Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!ncar!ico!nbires!stan!dworkin
From: dworkin@Solbourne.COM (Dieter Muller)
Newsgroups: comp.arch
Subject: Re: Self-modifying code
Message-ID: <2785@salgado.Solbourne.COM>
Date: 14 Oct 89 04:53:55 GMT
References: <4534@imagen.UUCP> <35636@apple.Apple.COM>
Reply-To: dworkin@Solbourne.com (Dieter Muller)
Organization: Solbourne Computer Inc., Longmont, Colorado
Lines: 40

In article <35636@apple.Apple.COM> baum@apple.UUCP (Allen Baum) writes:
>[]
>>In article <4534@imagen.UUCP> hedley@imagen.UUCP (Hedley Rainnie) writes:
>>
>>[running out of the registers]
>
>Except on the newer DEC-20s, it ran slower out of the register file than 
>out of real memory. oops.

Well, it depends on how you define `real memory'.  Registers are in
`fast memory'.  Quoting from the Hardware Reference Manual, Volume I
-- Central Processor, page 1-24:

	Fast memory times are for referencing as a memory location for
	an operand; a fast memory instruction fetch takes slightly
	more time than a cache access.

And, on the next page, a nice table of access times.  The relevant
entries are:

	Memory		Read/Write

	MA20 Core	.833/.40
	KL10 Fast	.067/.067
	KL10 Cache	.133/.133

Times are in milliseconds for a single word access.  So, if you had
cache installed, and your code happened to get placed in it (at the
whim of the monitor), then yes, code from memory would run faster
than code in registers.

However, if you were a financially-strapped site that decided not to
invest in such new-fangled things like cache, code would be much
faster in the registers.

	Dworkin
-- 
No, I'm this way normally.  Why do you ask?
boulder!stan!dworkin  dworkin%stan@boulder.colorado.edu  dworkin@solbourne.com
Flamer's Hotline: (303) 678-4624 (1000 - 1800 Mountain Time)