RE: [MLUG - DISCUSSION] VR3

["Anthony Benjamin" <EMAIL:PROTECTED>]

Ah, Tom's hardware, one of the most educated and unbiased sources on the
entire whole of the internet... 

Anyways, in both Linux and Win2k/XP, the kernel automatically schedules
multiple threads in the same process to multiple processors. Meaning,
unless you are running some single threaded program (Hello World.c),
more than likely you will see a speed increase of some sort. Not only
that, but if you're engaged in more than one running process (i.e.
compiling Java, checking email, writing some notes) You will see a
performance increase. In addition, many libraries are already inherently
multi-threaded (especially in the land of Microsoft).

So, yes, to see a performance increase in a specific application
requires an application that is multi-threaded. However, to see a
general performance increase, you could be running 10 instances of the
same 1 process, 1 thread program, and see an increase (via OS
scheduling). Proven by the same ridiculous website with the quote "In
fact, anyone who regularly runs several applications in Windows 2000 or
XP at the same time may also benefit from two processors."


-----Original Message-----
From: EMAIL:PROTECTED
[mailto:EMAIL:PROTECTED] On Behalf Of Kmicic
Sent: Thursday, May 02, 2002 8:11 PM
To: EMAIL:PROTECTED
Subject: Re: [MLUG - DISCUSSION] VR3

"Ross, Matt" wrote:

> C) Figuring out that two processors are better than one is simple
logic, 

No, it's not simple logic, it's actually a common misconception.

> for
> proving it, do a simple web search:
> http://www.apple.com/powermac/server/ - How it works on a mac
> http://physics.indiana.edu/~sg/CSE2001/singlenode.html - some
benchmarking
> results involving both single and dual

Correct me if I'm wrong, but in this experiment the higher the number,
the better the performance. Shows that dual Athlons are consistently
behind the same single processor machines on the tested tasks.

Here are some select quotes form Tom's hardware, which is not just any
website off of your random websearch:

"This comparison between the two workstation platforms from AMD and
Intel clearly shows that dual operation does not necessarily mean
increased speed for all applications. Rather, you need software that has
been specially adapted to multiprocessor operation, so that the load is
equally distributed between the two CPUs." 

"Our selected benchmarks, chosen specifically for dual-processing, show
that using two CPUs makes sense only with 3D rendering and MPEG encoding
(MPEG-4, Divx)."

I'll leave it at that - don't really have the time for a quote war. As
far as evidence to support my claims - it can be found in any
distributed computing textbook.

Paul
--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/

--
To unsubscribe, go to http://mlug.missouri.edu/members/edit.php

Archives are available at http://mlug.missouri.edu/list-archives/