Re: swap partition too small

[Tod McQuillin <devin@example.com>]

On Wed, 12 Jul 2000, Christopher Sekiya wrote:

> This does nasty things to performance, but ...
>
> [ description of swapfile ]

Is it really that bad?  On most OS's you just add a level of indirection
through the vnode (or similar) layer, which is probably at most 200 extra
machine instructions of overhead per swap access.

How do those 200 instructions compare with the time spent waiting for the
page to come back from disk?  I'd bet it's less than 1%.

(200 instructions on a 100MHz machine is 2 microseconds (glossing over
many details but the order of magnitude is right) -- disk access is
5-15ms)
-- 
Tod McQuillin