On Wed, 31 May 2023 at 13:14, Johnny Billquist <bqt%softjar.se@localhost> wrote:
On 2023-05-31 14:59, David Brownlee wrote:
On Tue, 30 May 2023 at 13:04, Johnny Billquist <bqt%softjar.se@localhost> wrote:
On 2023-05-30 13:59, David Brownlee wrote:
That's fine, the changes are all in (and will remain in) trunk (unless
someone comes up with a further way to have NetBSD support VAXen with
2GB of mapped memory :-p)
I don't think we support the newer NVAX page table format, so that would
be quite some work... :-)
Sounds like an interesting project (and a nice complement to the gcc
improvements allowing native compiling again). I Assume simh emulates
the NVAX page table format?
Pretty sure simh does *not* emulates the new page table format
introduced in the NVAX. The NVAX can also run with the old page table
format, so there is no actual need for the new format unless you want to
have more than 512M of physical memory.
But simh actually only emulates a rather narrow subset of VAXen, and
none based on the NVAX as far as I can remember.
Ahh.. it's just the KA49 in the 4000/90 which was NVAX, not the
earlier M60. My mistake.
That makes it more challenging, though finding someone interested in
implementin in simh could make a good collaborating project...
Anyone up for the challenge? :-p
Start with doing it in simh then. You could maybe get away with doing
something like the VAXstation 4000 series for a NVAX based machine, but
I think the only VAXen that actually could have more than 512M was the
7000, which means you'll need XMI and VAXBI implemented, along with
whatever controllers (disk and ethernet at least) to even boot.
Then you can test that with VMS, before we start playing with NetBSD...
It's going to be quite some work...
Definitely. Something to leave on a wishlist until the right
intersection of person and time arrive...