Subject: Re: Stupid Question
To: None <macbsd-general@NetBSD.ORG>
From: Christopher M. Hanson <firstname.lastname@example.org>
Date: 11/12/1994 16:29:28
Excerpts from internet.listserv.macbsd-general: 10-Nov-94 Stupid
Question by Henry B. Hotz@libra.lora
> This question may have been answered on macbsd-development, but I'm not on
> that list. I don't mean to criticise (since I love the work done so far),
> but. . .
> What is the big deal about support for fpu-less machines?
*I* don't grok either! "An OS without floating point isn't much fun."
Uh, pardoneme por favor, but isn't the only real FP operation that needs
to be handled by the OS the saving and restoring of the FPU state over
context switches? Otherwise, isn't FPU instruction generation a
function of the *compiler*, where *compiler* switches can be used to
turn it on & off?
It also probably won't be hard to write the code to detect an FPU's
presence on boot and decide whether to do the FPU save/restore at switch
time. One flag, one pair of CMP/BEQ instructions, more machines
Hmpf. This is the same problem 386bsd had at the start, and the
solution (I think) was the elimination of dependency on an FPU by the
kernel & c runtime library.
Who can't help at the moment because he's got a C610, a PB520, and a PM7100...
Chris Hanson, KSC, KTD-2 <email@example.com> <firstname.lastname@example.org>
"It's not my damn planet, monkey-boy!" -- John Worfin, ABBAT8D