End of Moore's law?

[Mark Woodward]

On 07/10/2010 03:46 PM, Daniel Feenberg wrote:
>
>
> On Sat, 10 Jul 2010, Mark Woodward wrote:
>
>> Moore's law as we have understood it may have stopped providing value
>> for merely buying something new. We have all been doing it for the last
>> 20 or so years.  Every year and a half or two years, you could spend the
>> same amount of money and get double the computer. 20MHZ 80386 a year and
>> a half later was a 33MHZ 80386 with twice as much RAM and hard disk. The
>> trend had been happening consistently for a long time.
>>
>> Now, something is funny. I bought an HP laptop about 2 years ago. 1.9Ghz
>> dual core, 2G RAM (Upgraded to 4G a year later) 160G HD (upgraded to
>> 320G a year later). My total for upgrades was about $120. I went to
>> Microcenter and looked for laptop, the upgrade of the model I currently
>> have had 4G RAM, 500G hard disk, and a 2.1Ghz dual core for about $250
>> more than what I purchased my current laptop.
>>
>> I recognize that it is a "better" machine than my two year old laptop,
>> but it ain't all that much better. $120 upgrades that I made a year ago,
>> make a $1000 new laptop almost senseless.
>>
>> This is bad news in a lot of ways. if I can't justify replacing a 2 year
>> old laptop based on performance improvements, I bet I am not alone. A
>> lot of companies are built on Moore's law and the upgrade treadmill and
>> the ripple effect is doubtless having an effect on just about everyone's
>> employment.
>>
>> Any thoughts?
>
> Economists talk about the "law of diminishing returns", which holds 
> (roughly) that going from 1 mHz to 1000 mHz is a lot more important 
> than going from 1001 mHz to 2000 mHz, so what you are observing isn't 
> unexpected.

Of course you are right, and I applaud you for thinking of something 
that I damn well should have, but didn't. So, on top of computers not 
getting much faster/better every year at the rate in which they did, the 
increases that we do get have less impact because to actually make an 
improvement you need to be much better.


>
> Daniel Feenberg
>
>
>> _______________________________________________
>> Discuss mailing list
>> Discuss-mNDKBlG2WHs at public.gmane.org
>> http://lists.blu.org/mailman/listinfo/discuss
>>