Memory Mangement Problem in 5.1-RELEASE

[Terry Lambert]

Ahmed Al-Hindawi wrote:
> dude, I have a third of my memory free!!

Dude, there's a difference between "free" and "available".

Dude, what makes you think that the swap in use doesn't refer
to pages that are also in main memory, but marked clean because
they've already been written to a backing store, but can be
instantly reactivated in main memory without reading from swap?

Dude, why do you thing FreeBSD is ever reading from swap at all
when you are in this situation, and is not just writing it, in
case you have a demand spike for memory so it can satisfy it
immediately, instead of having to futz around with delaying your
request until it can write the dirty things in main memory to
swap?

Dude, have you ever been through a drive through or even in line
inside at a fast food place like Wendy's, where they send order
takers into the line to proactively take your orders, so that all
you have to do when you get to the front of the line is hand them
your order ticket, instead of stating your order and delaying the
whole line behind you by however long they were already delayed
plus the time it takes you to give them your order?

Dude, have you ever heard of "queueing theory" or have you ever
google'd for "pool retention time" and "latency" in the same
query?

Dude, stop saying "dude".  8-) 8-).

There are good reasons that FreeBSD acts the way it does, and why
it is more responsive under a high load than certain other OS's
written by people who are at best undergraduates and at worst High
school students, and never cracked open a copy of "Knuth's Algorithms"
in their lives, let alone owned one and put it up on their bookshelf.

-- Terry