Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!amdcad!rpw3
From: rpw3@amdcad.AMD.COM (Rob Warnock)
Newsgroups: comp.arch
Subject: Re: Register usage
Message-ID: <25664@amdcad.AMD.COM>
Date: 17 May 89 18:18:07 GMT
References: <952@aber-cs.UUCP>
Reply-To: rpw3@amdcad.UUCP (Rob Warnock)
Distribution: eunet,world
Organization: [Consultant] San Mateo, CA
Lines: 30

In article <952@aber-cs.UUCP> pcg@cs.aber.ac.uk (Piercarlo Grandi) writes:
+---------------
| In article <8905110956.AA12655@decwrl.dec.com> neideck@nestvx.dec.com writes:
| > There is another very interesting paper by David comparing register window
| > schemes of varying organization with this global allocation stuff and this
| > seems to suggest that a slightly bigger global register file beats register
| > windows
| The 29k guys will cheer...
+---------------

Not really.

Again, to clear up the slight misunderstanding, by current 29k software
convention [with hardware support from the "gr1+local" address adder],
the Am29000 *is* a (variable-sized) register window machine. The "local"
regs are allocated with a strict stack discipline, which, as Neideck
implies, may not be quite as good as a super-global register allocation
strategy.  However, with much less than 1% of cycles being spent on
register cache spilling/filling (all benchmarks except "Ackerman"),
it's probably good enough. [And of course since the current register
discipline *is* a software convention, nothing prohibits a future
compiler from switching to a super-global allocation strategy...]


Rob Warnock
Systems Architecture Consultant

UUCP:	  {amdcad,fortune,sun}!redwood!rpw3
DDD:	  (415)572-2607
USPS:	  627 26th Ave, San Mateo, CA  94403