POV-Ray : Newsgroups : povray.off-topic : RIP Gary Gygax : Re: RIP Gary Gygax Server Time
11 Oct 2024 07:13:31 EDT (-0400)
  Re: RIP Gary Gygax  
From: Darren New
Date: 11 Mar 2008 14:43:11
Message: <47d6e0cf$1@news.povray.org>
Nicolas Alvarez wrote:
> Saying "people don't restart the app all the time, so startup times 
> don't matter" is just like saying "computers are fast nowadays, so 
> optimization doesn't matter".

The first mainframe I worked on had swapping, not paging. The OS was 
organized that you had one page (Job Information Table) that stored all 
the important things about your process - credentials, quotas, etc.

When you got swapped out (rather, when you allocated pages), the OS 
looked for a space on the swap disk that would hold your process. But 
the way it did it was to first find a page for your JIT, calculate how 
many sectors would pass under the heads of the swap disk while it 
rewrote the channel program in the JIT to include the right memory 
locations, and then swapped out the rest of the pages, again rewriting 
the channel program to put them in a time-optimal order.

When you got swapped in, the system read the JIT, allocated the memory, 
and stored into the JIT the instructions for the disk drive as to where 
to load each block of disk into memory, and handed it to the IO 
processor just as the first sector was coming under the head.

Nowadays, you have disk drives that don't even tell you what the layout 
of the sectors on the disk are. And the processor actually takes 
interrupts for every block transferred. It's impossible to figure out 
how many instructions the processor will execute while any given sector 
passes under the heads.

Programmers these days, they just don't know how to write efficient code.

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.