From owner-freebsd-questions Thu Aug 3 21:25:47 1995 Return-Path: questions-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.11/8.6.6) id VAA15565 for questions-outgoing; Thu, 3 Aug 1995 21:25:47 -0700 Received: from bubba.tribe.com ([205.184.207.7]) by freefall.cdrom.com (8.6.11/8.6.6) with ESMTP id VAA15558 for ; Thu, 3 Aug 1995 21:25:46 -0700 Received: (from archie@localhost) by bubba.tribe.com (8.6.11/8.6.9) id VAA06637 for questions@freebsd.org; Thu, 3 Aug 1995 21:25:14 -0700 From: Archie Cobbs Message-Id: <199508040425.VAA06637@bubba.tribe.com> Subject: Re: 2.0.5 Eager to go into swap To: questions@freebsd.org Date: Thu, 3 Aug 1995 21:25:14 -0700 (PDT) In-Reply-To: <199508032331.AA27711@diamond.sierra.net> from "Jim Howard" at Aug 3, 95 03:34:28 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Content-Length: 1241 Sender: questions-owner@freebsd.org Precedence: bulk Here's my own 1.5 cents data point on this memory hogging issue. I've used Linux a lot and now I'm starting to use FreeBSD a lot, and I can tell you the difference in memory usage is very noticable to me. I'm running the Mach64 server (albeit at 1280x1024) and right now it's using 19 Megs according to the VSIZE of "ps"... that's huge! I have 16 M of on-board RAM plus 41 M of swap, and I still run out of memory. That's ridiculous considering that I don't really do very much... :-) On Linux I almost never run out, and that's with only 16 M of RAM and 32 M of swap. > Maybe that's why nobody wants to deal with this issue--it collapses into a > flame war and nobody can do anything about it anyway! Hmmm.. bad attitude! If Linux can do it right, then so can FreeBSD... or at least there is evidence that maybe FreeBSD can do it better. What if we ported Linux's libraries? Would that help? How hard would it be to replace the existing malloc() with GNU malloc()? What's up with the X server? Is that all because of malloc()? Yow. On a related note, does the kernel ever promise more memory than it can actually deliver? Or do all calls to malloc() reserve swap (at least)? In either case, is this behavior configurable? -Archie