Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 08 Jul 1999 07:35:16 -0600
From:      Brett Glass <brett@lariat.org>
To:        Greg Lehey <grog@lemis.com>
Cc:        Phil Regnauld <regnauld@ftf.net>, David Scheidt <dscheidt@enteract.com>, Yann Ramin <atrus@matadore.montereyhigh.com>, freebsd-chat@FreeBSD.ORG
Subject:   Re: Intel's flops (was: IA64)
Message-ID:  <4.2.0.56.19990708073150.03f9d7d0@localhost>
In-Reply-To: <19990708165644.G6035@freebie.lemis.com>
References:  <4.2.0.56.19990708010021.041462e0@localhost> <19990708081844.17503@ns.int.ftf.net> <Pine.BSF.4.10.9907071125170.20161-100000@matadore.montereyhigh.com> <Pine.NEB.3.96.990707120716.77528A-100000@shell-2.enteract. <4.2.0.56.19990707200123.00b36480@localhost> <19990708081844.17503@ns.int.ftf.net> <19990708153710.B6035@freebie.lemis.com> <4.2.0.56.19990708010021.041462e0@localhost>

next in thread | previous in thread | raw e-mail | index | archive | help
At 04:56 PM 7/8/99 +0930, Greg Lehey wrote:

 >My recollection was that the 432 was just plain too slow.  The
>instruction set was not conducive to performance.

Actually, the instruction set was quite good. It was a bit stream,
and sub-instructions that did things like add indexing or indirection
could be added by tagging on a few bits. Much more efficient, in terms
of bus bandwidth, than RISC.

Trouble was, the CPU didn't fit on a single chip, and the chips that
comprised it had very large die areas. Limitations in the logic of
the time made the interface between the chips slower than it should
have been, and there was no way to speed things up via pipelining or
FIFOs; RAM was too dear.

Today, that architecture could be built right. Then, it was too far
ahead of its time.

--Brett



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-chat" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4.2.0.56.19990708073150.03f9d7d0>