I am not sure I like their approach, but as you
have already figured out, there is some advantage
to throwing memory at it.

Speaking somewhat out of a little ignorance, isn't
there a BUFPCT kernel tunable that instructs how
much system memory to assign to the disk buffer
cache?

What would be cool is to be able to dynamically 
change this on demand.


Andy

> -----Original Message-----
> From: mirian@cosmic.com [mailto:mirian@cosmic.com]
> Sent: Wednesday, April 12, 2000 11:01 AM
> To: tech-kern@netbsd.org
> Subject: disk caching
> 
> 
> So I've noticed that linux seems to be able to fudge a heck of a lot
> of disk io performance from just throwing huge amounts of cache at the
> problem.  Although inelegant, this approach really does seem to move
> disk io right along.  I'm wondering if the same effect can be 
> duplicated
> in NetBSD, or if NetBSD has an even better approach to the problem.
> 
> Currently linux does have measurably better filesystem 
> performance than
> NetBSD (even when fses are mounted asynch/noatime) at least according
> to hbench (I measured NetBSD 1.4.2 against linux 2.2.14), so this is
> more than idle curiosity.
> 
> cheers,
> --Mirian
>