[r-t] Big searches

Mark Davies mark at snowtiger.net
Thu Dec 27 11:51:44 UTC 2007


Mike Ovenden writes,

> It took several months to complete (and I had to give Cathy a way of
> stopping it when she wanted to use the machine), but sadly I have no idea
> how many nodes.

That's fascinating, cheers Mike. What OS were you running under - Windows? I
found even on a single-core machine it works well if you put the process
priority down to "Below Normal"; you'll soak up spare cpu cycles nicely, but
the search won't stop foreground programs running, nor interfere with
background threads which some programs use to support the UI. On a dual-core
box you don't notice even if the search process priority is high.

If you've still got the box, you could make an estimate of total nodes
searched by letting it run for a day and seeing how many it has got through,
then scaling up to however many months it was.

My search is currently on hour 653 at 57% complete, with 156 trillion nodes
searched. How exciting! Disappointingly, no better compositions have been
found than the one discovered about 0.1% of the way in.

Now I think on it, I'm pretty sure Mr Grimmett has run (may still be 
running) some big Stedman searches. Can you give us any stats Clarrie?

MBD




More information about the ringing-theory mailing list