From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,a3ca574fc2007430
X-Google-Attributes: gid103376,public
X-Google-Thread: 115aec,f41f1f25333fa601
X-Google-Attributes: gid115aec,public
From: kenner@lab.ultra.nyu.edu (Richard Kenner)
Subject: Re: Ada and Automotive Industry
Date: 1996/11/29
Message-ID: <57mhhf$cji@news.nyu.edu>#1/1
X-Deja-AN: 201392893
references: <57airn$7d0@olcs.olcs.com> <57cvl3$uqc@news.nyu.edu>
 <57jh0l$k1p@news.NetVision.net.il>
organization: New York University Ultracomputer Research Lab
newsgroups: comp.lang.ada,comp.realtime
Date: 1996-11-29T00:00:00+00:00
List-Id: <comp.lang.ada>


In article <57jh0l$k1p@news.NetVision.net.il> Eyal Ben-Avraham <eyal@dsp.co.il> writes:
>I agree with Otto Lind. Alltough gcc's RTL, optimization passes, and local/
>global register allocation are great. The reload phase is inadequate
>for accumulatr based machines (the
>SMALL_REGISTER_CLASSES/CLASS_LIKELY_SPILLED_P
>macros are not enough).

Actually, I would say just the opposite: the reload phase has gotten quite
good over the years in making do with just the barest minimum of 
spill registers.  The problem is that the register allocator's don't do
nearly as well, and, more importantly, can't have deal properly with reload
claiming a register they've used as a spill register.