Path: utzoo!utgpu!water!watmath!clyde!bellcore!faline!thumper!ulysses!andante!mit-eddie!bu-cs!purdue!decwrl!labrea!sri-unix!garth!smryan
From: smryan@garth.UUCP
Newsgroups: comp.arch
Subject: Re: RISC machines and scoreboarding
Message-ID: <842@garth.UUCP>
Date: 30 Jun 88 21:52:43 GMT
References: <1082@nud.UUCP> <2438@winchester.mips.COM> <1098@nud.UUCP> <318@mucmot.UUCP> <1639@osiris.UUCP>
Reply-To: smryan@garth.UUCP (Steven Ryan)
Organization: INTERGRAPH (APD) -- Palo Alto, CA
Lines: 18

In article <1639@osiris.UUCP> phil@osiris.UUCP (Philip Kos) writes:
>In article <318@mucmot.UUCP>, ron@mucmot.UUCP (Ron Voss) writes:
>> Are there really *good* reasons to put more trust in RISC optimizers?

Hey! Why not ask somebody who has to write the compiler? Don't you trust us?

No? Smart idea. Seriously, the reason given for delayed branches is that
(1) the compiler can do it, (2) nobody uses assembly anyway.

I would like to think that (1) is true, but compilers have just as many
problems as anybody else. (2) I have not yet heard of any machine in general
production without an assembler. Even if that were the case, because (1) is
false, somebody has to debug the compiler. It is hard enough to backtranslate
assembly through the compiler to the source language on a conveniential
machine.

Given the current ratio of software cost to hardware cost, I feel the more
extreme proposals of RISKs are misguided.