Author: Robert Hyatt
Date: 11:10:08 11/30/97
Go up one level in this thread
On November 30, 1997 at 09:28:41, Kai Lübke wrote: >On November 28, 1997 at 11:28:37, Ed Schröder wrote: > >>I want to be ready for the future. Better leave all the optimizations >>to the compiler then do it yourself manually. Todays compilers are good >>enough in optimizations. Furthermore you have the choice to release the >>same program for multiple processors types: >>- Pentium version; >>- MMX version; >>- Pentium PRO version; >>- Whatever processor the compiler supports. >>Then the install program simply recognizes the processor and installs >>the correct (and tuned) version for that processor. >> >>Other advantages: >>- Programming in C++ goes at least 3 times faster than in ASM; >>- Porting an text based Rebel (Crafty alike) to an Alpha (or whatever) >>could be done in a few days then. >> >>- Ed Schroder - > >Or you could even compile it on a supercomputer (Cray etc.) and then >challenge the human world champion - maybe a worthy successor to >Deep Blue !? >I really like this idea. :-) > >--- >Shep Nice idea, but never happen. :) IE to go fast on a Cray, you have to specifically design the program for the Cray. Long vector registers, 64 bit words, etc. That's why Cray Blitz is so damned fast on that machine and can barely hit 5K nodes per second on a P6/200... it's not designed to run on a memory-starved architecture at all, and relies on vectors (arrays) of things for pattern matching in the eval, in move generation, in exchange evaluations, etc...
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.