Author: Robert Hyatt
Date: 20:25:00 08/09/02
Go up one level in this thread
On August 09, 2002 at 12:56:45, Gian-Carlo Pascutto wrote: >On August 09, 2002 at 11:46:58, Robert Hyatt wrote: > >>OK.. then this question. One process holds the lock. The O/S happens to >>schedule the _other_ process which is spinning on the lock. That process >>spins for whatever the scheduling time quantum is, typically about .4 seconds. >> >>That represents a .4 second slice of time when _no_ useful work is being done. >>While on a real dual, the second processor would run, clear the lock and the >>first would only spin for microseconds rather than hundreds of milliseconds... > >Oops, I never considered this. Thanks for clearing that up. > >>It just means your estimated speedup is going to be very pessimistic and will >>only be useful as a lower bound, with the actual speedup somewhere above that >>bound. > >Pleasant surprises are always nice ;) > >-- >GCP parallel stuff is neat. The longer you do it, the better you understand it and the related issues. Some in-depth operating system experience also helps as you can see from the above... spin-locks are for dedicated machines. mutex locks that block a process when a lock is already held by another thread is the way to go on non-dedicated or shared machines. I always expect dedicated of course...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.