lkml.org 
[lkml]   [1998]   [Jul]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Memory Rusting Effect [re: Linux hostile to poverty]
Hi,

Albert D. Cahalan wrote:
> Dave writes:
> > Shoot me if I'm wrong.. But throwing money at a problem to hide
> > inefficient programming sounds like a very Microsoft thing to do.
> > Isn't this why we're using Linux in the first place?
>
> What is "inefficient programming" for a 386 is often good code on
> a quad-Xeon with 64 GB of RAM. Scalability goes both ways, and it
> costs development effort.

Actually Dave was referring to the not-so-sane behaviour of the MM layer
in 2.1.x kernels, when compared to 2.0.x kernels. I wouldn't ever call
it "inefficient programming", but certainly I think the Memory Rusting
Effect should get fixed before we go for 2.2.0.

Comparing a 386 to a quad-Xeon with 64Gb of RAM has little to do with
this issue, BTW.
>
> The original Linux hardware was rather high-end in 1991.

It wasn't. The 386 was the cheapest widely available processor that
would support virtual memory in 1991. Linux couldn't have been written
on anything less than a 386.

> We could continue the tradition by dropping support for
> anything less than a Pentium II with 64 MB of RAM.

AFAIK that's not the Linux tradition. But I may be wrong.
>
> If you want something bloat-free, Minix is still available.
> You can get it for free now, and it runs in 640 kB or less.

Minix is an OS dedicated to teaching OS design. Comparing it to Linux is
completely misleading and misses the point (old debate).

I would say that this thread is going offtopic. Could we go back to
kernel development, please?

The original point was:
- There is something that Bill Metzenthen called "Memory Rusting
Effect". Bill provided us with many data points.
- It happens in 2.1.x kernels (specially on small memory machines), and
it's measurable. It still happens in 2.1.109.
- It doesn't happen on kernels 2.0.x.
- Since buying/installing more memory in all Linux boxes with < 16 Mb of
RAM around the world seems impractical, how do we fix the kernel so that
it behaves in a sane fashion again?


Regards,
---------------------
Andrew D. Balsa
andrebalsa@altern.org

-
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.altern.org/andrebalsa/doc/lkml-faq.html

\
 
 \ /
  Last update: 2005-03-22 13:43    [W:1.046 / U:0.000 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site