Path: utzoo!attcan!uunet!husc6!bloom-beacon!mit-eddie!ll-xn!ames!elroy!cit-vax!ucla-cs!minnie!kennel
From: kennel@minnie.cognet.ucla.edu (Matthew Kennel)
Newsgroups: comp.sys.apollo
Subject: Re: Open Dialogue - please summarize
Message-ID: <14021@shemp.CS.UCLA.EDU>
Date: 29 Jun 88 23:49:32 GMT
Sender: news@CS.UCLA.EDU
Organization: none
Lines: 50

I have a question and perhaps a problem.  I'm running Domain/IX on
a DN4000 workstation.

When I'm in BSD4.2 (or some reasonable facsimile thereof) I do a
'ps -v' to look at the memory that my processes take up.

I'm specifically interested in the 'SIZE' field.  It appears that according
to the man page SIZE is the total virtual size of a process, i.e. the amount
of virtual memory the process takes up whether on disk or in real RAM.

Now, I change over to SYS5 and do a 'ps -l'.
According to _this_ man page the 'SZ' field refers to :
"The size, in blocks, of the core image of the process."

But the numbers are ALWAYS the same!
So which is it?

Now here is my problem:
I'm running a BIG lisp image (~11M) on disk.  Whenever I use it, I notice that
its 'SIZE' never goes above some number like 4288, and proceeds to page fault
very heavily.  

There appears to be 2 possibilities:

1) 'SIZE' == virtual memory.  But I can't believe that an 11M saved image
   would only give itself 4M of virtual memory!

2) 'SIZE' = real memory.  I have 8M on my machine and so I would think that I
    could get more than 4M of real memory for lisp.  (BTW no other big processes
    were running, just the normal daemons and a csh)   And I also ran it
    on another machine that was practically identical, except that it had 16M of
    RAM.  Again, SIZE went up to some number around 4000 and I got tons of
    page faults.

    As a test, I wrote a C program that malloc's a whole lot of memory, and
    its SIZE field was way up in the 8000s.

    It appears that somehow LISP limits the amount of real memory it will take up.
    I want to increase that limit!

    I've tried playing around with the LISP variables LISP_DYNAMIC_LIMIT
    and LISP_RESERVED_LIMIT or whatever and these seem to have no effect on
    whatever 'SIZE' is, but only seem to control the partition of LISP's total
    memory among its internal structures.

If you have any ideas, please e-mail to
kennel@cognet.ucla.edu

Much thanks,
Matt Kennel