Path: utzoo!utgpu!attcan!uunet!mcvax!cernvax!hjm
From: hjm@cernvax.UUCP (Hubert Matthews)
Newsgroups: comp.arch
Subject: Re: HW v. SW (was RISC v. CISC --more misconceptions)
Message-ID: <873@cernvax.UUCP>
Date: 9 Nov 88 09:56:13 GMT
References: <866@cernvax.UUCP> <5195@cbmvax.UUCP>
Reply-To: hjm@cernvax.UUCP (Hubert Matthews)
Organization: CERN European Laboratory for Particle Physics, CH-1211 Geneva, Switzerland
Lines: 21

In article <866@cernvax.UUCP>, hjm@cernvax.UUCP (Hubert Matthews) (me!) says:
>
>> The INMOS T800 has an instruction bitrevword, which turns a
>> little-endian word into a big-endian word, effectively doing a
>> reflection in the middle.  Great for FFT shuffle routines.  In
>> software, it takes quite some time.  In hardware it takes just over 
>> 1 microsecond on a 30MHz part.  
>

OK, OK.  So I picked a lousy implementation of the bitrevword operation.
All I was trying to point out is that reversing the bits in a word can be
done quickly in hardware (should be one cycle), whereas as it takes an
order of magnitude longer in software.  INMOS's microcode isn't very good
on this one - I think it must be using 32 shifts to do it - but it they
used some hardware it would be 1 cycle.

I now realise that my example wasn't pedagogically suitable, so please
stop junking my mailbox :-)
-- 

	Hubert Matthews