On 24 Aug 2008, at 06:31, Thomas M. DuBuisson wrote:
That's really interesting -- I just tried this.
Compiling not using -threaded: 1.289 seconds
Compiling using -threaded, but not running with -N2: 3.403 seconds
Compiling using -threaded, and using -N2: 55.072 seconds
I was hoping to see a relative improvement when introducting an
opportunity parallelism in the program, so I made a version with two
MVars filled at the start. This didn't work out though - perhaps some
performance stands to be gained by improving the GHC scheduler wrt
cpu /
OS thread affinity for the Haskell threads?
For the curious:
-O2: 7.3 seconds (CPU: 99.7% user)
-O2 -threaded: 11.5 seconds (CPU: 95% user, 5% system)
-O2 -threaded ... +RTS -N2: killed after 3 minutes (CPUs: 15% user,
20%
system)
Thats quite a lot of CPU time going to the system.
Specs:
Linux 2.6.26 (Arch) x86_64
Intel Core 2 Duo 2.5GHz
Hmm thanks, that's interesting -- I was think it was probably caused
by OS X, but it appears to happen on Linux too. Could you try running
the old code too, and see if you experience the order of magnitude
slowdown too?
Bob
_______________________________________________
Haskell-Cafe mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/haskell-cafe
