19 Jun
2007
19 Jun
'07
10:30 a.m.
On 6/19/07, woodelf <bfranchuk at jetnet.ab.ca> wrote:
Providing the word length is a multiable of 4 bits
that is.
It sure would be "fun" to debug microcode issues with nine of those
puppies chained together.
-ethan
19 Jun
19 Jun
10:52 a.m.
New subject: AMD2901s (was e: Inside old games machines, was: Re: Simulated CP/M-68K?)
Ethan Dicks wrote:
On 6/19/07, woodelf <bfranchuk at jetnet.ab.ca>
wrote:
> Providing the word length is a multiable of 4 bits that is.
It sure would be "fun" to debug microcode
issues with nine of those
puppies chained together.
I don't see it harder than with 32 or 24 bits.
[36..21][20..17][16..1] seems to me logical way to do it
for basic CPU as that keeps the boards from being to large.
The Floating point hardware that sounds like the hard part.
Since I never played with a PDP-10 I can say what other
fun features you get to play with.
Remember PAL's are your friend. :)
-ethan
10:53 a.m.
New subject: AMD2901s (was e: Inside old games machines, was: Re: Simulated CP/M-68K?)
Ethan Dicks wrote:
There's probably nothing stopping you from using three of them to
implement a 9-bit processor and either just sink the other three bits,
or use them for processor-level data type enforcement or something or
per-instruction relative jumps or something even weirder. It would
probably be a pretty strange architecture.
Peace... Sridhar
Providing the
word length is a multiable of 4 bits that is.
It sure would be "fun" to debug microcode issues with nine of those
puppies chained together. 
21 Jun
21 Jun
10:02 a.m.
Dave McGuire wrote:
On Jun 19, 2007, at 1:30 PM, Ethan Dicks wrote:
Have logic analyzer, will travel..
heh. have pc based simulator, will debug...
:-)
writing code to emulate the lisp machine microcode was not that hard,
but getting it *exactly* right turned out to be difficult, mostly due to
not understanding some of the math pipeline issues. That uses 48 bit
wide microcode and 78181's as alu slices.
if one were planning to make a ks20 out of an fpga, with microcode, I'd
recommend writing a simulator for the microcode first and booting
something in simulation.
then, when you go to do the hdl you'll having something known to work,
at least in one context.
i've run the lisp machine verilog in modelsim up through the first disk
i/o. I would think a pdp-10 would not that much more complex.
(oh boy, i'll live to eat those words :-)
-brad
Providing
the word length is a multiable of 4 bits that is.
It sure would be "fun" to debug microcode issues with nine of those
puppies chained together. 
11:44 a.m.
writing code to emulate the lisp machine microcode was
not that hard,
but getting it *exactly* right turned out to be difficult, [...]
Heh. I have a MicroVAX-II simulator that's sort of on hold. I had
similar issues; there are some undocumented (as far as I can tell)
aspects of the hardware which the ROM POST code depends on. (For
example, the CPU has an instruction stream prefetch buffer, and the
hardware can run out of the prefetch buffer even though the stuff there
is not what would be fetched from those addresses now, under some
circumstances. I had to give the simulator a prefetch buffer. :-/)
Now, I'm stuck dealing with some serial line timing issues....
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
12:36 p.m.
der Mouse wrote:
Heh. I have a MicroVAX-II simulator that's sort of on hold. I had
similar issues; there are some undocumented (as far as I can tell)
aspects of the hardware which the ROM POST code depends on. (For
example, the CPU has an instruction stream prefetch buffer, and the
hardware can run out of the prefetch buffer even though the stuff
there is not what would be fetched from those addresses now, under
some circumstances. I had to give the simulator a prefetch buffer.
:-/)
I've scanned the 78032 manual and I'm sure it's online somewhere.
I'd be (very) surprised if the IB is not documented somewhere in
there. Console code generally pulls all kinds of tricks to try and
work reliably even in a context where the user could mangle almost
any bit of memory from the console. making full use of the hardware
is just par for the course!
Antonio
22 Jun
22 Jun
12:03 p.m.
Heh. I have a
MicroVAX-II simulator that's sort of on hold. I had
similar issues; there are some undocumented (as far as I can tell)
aspects of the hardware which the ROM POST code depends on.
I've scanned the
78032 manual and I'm sure it's online somewhere.
I'd be (very) surprised if the IB is not documented [...]
25 Jun
25 Jun
1:58 p.m.
der Mouse wrote:
Ah. Then "undocumented (as far as I can
tell)" here really means "I
haven't found the right manual yet".
I'll keep your message around and go digging for the scans you refer
to some time when I have the leisure. Thanks for the pointer!
Before I forget I should say that slipping "78032" into Manx gets
you the CPU Chip guide and the KA630-AA User's Guide (which is
equivalent to what was later usually called the CPU Technical
Manual).
Antonio
2:15 p.m.
Before I forget I should say that slipping
"78032" into Manx gets you
the CPU Chip guide and the KA630-AA User's Guide (which is equivalent
to what was later usually called the CPU Technical Manual).
I have a scan of EK-KA630-UG-001, which has been very valuable to me; I
assume that's the User's Guide you mention. I'll have to see if I can
dig up that "CPU Chip guide".
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML mouse at rodents.montreal.qc.ca
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B
6389
days inactive
6395
days old
9 comments
7 participants
participants (7)
-
arcarlini＠iee.org -
bfranchuk＠jetnet.ab.ca -
brad＠heeltoe.com -
ethan.dicks＠gmail.com -
mcguire＠neurotica.com -
mouse＠Rodents.Montreal.QC.CA -
ploopster＠gmail.com