Am Montag 01 März 2010 16:59:54 schrieb Michael Lesniak:
I've (again) a very strange behaviour with the parallel GC and would be glad if someone could either reproduce (and explain) it or provide a solution.
Sorry, can't reproduce it: $ for i in `seq 1 5`; do time ./mlPi 5000 +RTS -N1; done 7.54user 0.02system 0:07.57elapsed 99%CPU 7.53user 0.02system 0:07.56elapsed 99%CPU 7.70user 0.01system 0:07.72elapsed 99%CPU 7.55user 0.02system 0:07.57elapsed 99%CPU 7.57user 0.01system 0:07.58elapsed 99%CPU $ for i in `seq 1 5`; do time ./mlPi 5000 +RTS -N2; done 7.76user 0.02system 0:07.68elapsed 101%CPU 7.69user 0.02system 0:07.63elapsed 101%CPU 7.94user 0.04system 0:07.89elapsed 101%CPU 7.72user 0.02system 0:07.64elapsed 101%CPU 7.68user 0.04system 0:07.62elapsed 101%CPU $ for i in `seq 1 5`; do time ./mlPi 5000 +RTS -N2 -qg; done 7.56user 0.02system 0:07.58elapsed 99%CPU 7.58user 0.00system 0:07.58elapsed 99%CPU 7.55user 0.01system 0:07.57elapsed 100%CPU 7.57user 0.02system 0:07.58elapsed 100%CPU 7.56user 0.00system 0:07.57elapsed 99%CPU Seems your system has a problem synchronising the cores for GC.