laurie@cybavax.UUCP (Laurie Moseley) (02/20/86)
*** REPLACE THIS LINE WITH YOUR MESSAGE *** Unix running under other operating systems ------------------------------------------ My university is in the process of replacing some central service machines and there are two main groups of users whose needs must be met. 1. People who want Unix for both teaching and research i.e. most mainstream information technology people. 2. People who wish to run large finite-element Fortran jobs, and a variety of packages such as SPSS and graphics. We need to plan for either 2 machines, or one machine which will run two operating systems in parallel or one on top of the other. We have heard many informal reports of two operating systems at the same time on one machine, but so far have not been able to obtain any hard reports of performance. If anyone has experience of Unix on top of, or in parallel with, some other O/S we would be very grateful if you could share it with us. We are particularly interested in performance and response, but other points would be welcome too (e.g. any problems of maintaining software under 2 systems, security ,user-reaction) With thanks in advance Laurie Moseley ------------------------------------------------------------------------- Merci Danke schoen Gracias Grazie Spasibo Diolch yn fawr .... etc. ------------------------------------------------------------------------- -- If you have skill, you have power. If you have power, you need responsibility. Never demonstrate one without the other.
craig@dcl-cs.UUCP (Craig Wylie) (02/21/86)
In article <172@cybavax.UUCP> laurie@cybavax.UUCP (Laurie Moseley) writes: > > Unix running under other operating systems > ------------------------------------------ > > We need to plan for either 2 machines, or one machine which will run >two operating systems in parallel or one on top of the other. We have heard >many informal reports of two operating systems at the same time on one machine, >but so far have not been able to obtain any hard reports of performance. The IBM operating system VM allows multiple operating systems to co-exist on a single machine. I have heard talk that UNIX is available (rumor only) as an OS for VM. Craig. -- UUCP: ...!seismo!mcvax!ukc!dcl-cs!craig| Post: University of Lancaster, DARPA: craig%lancs.comp@ucl-cs | Department of Computing, JANET: craig@uk.ac.lancs.comp | Bailrigg, Lancaster, UK. Phone: +44 524 65201 Ext. 4146 | LA1 4YR Project: Cosmos Distributed Operating Systems Research Group
agn@unh.cs.cmu.edu (Andreas Nowatzyk) (02/23/86)
Take a look at ELXSI systems. They run 3 operating systems in parallel: BSD 4.2 Unix, SYS V Unix and Embos (their naitive OS, which is supposed to become VMS-like). The machine itself is quite interesting: a 64 bit ECL machine that supports multiple CPU's (up to 10, each said to be 5 times faster than a VAX780), multiple I/O channels and a large memory (192 Mbyte, currently based on 64K chips). They claim that there is no significant performance loss by running several OS's and that each has about the same performance. This info is based on a talk by an ELXSI person... It would be nice if an ELXSI site on the Net could comment on these claims. -- Andreas Arpa: agn@vlsi.cs.cmu.edu uucp: ...!seismo!vlsi.cs.cmu.edu!agn
guy@sun.uucp (Guy Harris) (02/23/86)
> I have heard talk that UNIX is available (rumor only) as an OS for VM.
It's called IX/370 and probably runs under VM, since UNIX is not the primary
OS for IBM mainframes. Amdahl also offers UTS, which runs under VM and
native. I'm sure somebody from IBM can give further details.
--
Guy Harris
{ihnp4, decvax, seismo, decwrl, ...}!sun!guy
guy@sun.arpa (yes, really)muth@amdahl.UUCP (John A. Muth) (02/25/86)
In article <1007@dcl-cs.UUCP>, craig@dcl-cs.UUCP (Craig Wylie) writes: > > The IBM operating system VM allows multiple operating systems to > co-exist on a single machine. I have heard talk that UNIX is available > (rumor only) as an OS for VM. > This is no rumor. This is fact. Amdahl sells UTS, a System 5.2 compatable UNIX that runs on 370 hardware. It will run with or without VM. I'm using it right now. AT&T markets UTS under the name of System V/VM. The AT&T product runs under VM only. IBM also sells UNIX for VM. They call it IX/370 (or something like that). I won't go into comparisons between UTS and IX/370 since I'm obviously not an unbiased observer. I am confident though. -- John A. Muth ...!{ihnp4,hplabs,sun,nsc}!amdahl!muth
phil@amdcad.UUCP (Phil Ngai) (02/27/86)
In article <2840@amdahl.UUCP> muth@amdahl.UUCP (John A. Muth) writes: >This is no rumor. This is fact. Amdahl sells UTS, a System 5.2 compatable >UNIX that runs on 370 hardware. It will run with or without VM. >I'm using it right now. AT&T markets UTS under the name of System V/VM. >The AT&T product runs under VM only. Tell us about full duplex ASCII support, please. Can I run vi? At 9600 baud? How many users can be supported? How about ethernet and TCP/IP? -- The Hyundai is faster than speeding molasses! Phil Ngai +1 408 749 5720 UUCP: {ucbvax,decwrl,ihnp4,allegra}!amdcad!phil ARPA: amdcad!phil@decwrl.dec.com
pdg@ihdev.UUCP (P. D. Guthrie) (02/28/86)
In article <10009@amdcad.UUCP> phil@amdcad.UUCP (Phil Ngai) writes: >In article <2840@amdahl.UUCP> muth@amdahl.UUCP (John A. Muth) writes: >>This is no rumor. This is fact. Amdahl sells UTS, a System 5.2 compatable >>UNIX that runs on 370 hardware. It will run with or without VM. >>I'm using it right now. AT&T markets UTS under the name of System V/VM. >>The AT&T product runs under VM only. > We have Amdahl's product here, and I am quoting practical figures, not anything official. >Tell us about full duplex ASCII support, please. Yes, just as in any other Unix. > Can I run vi? No problem. We all run vi and emacs with no problem. > At 9600 baud? Yes. > How many users can be supported? I have seen 190 on, and it is just beginning to get slow. At 150 it is still blindingly fast. > How about ethernet and TCP/IP? Be real -- this is SysV. How about shared memory on your 4.2 machine? My hat is off to Amdahl for this one!! [I have taken net.unix-wizards out of the cross-posting list as it really does not belong there ] -- Paul Guthrie `When the going gets weird, ihnp4!ihdev!pdg The weird turn pro' - H. Thompson
fetrow@entropy.UUCP (David Fetrow) (03/01/86)
VM does support multiple operating systems, we have a machine running
CMS and UNIX (or rather an IBM variant). I am not on that machine as
a UNIX user so cannot say much about it.
PR1MOS (Prime computers) has a UNIX (called Primix) that runs as a subprocess
and, alledgedly, the C compilers under both O.S.'s act identically.
I believe that Primix is currently available and that the IBM UNIX is still
udergoing some testing.
--
- Dave Fetrow
{ ihnp4, fluke, tektronix, uw-june }!uw-beaver!entropy!fetrow :UUCP
entropy!fetrow@uw-june.arpa :ARPA
fetrow@UWALOCKE :BITNET
74175,1724 :Compuservemeissner@dg_rtp.UUCP (03/01/86)
Disclaimer -- I work for Data General (C compiler), but will try to keep hype
out of this discussion.
At Data General, we have two versions of UNIX, one that runs in conjuction with
our properitary OS's (MV/UX), and one that runs native (DG/UX). The fortran
(and C and Pascal) compilers are the same under both systems. I don't have
performance figures, but our fortran is competive with VMS fortran.
The DG gear runs on a propritary architecture with machines from .5MIPS to
5MIPS for uniprocessors, and a 10MIP dual processor (DG/UX, the native UNIX,
currently does not support the dual processor, but is working on it).
MV/UX runs under either AOS/VS or AOS/DVS. MV/UX users can transparently
access AOS/VS files, processes, etc. Similarly, you can invoke MV/UX processes
from the AOS/VS (AOS/DVS) command line interpreter. Since MV/UX is based
on a library that translates UNIX system calls into AOS/VS (AOS/DVS) system
calls, there are some features of UNIX that can't eaisly be translated. The
underlying AOS/VS operating system has been out in the field for 5+ years,
and there are various tools to tune systems to increase performance. MV/UX
is currently based on system V UNIX (the next rev will be system V.2) with
some berkley features (csh, vi). There is limited TCP/IP (ie, you can't write
sockets in user programs yet) ethernet support, as well as X.25 ethernet/sync
support. AOS/DVS is a superset of AOS/VS for a distributed environment that
was announced last november.
DG/UX runs native on the hardware, and supports full system V.2, as well as
most of berkley 4.2. It has only been out for 1 1/2 - 2 years, and so is
still maturing.
In terms of performance, DG/UX seems faster for normal job mixes, though I
suspect that AOS/VS wins in cases like large fortranish finite element
cases, because you have more tuning capability.
Michael Meissner
...{ ihnp4, decvax }!mcnc!rti-sel!dg_rtp!meissnerguy@sun.uucp (Guy Harris) (03/03/86)
> In terms of performance, DG/UX seems faster for normal job mixes, though I > suspect that AOS/VS wins in cases like large fortranish finite element > cases, because you have more tuning capability. Huh? For CPU-bound jobs, the only difference should be in OS overhead. If you have only one process running, UNIX definitely doesn't impose much overhead and I'd be very surprised if AOS/*VS imposed much either. There may be more overhead in context-switching and the like. There may be more overhead in doing the I/O to read or write the data in under UNIX, I dunno. -- Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.arpa (yes, really)
sjl@amdahl.UUCP (Steve Langdon) (03/03/86)
In article <528@ihdev.UUCP> pdg@ihdev.UUCP (55224-P. D. Guthrie) writes: > ... > > How about ethernet and TCP/IP? > Be real -- this is SysV. How about shared memory on your 4.2 machine? UTS certainly is System V, but we do have an Ethernet driver in UTS/580. For TCP/IP you might find that talking to Wollangong would be useful. > > My hat is off to Amdahl for this one!! Thank you, we try to please. -- Stephen J. Langdon ...!{ihnp4,cbosgd,hplabs,sun}!amdahl!sjl [ The article above is not an official statement from any organization in the known universe. ]
mark@umcp-cs.UUCP (Mark Weiser) (03/03/86)
In article <193@dg_rtp.UUCP> meissner@dg_rtp.UUCP (Michael Meissner) writes: > >DG/UX runs native on the hardware, and supports full system V.2, as well as >most of berkley 4.2. It has only been out for 1 1/2 - 2 years, and so is >still maturing. > You could have fooled me. The C.S. Dept. here at maryland (where I work) has an MV/10000 that DG has been trying to get DG/UX running on for the past 6 months. The latest holdup is ethernet-tcp/ip, without which the machine is pretty useless to us. We've had the machine a year and have been able to do no work at all on it. Not so good for an O.S. out for 1.5-2 years. -mark -- Spoken: Mark Weiser ARPA: mark@maryland Phone: +1-301-454-7817 CSNet: mark@umcp-cs UUCP: {seismo,allegra}!umcp-cs!mark USPS: Computer Science Dept., University of Maryland, College Park, MD 20742
gwyn@brl-smoke.ARPA (Doug Gwyn ) (03/03/86)
In article <3276@sun.uucp> guy@sun.uucp (Guy Harris) writes (>): >> I have heard talk that UNIX is available (rumor only) as an OS for VM. >It's called IX/370 and probably runs under VM, since UNIX is not the primary >OS for IBM mainframes. IX/370 indeed runs on VM/SP and can coexist with other IBM OSes (e.g. CMS, MVS) on the same processor. IX/370 is a fully- supported product (which may actually mean something when IBM is the vendor); it is an implementation of UNIX System V, currently somewhere between UNIX System V Release 1 and Release 2, with a few IBM enhancements. Contact IBM for details.
geoff@desint.UUCP (Geoff Kuenning) (03/06/86)
In article <3310@sun.uucp> guy@sun.uucp (Guy Harris) writes: > Huh? For CPU-bound jobs, the only difference should be in OS overhead. If > you have only one process running, UNIX definitely doesn't impose much > overhead and I'd be very surprised if AOS/*VS imposed much either. There > may be more overhead in context-switching and the like. There may be more > overhead in doing the I/O to read or write the data in under UNIX, I dunno. An OS can slow a program by paging poorly (thrashing), and by swapping poorly (which is different from paging poorly). Also, most real large computations are run under multitasking operating systems (you can't afford to run dedicated), so the quality of the CPU scheduler, in terms of absolute overhead, CPU thrashing characteristics, and throughput, matters a bunch. UNIX, of course, falls down on almost every one of these. The exceptions are CPU thrashing (a one-second timeslice *can't* thrash a CPU bigger than a 4004) and absolute overhead (at least in the more recent USG and BSD systems where they got rid of all those linear searches). -- Geoff Kuenning {hplabs,ihnp4}!trwrb!desint!geoff
throopw@dg_rtp.UUCP (Wayne Throop) (03/09/86)
> In terms of performance, DG/UX seems faster for normal job mixes, though I > suspect that AOS/VS wins in cases like large fortranish finite element > cases, because you have more tuning capability. > Huh? For CPU-bound jobs, the only difference should be in OS overhead. I expect that Mike was thinking of "large finite element cases" as being page fault bound, not CPU bound. Otherwise, Guy is quite correct... CPU bound benchmarks execute identically (more or less) on MV/UX (hosted) and DG/UX (native). -- Wayne Throop at Data General, RTP, NC <the-known-world>!mcnc!rti-sel!dg_rtp!throopw