From: Gary Miller (garymiller@starband.net)
Date: Thu May 22 2003 - 17:56:21 MDT
>> The computer assets per AI worker are on the order of $2,000,
>> adjusting for depreciation. Typical CPU is the AMD 2000+, and
>> desktop computers average about one year old, with the power users
>> receiving a new computer each year.
Stephen in your opinion, what percent faster would your workers be able
to complete their jobs if their systems were 10 times faster?
I would assume their productivity would not increase a 1000% also?
I keep hearing people say just because the systems will be faster that
this
will have a major impact on making making FAI occur faster.
My intuition tell me this is not correct, since when I double the power
of my
CPU I am lucky if my productivity increases by more than 6 or 7 percent
and I
Am using my machine for AI research also!
-----Original Message-----
From: owner-sl4@sl4.org [mailto:owner-sl4@sl4.org] On Behalf Of Stephen
Reed
Sent: Thursday, May 22, 2003 5:52 PM
To: sl4@sl4.org
Subject: Re: Failure of AI so far
On Thu, 22 May 2003, Dani Eder wrote:
At Cycorp, we have a staff of approximately 60 workers (aside from
admin) who mainly use white box AMD computers running Linux. We also
have a white box server farm for email, file storage, databases, web
sites and so forth that is very much what one would expect for a
computer software company our size.
The computer assets per AI worker are on the order of $2,000, adjusting
for depreciation. Typical CPU is the AMD 2000+, and desktop computers
average about one year old, with the power users receiving a new
computer each year.
We are able to hire more staff by using commodity computers.
I believe that we are the largest government funded AI organization.
-Steve
> Using my own workplace, the Space Station software
> test lab, as an example of a government funded
> computer lab, we have about $100K of equipment per
> person. So a well-funded AI project with half a
> dozen people (my guess at how large such a project
> would be today) might afford $600K in computers.
>
> This will buy you at present about 2.5 Tflops in
> a commodity cluster type setup. So in the next
> few years I expect to start seeing some real progress
> in AI research because the hardware is starting
> to get powerful enough.
--
===========================================================
Stephen L. Reed phone: 512.342.4036
Cycorp, Suite 100 fax: 512.342.4040
3721 Executive Center Drive email: reed@cyc.com
Austin, TX 78731 web: http://www.cyc.com
download OpenCyc at http://www.opencyc.org
===========================================================
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT