Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!purdue!mentor.cc.purdue.edu!l.cc.purdue.edu!cik
From: cik@l.cc.purdue.edu (Herman Rubin)
Newsgroups: comp.arch
Subject: Re: Unaligned Accesses (was Re: How to use silicon)
Summary: An example where unaligned is needed
Message-ID: <1189@l.cc.purdue.edu>
Date: 25 Mar 89 13:37:21 GMT
References: <37196@bbn.COM> <1989Mar16.190043.23227@utzoo.uucp> <59@microsoft.UUCP>
Organization: Purdue University Statistics Department
Lines: 25

In article <59@microsoft.UUCP>, w-colinp@microsoft.UUCP (Colin Plumb) writes:
> mlord@bnr-public.UUCP (Mark Lord) wrote:
> > The "break into multiple aligned arrays" approach can work fine as long
> > as one is dealing with arrays (of structures) to begin with.  Unfortunately,
> > I suspect that this is not of much use with dynamically allocated linked
> > lists of structures, which is very likely when we are talking about multiple
> > megabytes of data.

			..........................

Here is an actual example where unaligned is at least highly desirable.  One
has an array of bits; we may assume byte alignment is free.  It is necessary
to take a 24-bit block at a time to the right 24 bits of a 32-bit word.  These
24 bits can be anywhere in the block, but this has to be done repeatedly, and
can be vectorized if desired.  Bits used may not be reused.  It is also
necessary to keep track of pointers, as the input array may need to be
refilled, almost without warning and possibly at an unpredictable time. 
The refill procedure is somewhat costly.

I believe that the penalty for unalignment will be cheaper than anything else
if the necessary hardware is there.
-- 
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907
Phone: (317)494-6054
hrubin@l.cc.purdue.edu (Internet, bitnet, UUCP)