To avoid this in future, you can use the strict-concurrency package on hackage, which has some stricter container types, with their strategies. MVars and Chans are the main examples of where a stricter strategy is sometimes useful. kolar:
Yes, that's is. Thanks. My fault - missing wood seeing trees. ;-)
Best regards,
Dusan
Roberto Zunino wrote:
Dusan Kolar wrote:
Dear all,
I've thought the following three (dummy) programs would run some of their parts in parallel (on dual core) if compiled with option threaded (smp). The truth is that only the first one exploits multicore CPU. Why?
h1 <- forkIO $ putMVar v1 $ fibs (n-1)
You are putting an unevaluated thunk in the MVar. Try:
h1 <- forkIO (putMVar v1 $! fibs (n-1))
Zun.
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe