lkml.org 
[lkml]   [1998]   [Nov]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: Linux-asm (was A patch for linux 2.1.127)
Date
"A week of mondays ago Ragnar Hojland Espinosa wrote:"
>
> > produce a speed increase of at least a factor of 50% _overall_ (_all_ the
> > kernel) before it gets allowed in. 30% faster cuts no ice with me. Machines
> > run 30% faster every two months, practically.
>
> 18 * 30 / 100 = 5.4 months, actually.. :)
>
> I have to agree with you there, except for little snippets of code. If

Thanks for agreeing .. but if this is "Moore's law", you've probably
stated it wrong, just like almost everyone does! Moores law says that
circuit density doubles every 18 months, not that speed doubles every 18
months :-(.

Supposing speed to be proportional to linear dimension (resolution),
hem, hem, cough, that means that speed goes up by 1.414 every 18 months.
So we'd actually have to wait a year for a 30% speedup on that
reckoning.

My opinion is that we are getting faster speedups than that, so that
speed can't only be proportional to mask resolution. We are also seeing
speed improvements due to better materials and design, and possibly
quantum leaps in technologies. Let's see ...

I have been reckoning I've seen a three-four times speed increase every
4 years so far. Whenever I buy a new laptop, in fact. Say that's 4.5
in three 18 month segments, then we are looking at the cube root of 4.5,
which is 1.65, speed increase every 18 months.

So 30% speed increase would amount to 8 months hardware lead in speed.
I don't think that's worth it. I also don't think we ought to rewrite
the microcode of the pentium II to support us better, although I think
that's possible ...

Peter

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:45    [W:0.112 / U:1.100 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site