Author: Tom Kerrigan
Date: 06:27:06 01/20/00
Go up one level in this thread
All of the information you are requesting is well-known and posted on many web sites. The Alpha is a 64-bit chip and it has been around since the dawn of time. 1991 or so, if memory serves. Every major RISC chip vendor has made 64-bit versions of their chips. The ones that come to mind are the HP-PA 8000, the UltraSparc, and certain MIPS chips (I think). AMD is expected to release a chip in 2001 that uses a 64-bit version of the x86 instruction set. Intel's 64-bit chip, the Itanium (formerly known as Merced), is an engineering travesty. It was supposed to be released in 1997, but there's no way it will be released before October 2000 (and it might be delayed again). Prototypes were distributed last month, but only a few. For example, Microsoft only has 1. (!!) The prototypes can only run at ~400MHz, too, for reasons that aren't clear. The chip is so huge that it can't even be made with a 0.25 micron process. That means it needs an amazing amount of power, produces an amazing amount of heat, and costs as much as a small island. Compaq has posted a paper explaining why the Alpha is superior to the Itanium: http://www.digital.com/hpc/ref/ref_alpha_ia64.pdf They have me convinced... Anyway, the performance increase you can get from a 64-bit chip depends entirely on how many of your algorithms are 64-bit. -Tom On January 20, 2000 at 08:34:52, leonid wrote: >Hello! > >Where is the new 64 bits chip? > >I waited for the new 64 bits chip in 1999 - it never came. Feel me impatient in >2000 but found no noises about it in the new computer revue that I bought. Can't >wait any more quietly since I expect to rewrite my chess logic for 64 bits >machine and gain a lot of new speed. If you know somehow about the 64 bits chip, >that will be done by Intel or other company, please say what you found about new >chip. > >If you want to say your opinion about what is realistic to expect from the next >64 bits "gold rash", you are welcome. Maybe what should be your expectation that >next chess game will be able to perform in the nodes/second rate. How many plies >new chess game will be able to see in the time between 1 and 10 seconds, if it >will be looking for response by simple brute force logic. Or just whatere you >find interesting to mention about above topic. > >Thanks for responding! >Leonid.
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.