost?
> >
> >--dave
> Thank you for clarification.
> After simple performance testing -
> - Managed X10 2.5.2 is ~10-15% better then X10 2.5.0. ( ~95-100
> seconds vs ~110-115 )
> (For Karger MinCut algorithm with 200K+ cycles. In each cycle new
> Random or setSeed.
> On Ubuntu 10.04 x64. X10&jdk8 32b)
> Could not definitely tell was the speed up because of the new
> implementation of Random or
> other X10 enhancements.
>
Great! When doing the initial development, I had measured speedups of up
to 40% in a microbenchmark that did nothing but generate random numbers.
The new implementation is significantly faster.
--dave
------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
X10-users mailing list
X10-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/x10-users
