Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!wuarchive!usc!jarthur!nntp-server.caltech.edu!seismo.gps.caltech.edu!bruce
From: bruce@seismo.gps.caltech.edu (Bruce Worden)
Newsgroups: comp.lang.c
Subject: Re: Crash a RISC machine from user-mode code:
Keywords: RISC
Message-ID: <1990Aug13.053147.11714@laguna.ccsf.caltech.edu>
Date: 13 Aug 90 05:31:47 GMT
References: <49041@seismo.CSS.GOV> <477@demott.COM>
Sender: news@laguna.ccsf.caltech.edu
Organization: California Institute of Technology, CA
Lines: 38

In article <477@demott.COM> kdq@demott.COM (Kevin D. Quitt) writes:
>In article <49041@seismo.CSS.GOV> stead@beno.CSS.GOV (Richard Stead) writes:
>>
>>Do VAX-CISC programmers spend their days branching to random data?
>>  Who could possibly care
>>that a random instruction sequence crashes a risc box?
>
>    You certainly have a colloquial attitude.  Do you, perhaps, work
>only on single-user machines? Just because you can use something without
>breaking doesn't mean it's seriously flawed.  If there is *any* set of
>instructions that, from the user level, can crash a privelege-based
>system, then that system is *broken*. 
>    Again, just because it's broken doesn't mean it can't be useful for
>limited applications - like single user operation.  But when you try to
>stress the system, it's going to fail.

...and this attitude is simply asinine.  To insist that something will 
fail under certain circumstances when it so clearly does not is the 
height of arrogance.  Our installation has been using Sun Sparc 
hardware for a couple of years and this "problem" has never surfaced.  
And this is not a CS department brimming with talented programmers.  None 
of the people writing programs are trained programmers.  The servers here 
are heavily stressed day in and day out running not only the usual collection 
of editors, compiliers, system and network processes, but also running 
numerical models and other user code.  These systems provide excellent 
service for dozens of users; they have most certainly not "failed".  Not 
bad for a "*broken*" system useful only for single user operation.

Wake up.  Risc may not be perfect (is anything?), but it is providing 
price/performance that has never before been available.  Even IBM has 
jumped on the Risc/Unix bandwagon.  Meanwhile, Sun streaks to yet 
another quarter of record sales and profits selling these seriously 
flawed systems.

--------------------------------------------------------------------------
Bruce Worden                            bruce@seismo.gps.caltech.edu
252-21 Seismological Laboratory, Caltech, Pasadena, CA 91125
--------------------------------------------------------------------------