Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hyperthreading vs. dual configuration performance? Somewhat OT

Author: Matt Taylor

Date: 15:21:23 02/22/03

Go up one level in this thread


On February 22, 2003 at 07:06:53, Robert Pawlak wrote:

>Matt,
>
>I mostly agree that the price differential is not what it used to be. However,
>what will happen in a few months is that people will start to buy the HT enabled
>P4 systems. When that occurs, your typical user won't even know that he is
>getting something akin to a dual system. But eventually, they will become more
>common, and people will have to write code with multiprocessing in mind.
>
>It lowers the bar, by making it the default configuration.
>
>Bob

It lowers the bar, but threading is hard. Threading is quite possibly the
hardest part of programming. It is hard to get right, and it is hard to debug.
Many programmers don't understand threading at all. Furthermore there is extra
work to take advantage of HT, and the gains aren't as impressive as they are for
a real SMP system. Workstations have mingled with SMP for quite some time now,
so the only applications that might be affected would be desktop applications.

Multiprocessing is very compelling for a single application that requires
performance and can execute in parallel. However, it is also compelling for
multiple applications that require performance or applications that are unable
to take advantage of parallelism at this level. Many desktop applications don't
require extreme performance (i.e. word processing, browsing Internet, etc.) The
few that do have already been penetrated by the SMP market.

HT's prevalence might motivate a few people to go ahead & take advantage of SMP.
However, I think the majority of programmers interested have already been
confronted with SMP.

-Matt



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.