Very interesting. I've had quite good results with fork/join but for simply 
processing large arrays in chunks. I've had a quick look at the fork/join 
implementation and it looks like you create a new task for each recursive 
step but I could be wrong. If so, try adding a threshold to determine 
whether the size of the partition is worth forking (e.g. 20,000+). WIthout 
the conditional you will create many short running threads for sorting the 
small partitions. 

On Tuesday, 7 May 2013 23:10:58 UTC+1, Marceli Narcyz wrote:
>
> I benchmarked a few Java parallel compute frameworks with a simple quick 
> sort:
>
> https://github.com/marcelinarcyz/quicksortjmh
>
> Can anyone offer any improvement suggestions of better alternatives? I 
> know there are faster sorting algorithms, but I'm interested in the compute 
> framework piece itself. thanks!
>

-- 
You received this message because you are subscribed to the Google Groups "Java 
Posse" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/javaposse?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to