Subject: Re: Why it gives me always different results
From: rpw3@rpw3.org (Rob Warnock)
Date: Fri, 08 Feb 2008 22:46:37 -0600
Newsgroups: comp.lang.lisp
Message-ID: <GY6dnS08v_awrTDanZ2dnUVZ_uTinZ2d@speakeasy.net>
tim <TimJ@internet.com> wrote:
+---------------
| Alan Crowe wrote:
| > Tim Bradshaw <tfb+google@tfeb.org> writes:
| >> Do implementations typically have *any* free tag bits (and if so, why)?
| > 
| > Doesn't the move to 64 bits change this? If I were thinking about
| > designing a new language I would be tempted to say it was 64 bit only,
| > with 48 bit addresses and the luxury of 16 tag bits.
| 
| Hmmmm... we are using about 36 bits now in large systems, and memory sizes
| are going up by roughly 1 bit per 12 months. So 48 bits will run out in
| only 12 years.
+---------------

Sooner than that. Large SGI Altix systems already need 44 physical
address bits to be able to address main system RAM (yes, they can
put ~14 TB of global cache-coherent RAM on their largest ccNUMA boxes).


-Rob

-----
Rob Warnock			<rpw3@rpw3.org>
627 26th Avenue			<URL:http://rpw3.org/>
San Mateo, CA 94403		(650)572-2607