Subject: Re: 240/260 CPU hacking....
To: Chris Tribo <t1345@hopi.dtcc.edu>
From: John Maier <jmaier@midamerica.net>
List: port-pmax
Date: 03/20/2001 17:58:40
On Tuesday, March 20, 2001 3:42 PM Chris wrote:


> >     b) What is the LSI L1A7242 and L1A7231?  LSI was recently bought by
> > Philips and there is
> >     little information on LSI chips.
>
> As per a recent email of the thread "No heatsinks on 5000/240
> cpu?" those would be the CPU ASIC and the Motherboard ASIC. Definitely
> custom jobs specifically for DEC, not likely you will find much doco on
> either of those.

I figured so much.  It looks as if "upgrades" of an existing daughter-board
is the most plausable solution.

> Since the 5000/2x0 ran both R3x00 and R4x00, my guess is that it's
> 64 on the CPU card and possibly _to_ the motherboard ASIC, but no further
> with the R4x00

Humm, since those LSI chips are ASICs for the MB; I would guess, in light of
current info, that the data/address path is 32bit, which is the LCD for both
the R3000 and R4400 chips.  You can downgrade a 64bit path to a 32bit path
with multiplexing, but 32bit to 64bit...that's a lot different.

>
> > 4) There is no reference to 144pin PGA R3000 packaged CPU anywhere,
anyone
> > know exactly what make and model of R3000 DEC used?
>
> No clue, it was probably also a custom design just for DEC.

No, my guess [for now] is it a design that's no longer in use.  Info on the
R3000 is getting scarce, on MIPS's page, and a lot of MIPS manufactures are
going for the ASIC core and embeded market these days.  The original R4x00
doc refered to the R4400 and R4600 in a PGA form factor and I have a SGI
Indy here, and the R4600, on the daughter card, is a PGA form factor.
>
> > 3) Does the SG-615P chip, on the daughter card, control the CPU
frequency?
> > If so, does it also generate the clock for the main board?
>
> No clue.

It has 20.0000M silked on it and from what I've read the R4x00 can be run at
full CPU or x2, and I think the R3000 (at least the R3081) can do 1x or 2x,
depending on the mode you set it for.  The question is, did DEC multiply it
ahead of time or send the clock straight to the CPU...? <hummmm>

> You might search the pmax mailing list for the thread "LSI
> Logic" which had some past discussions on what a viable CPU replacement
> would be, and perhaps some more tech info that would be useful. I would
> also recomment searching for some keywords there, or going to
> http://www.google.com/bsd

hehe, I went there a did a search on 'schematic decstation' and found a link
to a sync on green ciruit I build some time ago...

> Nonsense, this is probably the only forum, next to
> comp.unix.ultrix and the OpenBSD pmax mailing list. Good luck. I should be
> able to grab the part number and revisions of the CPU if you need them
> from a spare 240 Mobo I have at home.

Actually for cross reference sake, can anyone with a /260 look and tell me
what the part numbers are on the two big non heatsinked LSI chips?  Also if
anyone has gotten up the courage  to crack the heat sink off the MIPS chip
(dead DECstation or just insane curiosity) let me know what numbers lurks
beneath...

BTW if anyone has a(any) old DEAD 240 or 260 daughter boards around, let me
know.  Don't throw them away!  I *may* be asking you to send one my way "in
the name of science."
:-)

If I can get a pin out of the PGA R3000 I'll have my solder sucker out and
poised.

John Maier