Subject: Re: Is gcc slow? Or is our gcc slow?
To: None <firstname.lastname@example.org>
From: Niklas Hallqvist <email@example.com>
Date: 04/10/1996 09:30:42
>>>>> "David" == David Gilbert <firstname.lastname@example.org> writes:
David> This all reminded me that I wanted to ask if anyone thought it
David> would be possible to port NetBSD to a 68010-based architecture.
David> I have access to a few copies of a 'Sylogics' computer. They
David> have a 68010, 4 serial ports, some form of early array
David> processor (some sort of DSP?), 16 meg of memory, an ethernet
David> board and an ST-506 disk controller.
David> When they were sold, they were sold with some form of UN*X.
David> Some of this, I'm guessing at, but most of the guesses seem
David> However, I don't know if the 68010 has the abilities that
David> NetBSD needs in a chip --- I know that the 68000 doesn't. I
David> also don't have a lot of documentation on the machines ---
David> besides the fact that all the chips seem rather stock (16450
David> serial chips, for instance).
Well the '010 does not have a MMU, but several vendors added custom
MMU HW, do you think Sylogics might have done that too? And then you
just need to understand hat MMU :-) The '010 had restartable
instructions as oppposed to the 68000 so it was easier to make
MMU-requiring OSes work for that CPU. Mind you it is *not* impossible
to port NetBSD to a straight 68000, you just need two of them running
slightly out of sync, so when a trap occurs the other CPU can be used
to restart the failing instruction. Cute, eh :-) Yes, there existed
such machines, I think Apollo did one, possibly Sun too..
Now an Apple ][ port would be harder I guess :-)