This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tm_gccmail@mail.kloo.net: Re: gcc for any microcontroller?]
- From: Crispin Cowan <crispin at wirex dot com>
- To: danecek at ucl dot cas dot cz, gcc at gcc dot gnu dot org, tm_gccmail at mail dot kloo dot net
- Date: Mon, 24 Mar 2003 17:01:19 -0800
- Subject: Re: [tm_gccmail@mail.kloo.net: Re: gcc for any microcontroller?]
- Organization: WireX Communications, Inc.
- References: <20030324234249.GA4237@wirex.net>
Toshi wrote:
o Lack of standard addressing modes
Some processors have no or very few bits of displacement on the indirect
addressing mode, which causes gcc problems. Unfortunately, some people
are publishing papers which say "you don't need displacement addressing
modes because they can be easily simulated" (Crispin Cowan, etc) and
some processors have actually been built without displacement addressing
modes. This problem complicates many optimizations including CSE, loop
optimizations, instruction scheduling, etc.
Applies to: NIW architecture, IA64, etc.
Ok, I assume you are referring to this paper
http://216.239.39.100/search?q=cache:YcszBDIK49oC:www.dcs.ed.ac.uk/home/cxs/Papers/report413.ps.gz+crispin+cowan++displacement+addressing&hl=en&ie=UTF-8
Let me correct a few misconceptions:
* The paper does *not* say that "you don't need displacement
addressing because it can be simulated." The paper describes a
cute hack that attempts to expose parallelism in a narrow
instruction word by encoding an ALU and a load/store op in the
same 32-bit word. To make that work, we had to employ the ugly
kludge of losing displacement addressing on the load/store part.
* It is *not* a general claim that you can simulate displacement
addressing with arithmetic at no cost. It is only a specific claim
that you can get away with this in the context of the proposed NIW
architecture because you have available slots to compute the
arithmetic.
* The whole thing was a hair-brained scheme that was IMHO rightly
rejected by peer review when I attempted to publish it (I wrote it
10 years ago as a side-line while I was a young graduate student)
It was never actually published anywhere other than as a tech
report. Peer review matters, and it works, even when it is my
paper being rejected :-)
Thanks,
Crispin
--
Crispin Cowan, Ph.D. http://wirex.com/~crispin/
Chief Scientist, WireX http://wirex.com
HP/Trend Micro Immunix Secured Solutions
http://h18000.www1.hp.com/products/servers/solutions/iis/
Just say ".Nyet"