Terry Lambert wrote:
> Things tend to change considerably when you close in on the
> physical RAM approaching the physical address space in size;
> historically all the assumptions have been that this would not
> be the case. While there's some benefit to rexamining some of
> these assumptions, going to a 64 bit address space with IA64
> and Hammer architectures is just going to reset the assumptions
> back down.
Let me back-track a little here. It might be worthwhile to do
this for code pages for application software, if you end up
running a lot of instances of a single program image (as opposed
to "a single program, different images". Arguably, you should
maybe be using threads for that; however, it could be a win in
the case in the "Subject:" line.
I think in the case that spawned this thread that most of the
httpd's are not running the same vnode object (either a jail
local copy or a read-only nullfs mount yields a different
vm_object_t), so it wouldn't help there, but for a single
installation running a lot of copies of one program, it's more
likely to be helpful (e.g. "one big mail server" or "one big web
server").
-- Terry
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message