[concurrency-interest] Fork/Join Examples
joe.bowbeer at gmail.com
Fri May 25 00:05:06 EDT 2012
Thanks for the examples! I will certainly check them out.
Regarding a cache for recursive subdivision, search in JCiP and www for
Memoizer and you will find some of our attempts.
On May 24, 2012 3:41 PM, "Dr Heinz M. Kabutz" <heinz at javaspecialists.eu>
> In my latest newsletter, I show two examples how we can solve a problem
> faster using Fork/Join. The first is Fibonacci, but a better recursive
> algorithm than is found in the JavaDocs. However, this sum-of-squares
> algorithm requires us to do multiplication which is particularly bad with
> BigInteger. I thus also implemented the Karatsuba multiplication
> algorithm, including a way to do it with the Fork/Join framework.
> I also implemented a FibonacciCache. I believe that caching should maybe
> be solved in general for the recursive decomposition approach with
> RecursiveTask. In my cache, if one thread tries to do work with which
> another thread is already busy, then he waits until the other thread has
> completed his task. This prevents duplication of work. I bet that this is
> a common problem with recursive decomposition algorithms. Have any of you
> thought of writing a cache that solves this problem in general and then
> making that available in the JDK?
> Also, could we *please* change the example in the RecursiveTask JavaDocs.
> It is embarrassing having an exponential algorithm as an example of how
> you would use RecursiveTask. So if we add 1000x the number of processors,
> we can now solve case n+10?
> Here is a direct link to the newsletter for those who are interested:
> Dr Heinz M. Kabutz (PhD CompSci)
> Author of "The Java(tm) Specialists' Newsletter"
> Sun Java Champion
> IEEE Certified Software Development Professional
> Tel: +30 69 75 595 262
> Skype: kabutz
> Concurrency-interest mailing list
> Concurrency-interest at cs.**oswego.edu <Concurrency-interest at cs.oswego.edu>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Concurrency-interest