Author: Bruce Moreland
Date: 09:23:56 05/22/00
Go up one level in this thread
On May 22, 2000 at 11:11:14, ShaktiFire wrote: >On May 22, 2000 at 00:48:39, Paulo Soares wrote: > >>On May 21, 2000 at 19:37:58, Bruce Moreland wrote: >> >>>On May 21, 2000 at 19:05:58, Robert Hyatt wrote: >>> >>>>You are missing my point. Within 5 years, a single microprocessor chip is going >>>>to have more than one cpu. There are already prototypes. Several vendors have >>>>done this already, although none that are "PC" aware... >>>> >>>>But a dual or quad cpu chip is coming. Quicker than you might think. And it >>>>will still be able to run in a palm or whatever, if the computational demands >>>>continue to increase.. >>> >>>I think multiprocessor machines are great, but my question is why are they >>>useful for the average person, given current software? The average person isn't >>>doing more than one CPU intensive thing at once, if they are doing any CPU >>>intensive things, ever. >>> >>>The software has to take advantage of multiple processors so that it can speed >>>up tasks for single-processor humans, and that is a bitch. >>> >>>Aside from chess programs, I don't do anything that is CPU-intensive, except >>>maybe some games, which seem to run fine now on my 550 mhz Intel machine. In >>>fact, everything seems to run fine now. If I have to sit and wait for something >>>it is typically modem bandwidth (56K modem here) or internet lag. >>> >>>So if they are going to be common, why? What is the upside for the typical home >>>user or semi-casual business user? >>> >>>bruce >> >>I am a structural engineer, and I have a Spanish program, called "Cypecad", that >>needs a lot of processing speed and RAM memory. Last week I run a structure of a >>construction that delayed 1.5 hours on my PIII-450 with RAM=192Mb. >> >>Paulo Soares, from Brazil > >I remember several years ago, I needed to do some structural dynamics while >working in the aerospace field. This particular structure was presumed to >be melting and inelastic, so we could not do the typical linear eigenvalue >type structural dynamics. Instead we had to integrate the finite element model >time wise...step by step.... 40 hrs on a Vax machine. Compared to a typical >elastic analysis of maybe 3 minutes. Stories like this make my point. My mother doesn't have to do this. She reads email, surfs, and keeps the books for her golf group. When I worked as a software engineer I could have benefited from a faster compile cycle, but only because I worked on an enormous application. Most of the time the extra CPU would have been idle. The typical person plays one video game at once, which is not to say that on video game can't benefit from two processors in some way, I'm simply pointing out that it's not like you have several of them going at one time. If you find you need extra capacity, Belkin makes a great monitor/keyboard/mouse switch, so you can simply pile up more computers under your desk, without having to add more keyboards, mice, and monitors. I have at least two going at all times, I have one I turn on sometimes, and there's an extra I might be able to use but I'm worried about causing a fire. If I'm in charge of some programmers, fresh ones who don't know what they are doing, I'm scared to death if they find they have more than one processor and permission to use the extras. Every project I worked on (when I had a job) that had more than one thread was a complete disaster, it either ossified or became a biohazard. I don't think the typical application can use more than one processor very effectively. Maybe there is potential in some commonly used libraries, and this can help, but I think that the single thread execution model is perfectly fine for most applications. bruce
This page took 0.01 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.