MightyFrame™ Overview

View full PDF in new tab        Thanks AGAIN to Al Kossow of bitsavers.org for providing these!

See the related S/80 Overview

See the related S/640 Overview (looks just like the big MightyFrame)

Computerworld reports some prices in 1988   


Jeremy Chatfield, who worked at Convergent Solutions Ltd in London from 1983-1985, says:
"The MightyFrame was a lot more like a standard UNIX system - no subsystems running their own OS's, as in the convoluted architecture of the MegaFrame. And no peculiar expansion bus, as in the MiniFrame."


Here is that full message, pending further segmentation:

Hmm, let's see. 

CTOX equipment - no idea. I'd try eBay and CraigsList, Gumtree, etc. There are also a bunch of people who like old computers. I've come across people who maintain collections of early consoles, of machines that don't have 8-bit bytes (the Elliot 309, IIRC, had a 39-bit word), and strange devices like the Norsk Data Lisp Machines. I do remember talking about ten years to, IIRC, someone from the westernmost of the North West European countries - Denmark, Holland, Belgium, maybe Norway - who was running some 68k kit. Try meetup groups and hobbyists, tech history buffs? 

The boot problem - Maxtors were pretty noisy. I used to be able to diagnose some common optimisation problems just by listening to the drive head stepper motors. Can you hear a distinctive pattern of stepper head noises before the MightyFrame looks for the tape. There should be a seek pattern where the head is moved across the disk to look for boot sectors. If you're not getting that, then, not sure what's wrong. If there's no distinctive boot-sequence ratcheting noises, then it could be the drive isn't jumpered as Drive 0 (boot drive, IIRC), wiring, the drive, drive contents, Boot ROM (they were, IIRC, Flash - I'm not sure what happens to Flash after 20 years). 

You're sure that the drive was last used to boot a MightyFrame, and wasn't for example, a second drive used for data? IIRC there was room for several drives in a MightyFrame chassis, and it might never have been formatted as a boot drive. One of the clues would be the original configuration of the drive select jumpers. IIRC, you could jumper a drive to be one of four devices on the bus. I think MightyFrames only booted from Drive 0. Each drive had to have a unique ID, and a drive used in ID 1, 2 or 3 wouldn't have a boot loader on it. 

IIRC the Convergent OS tapes weren't tar, they were 'cpio'. That was a different archive format, incompatible with 'tar'. 

The Unix PC came late in the series. If I recall correctly, the first machine was the Burroughs B25/MegaFrame, and the other CTOS based MiniFrame, then the MightyFrame came out, then the UNIX PC. Early UNIX kernels were very hardware dependent. I worked for a while porting UNIX software to over 100 Unix variants. It was very rare, except on the PC based systems, that an installation mechanism and kernel for one system would have any useful impact, even if the processor was the same. I'd be very wary of the idea that a UNIX PC boot/kernel would work in a MightyFrame. MightyFrame, MegaFrame and MiniFrame had radically different kernels, as a consequence of the underlying hardware and expectations. On a MightFrame, for example, the sub-processing systems (File Processors, comms processors, etc) all booted CTOS and then the main CPU would boot to run UNIX, using the FPs to prvide the file system). There'd be little chance that a MegaFrame boot or kernel would be useful for a MightyFrame, MiniFrame or UNIX PC. The MiniFrame... Very strange expansion bus and IIRC a lot of the early boot sequence was probing that bus to find out what was present. My suspicion is that Convergent engineers would think of every machine series as being a unique boot experience, with unique hardware challenges. It probably wouldn't occur to them that a kernel could be used in multiple different motherboards with different buses. I certainly wouldn't have thought that way in the early to mid 80's. 

It took the PC, with the common motherboard and standardised hardware interfaces, to produce common boot systems that would work comfortably across different vendors' systems, and even in 1994 (last time I actively did kernel work as a paid job), we were still finding PC UNIX installation mechanisms that were part of the whole vendor-lockin model of thinking that lead to hardware-unique boot sequences. 

Also, the kernel subsystems were statically compiled in. Loadable kernel modules were a later addition - started coming in the early 90's. That made probing for different buses very difficult. I helped debug and solve some cases in 90-94 where hardware conflicts even in a single series of machines, could result in failed device operation, and having to compile hardware specific kernels to make specific combinations of hardware work. 

So... I'd be very surprised to hear that a UNIX PC install would work on a MightyFrame - unless the core hardware was pretty much identical (same hard disk interfaces, same motherboard memory subsystems, comms chips, etc, etc). Comparing the core hardware specs would be route to go. If there's differences in the devices, different drive interfaces, etc, different CPU, different comms chips, then the chances of success will drop, rapidly. 

Got to go - working on a project today. I'll try to answer some of your other questions later - but it has been a very long time since I worked on these machines, and I only worked on them for about a year, without kernel code (part of the reason I left - I was effectively used as a support tech, not a kernel capable coder, and the puzzles were limited without kernel code). 

Cheers, JeremyC.

Admin's Link 


Clarence Dold (LinkedIn) said, in a Google+ Hangout Chat on August 23, 2013:

"What kernel changes were made from UNIX System V Release 3 to support CTIX for the MightyFrame? The Mightyframe ran SysV 2.2, never SVR3."


So with that in mind, here's a link to UNIX resources, which includes System V Release 2, but not 2.2.  What's the difference?

No comments:

Post a Comment