POV-Ray : Newsgroups : povray.off-topic : Parallel processing : Parallel processing Server Time
3 Sep 2024 13:14:32 EDT (-0400)
  Parallel processing  
From: Invisible
Date: 18 Jan 2011 10:16:30
Message: <4d35aece@news.povray.org>
As best as I can tell, desktop computing seems to have reached an 
impasse where the only way to increase processor performance is to add 
more cores and hope that multi-threaded applications start being 
developed "real soon now".

As far as computer programming is concerned, writing programs which 
aren't single-threaded is a "hard problem". Oh, it depends on the task 
of course. But many programs are just really awkward to write in a way 
that utilises multiple cores.

Part of that is the design of the system, of course. The design worked 
OK when there was only one processor, but having several starts to 
stress the design assumptions. Multiple cores fight over available 
memory bandwidth, unified cache, and cache coherence.

However, I'm beginning to wonder whether the whole concept of "computer" 
fundamentally revolves around sequential execution. I mean, think about 
it: Why do computers exist in the first place? Computers exist because 
humans suck at arithmetic. But why do humans suck at arithmetic?

Make no mistake, a human can do a facial recognition task in split 
seconds that would take many hours for a computer. Humans can resolve 
blurry stereoscopic images into 3D mental maps with an accuracy and 
precision that still makes computer vision experts sick with envy. Watch 
a game of snooker and you'll see humans computing things which are far 
from trivial for a computer to simulate.

In other words, the human brain has significant computational power. And 
it is of course perfectly capable of performing just about any 
arithmetic operation that a computer can perform. It's just not very 
good at it, in general.

So here we have a device, the human brain [or rather, a quite small 
subset of it], which is very good at certain mathematical operations, 
yet fairly bad at certain other mathematical operations. The interesting 
question is: why?

Hypothesis: The human brain is good at parallel operations, and less 
good at significantly serial operations.

Corollary: Computers are designed to perform significantly serial 
operations.

QED.


Post a reply to this message

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