Newsgroups: comp.arch
Path: utzoo!henry
From: henry@zoo.toronto.edu (Henry Spencer)
Subject: Re: Optimal Computer Architectures
Message-ID: <1990Nov13.200051.12329@zoo.toronto.edu>
Organization: U of Toronto Zoology
References: <212412@<1990Nov8> <3300209@m.cs.uiuc.edu> <8662@scolex.sco.COM>
Date: Tue, 13 Nov 90 20:00:51 GMT

In article <8662@scolex.sco.COM> seanf@sco.COM (Sean Fagan) writes:
>>SPARC seems to suck the big wazoo
>>on deep recursion, presumably because of its register-windowing
>>design.
>
>More to the point, *any* processor with register windows is not going to do
>too well on deeply recursive code.

Actually, the AMD 29k ought to do all right.  The problem is not register
windows vs. recursion, it is *fixed-size* register windows vs. recursion.
SPARC ends up saving and restoring a lot of registers that aren't actually
in use, because the allocation quantum is 16 registers.  On the 29k the
quantum is 2 registers (1 if you don't care about being floating-point
compatible with the 29050) and it should almost never be necessary to
save and restore unused registers.
-- 
"I don't *want* to be normal!"         | Henry Spencer at U of Toronto Zoology
"Not to worry."                        |  henry@zoo.toronto.edu   utzoo!henry