Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!sri-spam!ames!oliveb!intelca!amd!amdcad!neptune!brian
From: brian@neptune.AMD.COM (Brian McMinn)
Newsgroups: comp.arch
Subject: Re: AM29000 Booleans
Message-ID: <138@neptune.AMD.COM>
Date: Wed, 6-May-87 12:19:32 EDT
Article-I.D.: neptune.138
Posted: Wed May  6 12:19:32 1987
Date-Received: Sat, 9-May-87 01:19:25 EDT
References: <1270@aw.sei.cmu.edu>
Reply-To: brian@neptune.AMD.COM (Brian McMinn)
Followup-To: comp.arch
Organization: Advanced Micro Devices, Inc., Austin, Texas
Lines: 26

In article <1270@aw.sei.cmu.edu> firth@sei.cmu.edu (Robert Firth) writes:
> [regarding Am29000]
>
>  The machine has comparison instructions that
>  yield a Boolean result in a register.  The
>  processor description says that TRUE is
>  represented by a 1 in the MOST significant
>  bit.  Is this a typo?

No, this is not a typo.  All negative numbers are regarded as Boolean
TRUE.  Although this does not correspond with the "C" language
definition of TRUE, it did gain some speed in the hardware.  Since
most booleans are produced by compare instructions (things like
while(i--) to the contrary, but look at JMPFDEC before you complain
too loudly), we could define a boolean to be whatever was convenient.

In a hardware comparator, the "result" of the compare is generated
from the carry out of the two most significant bits.  Rather than
routing the result bit back to the lsb or propogating it to all 32
bits, we just stuffed it into the msb.  Another nice side effect is
that we only need to test ONE bit of the condition register on a
jump.
-- 
	Brian McMinn		1-(512)-462-5389
	Advanced Micro Devices	domainLand:  brian@neptune.AMD.COM
	Austin, Texas		bangLand:   ...!amdcad!neptune!brian