News for hackers. Stuff that matters.

threadman

1. Cilk++ has nested parallelism that works and provides guaranteed speed-up. OpenMP has nested parallelism, but it is a memory hog and not reliable.

2. Cilk++ guarantees space bounds. On P processors, Cilk++ uses no more than P times the stack space of a serial execution. In OpenMP, not so.

3. Cilk++ has a race detector for debugging and software release. With OpenMP, you are on your own.

4. Cilk++ has serial semantics. With OpenMP, you do not have this benefit – only a subset of OpenMP supports serial semantics.

5. Cilk++ has a solution for global variables (a construct called "hyperobjects"). OpenMP does not.

sorry, I'm not clear- do you mean you distribute it under GPL2, or that GPL2 allows use without releasing the resulting work?
do you distribute your work under GPL? If so, then you wouldn't see a change under CAPL.
Re:Cilk Programing Language Wins Most Influential Award on Thursday September 11, 2008 @06:02PM (#301820)
Here's the (alpha) documentation and code samples: http://www.cilk.com/resources-for-multicoders/for-developers...
Re:Cilk Programing Language Wins Most Influential Award on Thursday September 11, 2008 @06:01PM (#301819)
Here's the (alpha) documentation and code samples: http://www.cilk.com/resources-for-multicoders/for-developers...
Re:Multicore-enabling the N-Queens Problem Using Cilk++ on Friday September 05, 2008 @08:23AM (#295794)
TBB borrows the scheduling portion, but requires substantial code restructuring, and doesn't have a solution for the global variable problem analogous to Cilk's hyperobjects.
Re:Multicore-enabling the N-Queens Problem Using Cilk++ on Friday September 05, 2008 @01:26AM (#295596)
interesting point - worth a comment in the blog post

Older »