Path: utzoo!attcan!uunet!lll-winken!lll-tis!mordor!joyce!ames!mailrus!tut.cis.ohio-state.edu!rutgers!bellcore!faline!thumper!ulysses!andante!alice!debra
From: debra@alice.UUCP (Paul De Bra)
Newsgroups: comp.unix.xenix
Subject: Re: Size of kernels
Message-ID: <8192@alice.UUCP>
Date: 12 Sep 88 20:32:04 GMT
References: <gX-wbmB38k-082XJsY@andrew.cmu.edu>
Reply-To: debra@alice.UUCP ()
Organization: University of Antwerp
Lines: 21

In his article, Jay Mathew Libove writes:
>Someone mumbled in a question about getting big programs to run in
>small memory that their system has 640K and has a maximum user
>process size of 300K or so. That implies only 340K to the kernel.

>That person claimed SCO Xenix 2.2.1. I use SCO Xenix 2.2.1. My system
>has 2 megabytes of memory, and 1400K is available as my maximum user
>process size.

The Xenix kernel is actually smaller than 340 K, but it dynamically allocates
the I/O buffer pool depending on your total memory size. You can configure
this (/usr/sys/conf/xenix.conf) as described in the documentation. The default
maximum is 400k which is probably reached when you have about 2 Mbytes of
memory. But you can configure it to go up to about 1Mbyte. The middle model
kernel imposes a problem beyond that (on the 286). However, Xenix does start
to behave strangely with such large buffers.
Example: Loading Tex (around 800k) takes about 7 seconds on my AT, 2 second
with a ramdisk, but loading it a second time if it is in the 1Mbyte disk
buffer pool takes 30 seconds. I have found 512K to be a reasonable size.

Paul.