From mboxrd@z Thu Jan  1 00:00:00 1970
Date: Mon, 25 Sep 1995 19:08:24 -0400
From: jmk@plan9.att.com jmk@plan9.att.com
Subject: plan9 on a pc w/no fpu
Topicbox-Message-UUID: 272a65be-eac8-11e9-9e20-41e7f4b1d025
Message-ID: <19950925230824.-o48bclY_VXIUYfXVfvLiUZaRO8eEqc8meIOM4f_poo@z>

>>From the archive
	From:	forsyth@plan9.cs.york.ac.uk
	To:	9fans@cse.psu.edu
	Date:	Sat, 2 Sep 1995 17:56:51 -0400
	Subject: re: plan9 on laptop

	>>It is said about compiler that floating point requires hardware fpu. Does
	>>it mean that the system won't work with 486sx/slc, that is a frequent
	>>chip to install in laptops?

	there is no support in the 386/486 kernel for emulation of floating point.
	the 486sx seems to return not-a-number (NaN) for floating-point operations,
	if the processor is not set to trap on floating point operations (and
	plan 9 does not set that bit).

	although the bulk of the system will function quite happily without
	a floating-point unit, there are important exceptions.  sometimes these
	are surprising.  it has been some time since i tried a 486sx on the
	cpu/terminal kernel, but i believe that either 8c or 8l will fail.
	one of the more surprising programs to fail is troff, which now
	uses floating-point (at least, it did in the old release, and i don't
	think that's changed).  less surprising failures include awk.
	the window system and editors weren't a problem.

	...