View unanswered posts | View active topics It is currently Thu Jun 27, 2019 6:28 am

Reply to topic  [ 21 posts ]  Go to page Previous  1, 2
 Bitslice CPUs - machines built with the AMD2901 ALU chip 
Author Message

Joined: Wed Jan 09, 2013 6:54 pm
Posts: 1192
Very good! I see there's some material in this book: ... AJ&q=cesar

Cesar-SIMD-1988.png [ 74.77 KiB | Viewed 4164 times ]
Wed Jan 25, 2017 4:50 pm

Joined: Tue Jan 15, 2013 10:11 am
Posts: 114
Location: Norway/Japan
The book entry about CESAR is correct, except for a couple of inaccuracies - the host computer was never a 16-bit ND-100, it would be a 32-bit ND-500 (or possibly -5000/5700[Edit: Yes, 5000 series]), as I mentioned earlier, and later the host became an SGI computer. An ND-100 wouldn't be used as a host. 'PE's (Processing Element), used about the bit processors, weren't called PE back when an ND computer was host, they were called S-Elements. Some years later they were called PE, but by then the host was an SGI.

I have a 68020 Control Processor board here as well.

There were probably only 6 ever built of these computers. They were dedicated SAR processors (Synthetic Aperture Radar, needs lots of FFTs), so the market was a special niche. It turns out that we have one complete set of boards for one of them (~13), stored in a co-workers office for maybe 25 years (he used to work for the company making them. Not all of the 6 were deployed, so that's why he kept one.) But the boards will go, except for a few I'll keep for parts. So unless I can picture all of them by tomorrow there won't be any more photos. I have one picture to share for now, which is the bit slice processor board. If I can take photos of the rest I'll start a separate thread, about the CESAR computer.

This imgur link is supposed to have both the full-size photo and the description, but I can only see the description, for some reason:
The full-size picture is
Smaller picture follows here.

File comment: CESAR Triple Address Processor - bitslice
cesar-tr-add-proc-serial-ND0006026-small.png [ 1.7 MiB | Viewed 4154 times ]
Thu Jan 26, 2017 10:52 am

Joined: Tue Dec 11, 2012 8:03 am
Posts: 264
Location: California
Holy mackerel!! It looks like about 340 ICs (counting the mezzanine) and there might be more under the big ones! I've laid out a board with 500 parts, but much of it was analog, and a lot of those were resistors, diodes, and capacitors. The IC count was only about 30. Routing buses for the one above must have been a nightmare, requiring a huge number of layers. Troubleshooting and repairing with those white and yellow wires would be quite a challenge too.

_________________ lots of 6502 resources

Fri Jan 27, 2017 4:44 am
Profile WWW

Joined: Wed Jan 09, 2013 6:54 pm
Posts: 1192
Thanks for the photo, Tor - it's enormous! I see very few bypass caps, fewer than I'd expect. As for the chip count, for a prototype in 1988, perhaps one would hope to mop up a lot of the chips in semi custom devices, if there was a need to do it. (Do we see any programmable parts here? Presumably the ones with the sticky labels.) But for a very low volume product maybe that would never have happened. I notice the first FPGA was introduced in 1985, with 800 gate capacity, so perhaps the future for a system like this would be a number of small FPGAs.

(Edit, at 50MHz max, the first FPGAs might not have been up to the job)

Fri Jan 27, 2017 8:01 am

Joined: Tue Jan 15, 2013 10:11 am
Posts: 114
Location: Norway/Japan
I hadn't considered the size/density of the board :) (probably because it looks just about the same as a lot of boards used in the superminis of the time - at least the ND ones. The ND-5000 series used the same size and type of boards, approximately 37.5 x 40.5 cm.)

Although that mezzanine board makes it a bit more dense than most (as you have guessed, the area under the mezzanine board is fully populated with ICs too).

This system had 13 boards, the bit-slice board is just one of them. But it could have been much worse. The design was started in the early eighties, but in 1985 they decided to make a drastic change - moving from NMOS to CMOS. That would (and it did) reduce the number of boards from 130(!) to 13.

Attaching two pictures of one MALU board, with 128 VLSI S-elements (processors). It's populated on both sides. This board is very heavy.

File comment: Side 2 of MALU board, VLSI processors
malu-side2.png [ 1.51 MiB | Viewed 4136 times ]
File comment: Side 1 of MALU board with VLSI processors
malu-side1.png [ 1.56 MiB | Viewed 4136 times ]
Fri Jan 27, 2017 9:09 am

Joined: Mon May 22, 2017 12:11 am
Posts: 23
Location: NW Scotland
I used to work on a video terminal which was four AMD2901's
and truck load of MSI ttl logic, it was green screen and vector graphics
We had them in Nottingham University, and 2 engineering firms in Newark ( LIncs ), not Mass?
Who sacked the drawing office more or less overnight, and designed locomotives
for B.R. remember them aargh!!!
Think they might have become Westwood Terminals because we had a half a dozen at
Plessy sites in the south of the UK
I could be wrong, but don't care, I 've been wrong before.
Just remembered we had a couple in Loughborough at the Brush Works
( no not brushes, heavy duty railway locomotives, thanks to any American readers )

Best Regards Ian Hughes ( carronjack )

Some rules of Engineering :-
Interference fits don't
Sliding fits don't
The part most likely to go wrong, will be in the least accessable place

Thu Jun 08, 2017 10:50 pm
Display posts from previous:  Sort by  
Reply to topic   [ 21 posts ]  Go to page Previous  1, 2

Who is online

Users browsing this forum: No registered users and 1 guest

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group
Designed by ST Software