Three Geeks Rescue a 50-Year-Old IBM 360 Mainframe From an Abandoned Building (ibms360.co.uk) 129
In late April of 2019 Slashdot reader Adam Bradley and engineer Chris Blackburn were "sitting in a pub on a Monday night when Chris happened across a somewhat unusual eBay listing..."
They eventually submitted the winning bid for an IBM 360 Model 20 mainframe -- €3,710 (about $4,141 USD) -- and proceeded to pick it up from an abandoned building "in the backstreets of Nuremberg, Germany." (Where they tackled several issues with a tiny door that hadn't been opened since the 1970s.) By day Adam is a railway software engineer, but he's also been involved in computer history for over a decade at The National Museum of Computing in Bletchley, England.
Along with engineer Peter Vaughan, the three are now blogging "the saga that unfurled...and how we eventually tackled the problems we discovered." But after much beer, whisky, and Weiner Schnitzel, Adam assures us the story ends with a victory: The machine will shortly be headed to the UK for a full restoration to working order. We're planning to blog the entire process and hope some of you might be interested in reading more about it.
They eventually submitted the winning bid for an IBM 360 Model 20 mainframe -- €3,710 (about $4,141 USD) -- and proceeded to pick it up from an abandoned building "in the backstreets of Nuremberg, Germany." (Where they tackled several issues with a tiny door that hadn't been opened since the 1970s.) By day Adam is a railway software engineer, but he's also been involved in computer history for over a decade at The National Museum of Computing in Bletchley, England.
Along with engineer Peter Vaughan, the three are now blogging "the saga that unfurled...and how we eventually tackled the problems we discovered." But after much beer, whisky, and Weiner Schnitzel, Adam assures us the story ends with a victory: The machine will shortly be headed to the UK for a full restoration to working order. We're planning to blog the entire process and hope some of you might be interested in reading more about it.
Big Iron Rocks! (Score:5, Informative)
No, it wasn't fun doing channel programming (Score:5, Insightful)
It was tedious and entirely component specific.
As part of system admin training, you got training in how to do it, but I don't know anybody who considered it "fun".
Re:No, it wasn't fun doing channel programming (Score:4, Interesting)
I found channel programming (especially of unit record devices) a lot of fun when writing standalone personal utilities that could tolerate imperfect error handling. The big challenge in writing channel programs was making them robust, able to recover from various kinds of error conditions. That was tedious and time consuming.
Re: (Score:2)
Maybe, but they do belong on /.
Unlike you...isn't there an AOL chatroom somewhere you could be hanging around in?
Re: (Score:3)
Re:No, it wasn't fun doing channel programming (Score:5, Informative)
Surely you've heard of at least one guy [catb.org]...
Re: (Score:1)
I do.
LBL when in the 70's got to see disk drive races. Think quarter horse or drag racing but with large "washing" machines...
Trick was to get the head movements such that drive housing gets "vibrating", like an unbalanced washing machine. Then get it to start going in only one direction. Once two programmers got their "horses" ready. The race was on! Only real problems, power cable length and walls (hence stopping).
I got to see one of these races... Every think very slow at the start, Once the vibra
Re:Big Iron Rocks! (Score:5, Interesting)
The entire high end IBM line is amazing. People act like virtual machines are the latest hot item. IBM has been doing that since the 1970s. IBM designs things in a vacuum so some of the ideas are really odd. Like the larger machines getting microcode and system configurations injected from a management PC during boot up. They tried to implement as much as possible into the hardware to reduce the system load. Like external terminal controllers and disk controllers.
Re: (Score:3)
Re: (Score:2, Informative)
Hate to tell you, IBM did that ages ago too:
https://en.wikipedia.org/wiki/VM_(operating_system)
"Another copy of VM. A "second level" instance of VM can be fully virtualized inside a virtual machine. This is how VM development and testing is done (a "second-level" VM can potentially implement a different virtualization of the hardware). This technique was used to develop S/370 software before S/370 hardware was available, and it has continued to play a role in new hardware development at IBM. The literature
System 360 Model 20 not really a mainframe (Score:5, Insightful)
To call the 360/20 a mainframe is a real stretch whatever IBM's 1960s marketing literature might have claimed. Even 50 years ago,it was basically a mini computer that could (slowly) run DOS or TOS (Tape Operating System) and System 360 programs. That does not make the machine any less worth saving. Although a small machine, getting it into working order is going to be quite a challenge.
Re: (Score:1)
Wonder what the GCHQ thought it was getting when it got its IBM 360?
vs the ICL 1900, UNIVAC products on offer?
Re: (Score:3)
Re: (Score:1)
My first IBM 360 experience (Score:1)
In my first year of college all our programming were done with punch card machines. Then on the 2nd year, the computer lab had one TRS 80 PC for us to use.
On the 3rd year my college was one of the first U that was chosen by IBM for their 360 trial. When it was first installed, the 360 was all hardware, because many essential pieces of software weren't written yet.
So we, the 3rd and 4th year students became free 'lab mice' for IBM, for we had to create a lot of basic utilities for that thing.
But that is good
Re: (Score:3, Interesting)
My IBM 360 experience is being an IBM brat. Yes, my dad started working at IBM in 1956. One thing I know for certain is that the IBM 360 came out LONG before there was any such thing as a TRS-80. So there had to be plenty of software for that 360 on your campus that was installed the year after your lab had a TRS-80.
Or maybe you refer to a specific model. But I thought System 370 was out by the time there was a Trash 80 in existence.
Re:My first IBM 360 experience (Score:5, Informative)
So, well, all of that is basically not true.
I started college (Wayne State) in 1972. We had dual IBM 360/67's running MTS (Michigan Operating System - developed at University of Michigan). There was no TRS-80.
I got a student assistant job at the academic computing center. I wrote code in 360 assembler and PL/1. I was responsible for the MTS port of IEFSD095 ((Look it up - it still exists!) and wrote some "system info" kind of utility. Wayne State made some contributions to the MTS codebase. I had stacks of source listings labeled on the edges in magic marker in the office I shared with a full-time employee.
We had an IBM 1620 in my high school (Cass Tech in Detroit). We punched cards, first with machine language instructions - NOT assembly - pure machine language (the 1620 was a decimal machine) and then Fortran II. Big yucks were distracting a fellow student so that you could move a few cards around in their deck. (The smart ones figured out how to program the drum card on the keypunch to automatically punch sequence numbers into 73-80, so they could be sorted back into the right sequence...) Ironically, we also had some ancient non-working Univac mainframe (I mean MAIN FRAME, this beast had a 1KW ceramic power triode driving the 1mHz clock...) that somebody had donated in the back of one of my electronics classrooms. I read recently that they actually got it working a few years after I graduated.
The 360/20 was not really a mainframe, though it looked like one. And it was the most popular (most units produced) 360. Reduced instruction set, 16-bit registers, no floating point. The 1620 was probably more powerful.
The 360/67 was the first to have the "DAT box" ("Data Address Translation"). And it had the (dubbed by Computer Comix, the campus computer center underground newsletter...) "Virtue Machine". MTS was probably the origin of the term "virtual machine". We could run MTS on MTS. Or OS/360 on MTS. Or VMS on MTS. Or....
Anyway, I remember taking my 8008 wirewrap kit into work so that I could hook it up to one the "glass teletypes" on the mezzanine above the computer room. (There were a row of terminals that were a mix of ADM-3As and Selectrics). THERE WAS NO TRS-80. For several years.
Of course my iPhone has orders or magnitude more memory and computing power than the dual 360/67's that ran the whole campus timesharing system.
I remember when we cornered the market on Fairchild solid-state memory. We had 8 MEGABYTES of memory!
Later, they got an Amdahl. "King Frank" (Another Computer Comix reference) Franklin Westervelt, the director of the computing center) had worked with Gene Amdahl at IBM, and was one of the inventors of virtual machines.
BTW - thank you King Frank for forgiving me for writing "waddya wanna permit" and offering me a part-time student assistant job!
Re: (Score:2)
According to this benchmark compilation [roylongbottom.org.uk], the 1620 could do 4.6 kips, while the model 20 could do 2.0 kips. A race between snails...
Re: (Score:2)
Why were sorting routines so important early on in mainframe development, subject of so much?
Dropping a tray of punch cards was not uncommon.
Necessity is the mother of all.
Re: (Score:2)
Re: (Score:2)
Re: (Score:3, Informative)
Re: (Score:2)
Re: (Score:2)
IBM 360/370 systems are so cool. Unlike the PCs of today you could write things called channel programs that executed on the disk controller. You could customize/optimize the code to write out or read in data that is "tuned" to the rotation of the disk and move tracks or cylinders (vertical stacks of tracks) with amazing speed. The channels were processors themselves, to a degree, and controlling both your IO and your CPU was what made it all fun.
That sounds cool and all but for the uninformed (ie me) is there anything you can do with this that you can't do with modern machines or is this just a cool nostalgia thing?
Re: (Score:1)
Not with the power supply "on".
Re: (Score:2)
"IBM 360/370 systems are so cool."
That's because you run drinking water through it 24/7 to cool it before it goes into the sewer.
That's going to be expensive.
What does it say about English cuisine if (Score:3)
"I found Weiner Schnitzel (the greatest food ever invented)"!?!
This guy would be orgasmic if somebody ever gave him a twinkie.
Re:What does it say about English cuisine if (Score:5, Funny)
"I found Weiner Schnitzel (the greatest food ever invented)"!?!
To be fair, the guy is English.
Re: (Score:3)
The sort that's had a proper schnitzel in North-Rhine Westphalia, made from pork and covered in a rich mushroom sauce.
Re: (Score:2)
They've gone downhill since the big changes in how Hostess operates. They used to be locally produced in each market area, now they're frozen and trucked all over the country. The old Twinkies from past decades are certainly better. But I am more of a Hostess Snowball fan.
Re: (Score:2)
If this guy thinks Wiener Schnitzel is the greatest food ever, he obviously hasn't tried the awesomeness that is smoked beef brisket or a Philly Cheesesteak... (and yes I have had Wiener Schnitzel many times)
Re: (Score:2)
Smoked. No need to mention anything else. The underlying meat tastes the same regardless of what it is, all American BBQ tastes like smoked something. Same with the Philly Cheesesteak. I don't know why you bother putting steak in it at all. Omitting it would still taste like a cardiac arrest on a piece of bread.
Kind of like what the Germans do to schnitzels. Jaeger schnitzel, Zigeuner schnitzel, at some point you wonder why do they bother with the schnitzel at all if you can't taste it.
Forget your smoked, c
Re: (Score:2)
The worst is in CA.
On I-50 going to S lake Tahoe the German restaurant ('St. Pauli Inn', the beer not the red light district) makes OK schnitzel, with a couple of the usual sauces (Jaeger etc). The terrible version, for 'bay aryans', schnitzel covered in sweet and sour red sauce (delivered in 55 gallon drums, like to standard 'American Chinese' places).
Fuck you and your suggestion to boil brisket. You know _nothing_ about BBQ. Are you Irish, Scottish or English? Fucking meat boilers.
Re: (Score:3)
What if someone gave him a Cotoletta Alla Milanese?
Re: (Score:1, Offtopic)
Re: (Score:1, Funny)
I've never heard of "Weiner Schnitzel", I merely know "Wiener Schnitzel"
Well Wiener Schnitzel is from Vienna, while Weiner Schnitzel is from Veinna, obviously.
Re: (Score:2)
I used to have a weiner dog, so I'd be sad if anyone ever made schnitzel out of one.
It was not at all stuck up and aloof like a wiener dog.
Got a lot of fond memories from RFTA (Score:5, Insightful)
I worked for IBM and was responsible for mainframe memory test/quality in the early 1990s.
No surprise that non of the hardware was on wheels - it was meant to be run on a raised air-conditioned floor. IBM Poughkeepsie had a number of steel floored areas on the production floor that were kept as flat as possible (to within fractions of a millimetre) where the frames were placed during assembly and the covers were adjusted so the machines would look perfect.
No surprise that there is a lot of wire-wrapped connections in there. IBM was big on it as a way to customize hardware. Wire wrapping is extremely reliable and was relatively inexpensive when considering that each machine was a custom install and configuration. They may have some issues here when they're trying to bring up the machines and peripherals as, depending on who did the wiring, they may find clipped leads on resistor SIPs to customize them for specific termination needs (not a fun thing to deal with).
If they are going to get into the wiring, I hope somebody tells them that the data and address bits are the reverse to everybody else's convention; the most significant bit is "0" and the least significant bit is labeled "23" (for address, these machines had a 16MB address space) or "31" (for a 32 bit data word although don't quote me on that, I think some of them had 16 bit). Also, a "1" is a low (maybe negative) voltage and a "0" is a high voltage.
Good luck to them!
Re: (Score:2)
It sounds like you know a whole lot of stuff they would find useful.
Re: (Score:2)
Well at least one of them is on Slashdot and submitted the story... so I’d think there’s a decent chance he’ll be reading the comments.
Re:Got a lot of fond memories from RFTA (Score:5, Informative)
I am also an ex IBMer, though from a much earlier period (1973-79). I am more familiar with the machines from a programming rather than a hardware angle. System 360 used 32-bit words, but addresses in System 360 machine instructions were always 24 bits, limiting the maximum possible storage size (absolutely irrelevant in those days, but a major problem later)..Depending on the machine instruction, data that was addressed could be of various lengths (byte, half word=16 bits, word, double word=64 bits or as specified as specified within the machine instruction). Mostly, half word, word and double word operands needed to be aligned on the appropriate boundary, either as an absolute rule or to avoid severe performance penalties.
Re: (Score:2)
There was an actual reason for this. Before the System/360 there had been two separate types of mainframe: 'scientific' machines with long fixed words of up to 64 bits, for large integer and precision floating point computation; and 'business' machines, addressable by 6-bit bytes. These machine styles separated programming into two different worlds.
The 360 was intended to unify the two kinds of computation. Memory ws addressable by bytes, which were now 8-bit for the first time, but groups of 4 bytes "on a
Re: (Score:1)
Did they have "dual" addressing as the later mini lines of S32, S34 s36 did? Those cpus had 24bit addressing but the "24th" bit has not and address so only 8MB was really available vs 16MB.
When that bit was OFF, the the other 23 bits where real address space, all 8MB was accessible.
When the bit was ON, 17th to 23rd bits where set off, and only 16bit translated addresses. So the a user program only has 64KB size. This 64KB was anywhere in memory and not even contiguous. I think it was broken in 256B ch
Re: (Score:3, Interesting)
The early generation IBM-PCs also came with 'full wiring manuals', if you bought the Technical Reference Manual. I have The IBM PC-1 Techref that has schematic diagrams for the diskette drive, power supply, and keyboard in it. Also the hard drive. And of course, the commented source code for the BIOS.
Re:Got a lot of fond memories from RFTA (Score:4, Interesting)
PowerPC and S/390 still number bits the same way (0 = MSB), so plenty of people are familiar with it. The naming for data sizes (8 = byte, 16 = halfword, 32 = word, 64 = doubleword) is the same on PPC, S/390, MIPS and SPARC. In a lot of ways, x86 is the odd one out. You can see the S/360 heritage in a lot of "serious" systems.
Good start of a great story (Score:5, Insightful)
"In late April of 2019 Adam Bradley and Chris Blackburn were sitting in a pub on a Monday night..."
Because no great story ever started with someone eating a salad .
old guy here (Score:2, Interesting)
When I got into this game, .
If you had a business and wanted to get a computer, you needed to talk to a real estate guy beforehand
Just so you could put the thing onsite.
You needed an extra building with AC,
Power and other utilities and it had to be very close by.
Then you had to deal with ibm or Burroughs or Dec
These guys were not cheap.
You needed serious bank backing to even get this far.
Once it was in place you had to pay for the software to run your monstrous machine.
Software cost much more than hardware.
Fond memories of the start of my career (Score:4, Insightful)
I started in the 70s on a 370/125 (like they also found) running DOS. btw, the Field Engineers were called Customer Engineers (CEs), and the hardware manuals were giant blue manuals that were kept in the computer room. I swear, those manuals had the most detailed descriptions on how to repair anything on the machine.
Re: (Score:3)
oh yeah, about the cables. You may be able to find newer cables. They may or may not be compatable, IBM only recently stopped using big cables like them to connect the CPU to peripherals.
Re: (Score:2)
Bus and tag? Yeah invented in the 1960s and supported until not that long ago. Cables the size of a garden hose with huge rectangular connectors.
Re: (Score:2)
according to this list [ibmsystem3.nl]
they already have
what do they need another 360/20 for?
I think they keep System/360 on their want list in case someone has a 360/195 or one of the initial 1964 models stashed away. And spare parts are always useful.
Geeks Rescue 50-Year-Old (Score:3)
Now if they help it cross the street they'll get a merit badge.
Safe from the Gold Scrappers (Score:2)
Thank goodness this hardware is now (hopefully) safe from the gold scrappers. Because there is a ruthless subculture of scrappers out there on search-and-destroy missions against old mainframe hardware. There is lots of gold in machines where the individual flip-flops are sometimes made with individual transistors and each on it's own circuit board.
the 360/20 was not a mainframe (Score:1)
It was a mini, sort of the baby brother of the full sized 360s, typically used to run remote peripherals like printers. It had 8 registers instead of the usual 16. Max memory was 32k bytes though most were smaller. It had no floating point. According to wikipedia it had 16 bit data words but that sounds kind of weird to me. It was the size of one of those old horizontal-style filing cabinets, think of a refrigerator lying on its side and another one stacked on top of it. It won't run most of the inte
Lawrence Wilkinson's IBM 360/30 Saga (Score:4, Interesting)
A similar story (and a good read), posted to alt.folklore.computers in 1993.
http://www.ljw.me.uk/ibm360/Sa... [ljw.me.uk]
Re:Lawrence Wilkinson's IBM 360/30 Saga (Score:4, Informative)
Re: (Score:2)
Re: (Score:2)
more more (Score:2)
RCA 301/501 (Score:1)
Professor at IIT (Ill Inst of Technology) and I did the same thing around the early 70s. We got hold of an RCA 301/501 for free, hauled it to the 3rd florr in the EE building and got it running. A year earlier some friends and I tried to rescue an old Univac 1103A from IIT (I used to be the operator on it). Unfortunately no one would give us the money to move it and set it up..too bad.
Both of these systems had some "cool" for the time technology. The Univac had a mercury column memory as well as a decelerat
Wow, no mention of John Titor yet? (Score:2)
https://en.m.wikipedia.org/wik... [wikipedia.org]
The time traveller from the future seeking an IBM 360...or do I have the computer model wrong? Anyway its an interesting story to resurface here.
"Full" Restoration? (Score:2)
I don't see that this system was equipped with a disk drive but if it is I would really like to know if they can get it working again. It seems really doubtful.
The tape drives should be easier but the biggest problem in getting those old systems to work is always the mechanical parts. Keyboards and typewriter outputs and such. Takes lots of time an money. I'll be reading the blog.
Re: (Score:2)
the article shows the disks, not sure of the drives since I read it a few days ago. The cables would be the hard part.
The old days... (Score:1)
Started, in school, on a time-shared 370. Punch cards - the first terminal I got to use was after I'd finished all the computer courses (Philly Community College), and had gotten a job as a progammer there, so around '82.
Yes, channels was what made them so powerful.
We were running on MVS, with VMs - this was 1980, you *really* think VMs are a new thing? Hell, aound 2000 or 2002, someone at IBM was playing with Linux, and maxed out a mainframe with 48,000 virtual Linux machines, running IBM's VM, and had it
Re: The old days... (Score:1)