Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!claris!apple!baum
From: baum@Apple.COM (Allen J. Baum)
Newsgroups: comp.arch
Subject: Re: Branch Delay Annullment
Message-ID: <12211@apple.Apple.COM>
Date: 15 Jun 88 16:53:35 GMT
References: <22065@amdcad.AMD.COM>
Reply-To: baum@apple.UUCP (Allen Baum)
Organization: Apple Computer, Inc.
Lines: 19

[]
>In article <22065@amdcad.AMD.COM> tim@amdcad.AMD.COM (Tim Olson) writes:
>...this says that m88000 branch delay annullment ("squashing") takes
>place if the branch is taken, causing the branch to take 2 cycles.
>
>This is the opposite of what the SPARC annulled branch does -- it
>squashes untaken branches.  Squashing the untaken branches seems more
>effective to me. (..example showing top of loop duplicated in branch shdw)
>
>Any thoughts as to the benefits of annul-taken form?

In the looping case, obviously squash-on-non-taken works. However, in the
standard if-then-else case (which can occur inside a loop remember, so it
can occur even more frequently than loop-closers dynamically) just the
opposite is true. The HP Precision squashes according to the sign of the
branch displacement in conditional branches. Backwards branches (loop
closers) squash on non-taken, forward branches (if-then-else types, typically)
squash on taken. The best of both worlds, I think.

--
{decwrl,hplabs,ihnp4}!nsc!apple!baum		(408)973-3385