Return to BSD News archive
Xref: sserve comp.unix.bsd:12570 comp.os.linux:54487 comp.os.386bsd.misc:901 Newsgroups: comp.unix.bsd,comp.os.linux,comp.os.386bsd.misc Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!spool.mu.edu!caen!hellgate.utah.edu!fcom.cc.utah.edu!cs.weber.edu!terry From: terry@cs.weber.edu (A Wizard of Earth C) Subject: Re: BSD UNIX Message-ID: <1993Sep7.162843.19294@fcom.cc.utah.edu> Sender: news@fcom.cc.utah.edu Organization: Weber State University, Ogden, UT References: <CGD.93Sep3160517@erewhon.CS.Berkeley.EDU> <michaelv.747102277@ponderous.cc.iastate.edu> <CCu0s1.29o@ssesco.com> Date: Tue, 7 Sep 93 16:28:43 GMT Lines: 46 In article <CCu0s1.29o@ssesco.com> rhealey@gorp.ssesco.com (Rob Healey) writes: [ ... About NetBSD on the Amiga ... ] > Hey now! B^). I'll pit the 3000's 0.9 I/O performance against any > 386 or 486 based system any day! > > I hear the screen/console code doesn't take full advantage of > the hardware yet tho. And there is the issue of shared, or > better yet dynamic, librarys. > > I run SVR4 on my Amiga tho so I'm not 100% up to date on 0.9's > status. The main problem I have with the Amiga version right now is that it requires a match coproceser and anMU with all the instructions. This means that an 030 is needed, or an 020 with a lot of extra chips, and that the 040 can't be used until the memory management has been rewritten (both a memory management rewrite and a software math coprocesser are reportedly works in progress, to be fair). The screen/console differences are really another good argument for a hardware abstraction layer for the video, keyboard, and mouse I/O; it would be best if an X server for *BSD only needed a recompile for each processer type, and a 680x0 binary should run unchanged on MACs, Amigas, etc. The difference between shared and dynamic libraries boils down to when the offset table is filled out. While load tends to be faster with dynamic, you do take run-time hits for initial reference of routines with dynamic binding; I consider this a loss (it *is* a performance hit that should be taken at link time, not runtime -- an application is typically linked once and run many times). Also, depending on implementation, you end up either losing shared text for each fixup (if it is done wrong) or losing shared text for the GOT for the jump-through and reference-through tables which could otherwise be marked as shared pages after initial fixup (thus increasing the per-instance cost of programs like xterm, etc. -- the size of the GOT can be significant for libX + libXt + libc + ...). All in all, I'll probably be buying an Amiga to run BSD some time in the near future, but it won't be tomorrow. Terry Lambert terry@icarus.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.