Path: utzoo!news-server.csri.toronto.edu!rutgers!usc!rpi!crdgw1!crdos1!davidsen
From: davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr)
Newsgroups: comp.arch
Subject: Re: register save
Message-ID: <3238@crdos1.crd.ge.COM>
Date: 11 Mar 91 16:05:19 GMT
References: <3219@crdos1.crd.ge.COM> <12234@pt.cs.cmu.edu> <1991Mar07.014907.11081@kithrup.COM> <7613@mentor.cc.purdue.edu>
Reply-To: davidsen@crdos1.crd.ge.com (bill davidsen)
Organization: GE Corp R&D Center, Schenectady NY
Lines: 26

In article <7613@mentor.cc.purdue.edu> hrubin@pop.stat.purdue.edu (Herman Rubin) writes:

| I do not think our current instructions are complex enough, in many situations.

  I'm sure that will make it into other quote files besides mine... but
I know what you mean. To justify a complex (microcode) instruction, I
believe you have to satisfy the following:

	/- be faster than doing the same thing in multiple instructions
       <                - OR -
	\- be as fast and smaller than the hardcode to save space

	- AND -

	not use chip space which could be used to make the *overall*
	performance better by doing somthing else.

You can sum this up as "best use of the real estate" or "best overall
performance" and many other people have done so. I have no attachment to
hardcoded instructions, I just want top performance overall, without any
horible "worst case" programs which run orders of magnitude slower than
you would expect.
-- 
bill davidsen	(davidsen@crdos1.crd.GE.COM -or- uunet!crdgw1!crdos1!davidsen)
        "Most of the VAX instructions are in microcode,
         but halt and no-op are in hardware for efficiency"