I agree with Roman. Even if it was possible to get a hybrid port to work,
we're looking at so much kludge code to make it fly that the speed
penality to classic m68k would effectively kill the port. If we try and
focus on classic m68k (that is getting m68k ASM working on coldfire),
we'll be hurting the coldfire port, its going to add another layer of
complexity to an already complex system, and TBH, its hard for me to
justify such a move because the focus should be on the newer coldfire vs.
the older m68k.
Two seperate ports would be ideal as we don't have to canalbize one port
in favor of the other , but it doesn't seem the release masters
will allow that to fly. As roman said, politicial issues will always trump
a technincal one.
Michael
On Tue, 4 Mar 2008, Roman Zippel wrote:
Date: Tue, 4 Mar 2008 23:12:19 +0100
From: Roman Zippel <[EMAIL PROTECTED]>
To: debian-68k@lists.debian.org
Cc: Wouter Verhelst <[EMAIL PROTECTED]>,
Finn Thain <[EMAIL PROTECTED]>,
Christian T. Steigies <[EMAIL PROTECTED]>,
Ingo Juergensmann <[EMAIL PROTECTED]>
Subject: Re: FOSDEM thoughts
Resent-Date: Tue, 4 Mar 2008 22:15:33 +0000 (UTC)
Resent-From: debian-68k@lists.debian.org
Hi,
On Tuesday 4. March 2008, Wouter Verhelst wrote:
A hybrid port would just emit two opcodes in such a case, of course.
This will indeed degrade performance slightly.
The incompabilities are all over the place, we would be basically restricted
to a very basic instruction set and due to the floating point differences it
wouldn't even be compatible with the current ABI. I'm not overly optimistic
that a few optimized libraries would save us, e.g. scripting languages are
already slow enough right now.
bye, Roman
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
