Author: Sune Fischer
Date: 17:35:01 07/14/03
Go up one level in this thread
On July 14, 2003 at 19:48:05, Russell Reagan wrote: Thanks Russell, it's hereby been bookmarked :) Though, as you can understand I've run my own experiments ;) Theory doesn't always replace hands on experience, IMO. I don't remember the exact times, but typedef'ing a move as unsigned int instead of signed, yielded a measurable speed increase. Also I tried some of the piece square and scoring tables as signed ints rather than chars, that ran also faster in some cases. Definitely modern processors like ints. I could run the tests again, was it not because I'm in the midst of tracking a horrible bug, the program is sadly gutted at the moment. -S. >On July 14, 2003 at 19:38:40, Sune Fischer wrote: > >>I once heard that working with byte size variables is slower than native >>integers (32 bit) sizes, so I'm not fully convinced 8 bit casting it that much >>faster(?). > >Hi Sune, > >You might find this site interesting, "C++ Optimization Strategies and >Techniques". Some 8-bit ops are the same speed as 32-bit ops, but some are >slower. > >http://www.tantalon.com/pete/cppopt/main.htm > >There is one section that covers "Relative Costs of Common Programming >Operations". It tells you approximate how long different operations take on >different types (8-bit, 32-bit, signed, unsigned, etc.). Below that table there >is also one regarding converting from one type to another. For instance, it says >converting from an 8-bit value to 32-bit value is 70% of the speed of 32-bit to >32-bit. > >http://www.tantalon.com/pete/cppopt/appendix.htm#AppendixB_RelativeCosts
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.