Author: J. Wesley Cleveland
Date: 15:19:34 04/19/02
Go up one level in this thread
On April 19, 2002 at 16:09:01, Robert Hyatt wrote: >On April 19, 2002 at 02:20:45, Michael Williams wrote: > >>On April 18, 2002 at 21:46:43, Robert Hyatt wrote: >> >>>On April 18, 2002 at 16:33:55, Martin Andersen wrote: >>> >>>>On April 18, 2002 at 16:10:02, Sally Weltrop wrote: >>>> >>>>>A Japanese machine records the fastest "floating point" calculation >>>>>speed - over 35 trillion calculations per second. This is five times >>>>>faster than the previous record holder, IBM's ASCI White system. >>>>> >>>>>http://www.processrequest.com/apps/redir.asp?link=XbddafaeBG >>>> >>>>I'm no expert, but I don't think chess programs use floating >>>>point calculations. >>>> >>>>Martin >>> >>> >>>Only because on PC machines, integer math is faster. If FP was faster, >>>we'd all be using that. On some machines, it is faster.. >> >>Would you be so kind as to elaborate on this (fp)? >>I'd really appreciate it, and I'm also pretty ignorant on such matters. >>Thanks in advance. > >Not sure what you mean. Floating Point (FP) is a different way of representing >numbers so that you get an exponent thrown in so that very large/small numbers >can be represented, while with integers everything is in units of 1... > >Most machines do FP in a separate pipeline which means that some FP can proceed >concurrently with other integer operations, which can make using them faster >than doing purely integer calculations. Speaking of this, did you consider using doubles for node counters in crafty instead of 64 bit ints? My tests show it is faster on x86, and would have solved all those problems about declaring and formatting 64 bit ints.
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.