Path: utzoo!attcan!uunet!husc6!bloom-beacon!mit-eddie!uw-beaver!fluke!mce
From: mce@tc.fluke.COM (Brian McElhinney)
Newsgroups: comp.arch
Subject: Re: RISC machines and scoreboarding
Message-ID: <4180@fluke.COM>
Date: 21 Jun 88 15:47:30 GMT
References: <1082@nud.UUCP> <2438@winchester.mips.COM>
Sender: news@tc.fluke.COM
Organization: John Fluke Mfg. Co., Inc., Everett, WA
Lines: 23

In article <2438@winchester.mips.COM> mash@winchester.UUCP (John Mashey) writes:
>Anyway, I think the bottom-line is that full scoreboarding:
>
>	Is not very relevant to the integer part of the CPU.
>
>	Might be useful for long multi-cycle operations (such as FP),
>	maybe
>
>	Might be helpful for compilers that can't do code scheduling,
>	even though it is silly to do high-performance pipelined RISCs
>	without having a good reorganizer.

Register scoreboarding is also useful to software types and other end users of
this complicated compiler technology.  No matter how mature the compiler is, it
is still a large software program and will have bugs.  It may not be worth the
hardware trade offs, but the software costs should also be considered.
Anything that makes the software "safer" seems like a win to me, but I'll admit
I'm biased!


Brian McElhinney
Software Engineer
mce@tc.fluke.com