I had never heard of that machine until I posted this and someone sent me a link to it. The architecture of the ZedRipper is actually incredibly similar - each CPU has its own dedicated 64KB of memory and then an I/O mapped high-speed synchronous ring bus connects them all. On a MHz * CPUs basis, mine should actually be slightly faster =)

Bob Bane hangs out here regularly!

By far the most important feature of ZMOB was that each board had four differently colored software blinkable LEDs, all arrayed behind the smoked plexiglass cover of the rack they were all plugged into. It looked really cool and was great for hypnotizing and mesmerizing people, which was how we got a big NFS grant (which we spent on Vaxen and Suns and Xerox workstations).

https://news.ycombinator.com/user?id=bane

Here's a photo of some UMD people in front of ZMOB in the Department of Research Simulation, with some white Xerox file servers and black Sun server in the background:

https://i.imgur.com/zLjYTS3.jpg

That was probably me, I posted a link in the Hackaday comment section. Don, I forgot about the colored LEDs. I need to find a picture!

Sorry, the only photo I have is monochrome (linked above). Maybe Bob Bane has some color photos! It ran a "Dining Philosophers" simulation that did a lot of cool blinking, and I ported FIG-FORTH to it, which was great for writing LED blinking programs.

https://www.cs.umd.edu/sites/default/files/zelkowitz-report....

Laboratory for Parallel Computation and Z-MOB Around 1980, Dr. Charles (Chuck) Rieger (Figure 3.12) designed a computing system to run as a network of processors. The initial design consisted of 256 Z80A processors. Each contained 64K memory, linked together on a “conveyor belt” (a 48-bit wide slotted ring architecture), as packets of information were passed from one processor (or “moblet”) to another. This “mob of Z80s” soon became known as Zmob. The system was programmed and controlled by a host VAX 11/780 computer. An Air Force Office of Scientific Research grant was obtained for building the basic Zmob hardware.

It was apparent to both Minker and Basili (after addressing the crush of student majors) that to be a top department, we needed more equipment. Minker was behind the acquisition of the VAX 11/780 (Mimsy) in 1980, which was a good start, but was insufficient. We needed terminals on every desk, additional machines like Mimsy, and a laboratory under our own control. Computers were our technology and if we couldn’t have and control our own machines, our research in this area would always be hampered.

In 1982 Basili organized a proposal team, consisting of Agrawala, Minker, Rosenfeld, Stewart, and Weiser. They met and came up with the idea of a parallel computation laboratory, using some of the ideas in Zmob as a starting point. A Coordinated Experimental Research (CER) proposal was written to NSF, and in 1983 it was funded as a 5-year $4.3M grant. This grant greatly expanded the departmental laboratory, purchasing machines tove, gyre and gymble, as discussed earlier.

PRISM, under the direction of Minker, was the core of the CER project. It was conceived as the software system designed to run on Zmob, and was run as a simulator on the VAX. It used Horn clause logic (similar to the design of Prolog) to implement AND/OR parallelism. Users could specify the number of problem solving machines (i.e., moblets) to use, the number of database machines to use, and the number of machines to handle executable procedures. Statistics were collected to determine how effective the search strategy was. Users could also run PRISM in a sequential Prolog-like manner.

Continued Zmob research was funded by this grant and a version consisting of 128 moblets was built and installed in the Department laboratory (Figure 3.13). Moblets were accessed by physical address or by pattern matching. Programming in Prolog like languages was seen as the mechanism for using Zmob. Mobix was the operating system designed to run on Zmob, which would hide many of the complexities of the Zmob hardware.

Due to inexperience in hardware design, Zmob ran, but not as reliably as desired. The Z80s were 8-bit processors and became obsolete too soon. 16 bit and 32 bit versions of Zmob were designed, and a 16 node ring of 16-bit processors using the Motorola 68010 processor, called McMob, was built and used for several years. Initial design work for a further advancement, called Chessie, was started, but a prototype Chessie was never built. Work continued for several years, but after the NSF grant that suppported most of the development ended, work slowly stopped on that activity.

Mark Weiser’s foot “mole” of 1986 was another interesting idea that grew out of the CER grant. Instead of using a mouse controlled by the hand, a foot pedal was developed, eliminating the need to move the hands off the keyboard in order to control the screen. Several variations were built and tested on Sun workstations, but the concept never became fully operational. Unfortunately for the Department, Weiser soon left for a research position at Xerox Palo Alto Research Center and this activity also ended.

https://en.wikipedia.org/wiki/Mark_Weiser

Funny story, in my undergrad I worked at UMIACS and we had to install a set of foot pedal mice buttons in an office next to Minker's. The user would move the mouse on their desk, but click with their feet.

I used to joke that in the future, EMACS would require foot pedals to handle all the different modifier keys it would want.

I love all of the weird computers that got built by academics in the 1980s! That sounds like such a cool machine.

This is why I read hacker news! Awesome story thanks!

I’ve recently been working on an ESP32 variant of something similar [0]. ZMOB-like designs are so much fun to think about.

[0] https://twitter.com/eismcc/status/1201310140416217089?s=20

My own experiment has recently been a Z80 hooked up to an Arduino - which fakes I/O and RAM accesses to/from the chip.

I got as far as getting BASIC to run, but struggled porting FORTH to z80. Existing Z80 forth interpreters I found need almost 64k but wiring up some static-RAM hasn't yet given me a functional system so I'm hazy on what is wrong; either my real-RAM or the interpreter itself.

My issue when I build my homebrew computer in college was that I didn’t think to put biasing resistors between the z80 and SRAM, since it’s TTL.

You probably didn’t do something that silly though.

That's great! The idea of a cluster with a wireless interconnect is hilarious to me for some reason.

> [...], interprocessor communications system called the "conveyor belt"

Is that where the Mill architecture has the belt concept from? I think i heard some belt concept in the Mill's architecture's context.

Nah, looks more like an NoC fabric.

The belt is more a temporal view of the bypass network instead of a register file.