Subject: Re: Is gcc slow? Or is our gcc slow?
To: Olaf Seibert <email@example.com>
From: Peter Svensson <firstname.lastname@example.org>
Date: 04/12/1996 20:33:17
On Fri, 12 Apr 1996, Olaf Seibert wrote:
> der Mouse <mouse@Collatz.McRCIM.McGill.EDU> wrote:
> I have a MicroVAX 2000 running VMS, and VMS has a self-tuning mechanism
> (autogen). And indeed, the first time I used it it tuned itself into an
> unusable state, which made an essential command fail, which in turn made
> it impossible to login. Basically, what autogen was doing was to adapt
> perfectly to a situation of extremely light load, which was the case in
> a time perod before tuning but which resulted in too conservatively set
> system parameters to do anything at all.
But didn't autogen warn you that the system should have run at least a
week or two with normal load prior to using the feedback information?
The feedback mechanism had to be given resonable lower limits after some
os-extensions were installed, notably TCP/IP stacks often required more
of some resources to boot, especially on machines which were previously
trimmed. Minor changes, such as more users, more memory added, more swap
spindels added or removed etc. are handled smoothly.
Hm, I just realised we are dicussing VMS on NetBSD-current, so I will
stop right here.