Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!rutgers!sri-unix!sri-spam!ames!ucbcad!zen!ucla-cs!frazier
From: frazier@CS.UCLA.EDU
Newsgroups: comp.arch
Subject: Re: register windows
Message-ID: <8758@shemp.UCLA.EDU>
Date: Wed, 21-Oct-87 13:01:38 EDT
Article-I.D.: shemp.8758
Posted: Wed Oct 21 13:01:38 1987
Date-Received: Sat, 24-Oct-87 05:01:15 EDT
References: <201@PT.CS.CMU.EDU> <8801@utzoo.UUCP>
Sender: root@CS.UCLA.EDU
Reply-To: frazier@CS.UCLA.EDU (Greg Frazier)
Organization: UCLA Computer Science Department
Lines: 28
Keywords: register windows, interrupt latency
Summary: Not bad on context swaps, either

In article <8801@utzoo.UUCP> henry@utzoo.UUCP (Henry Spencer) writes:
>                                 Register windows may end up being a net win
>despite context-switch overhead.
>-- 

There isn't that much of a context-switch overhead, either.  Saving
the register windows does not dominate the cost of a
context-switch.  Context switches require hundreds of instructions,
while a pipelined system should be able to save the register file
in about 100-150 instructions.  This cost is more than offset by the
savings incurred in the calls between process swaps, even in a system
which performs the swaps fairly often.  In addition, smart systems 
only restore the "top" window
of a swapped out process.  This prevents filling the register file
with windows which are then have to be saved for further calls, and
incurs no (read little) extra cost if the process, once active,
proceeds to execute a series of returns and the windows have to be
restored after all. 

$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$

Greg Frazier              Your feet are frozen to the floor...  -more-
CS dept., UCLA

Internet: frazier@CS.UCLA.EDU
    UUCP: ...!{ihnp4,ucbvax,sdcrdcf,trwspp,randvax,ism780}!ucla-cs!frazier

$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$-$