3000: Cards and punching and tape, oh-29!
May 28, 2014
The Hewlett-Packard System/3000 -- that's what the computer called the 3000 was first known as during the era when punched cards and tape could drive its data. The 3000-L mailing list popped back up to life last week with stories about the era when hanging chads and IBM 029 punch machines were a working part of MPE's four decades of historic service.
History for an active operating environment whose pedigree includes punched tape and punched cards -- that's pretty much exclusive to the HP 3000. Punching pedigree is a mark of utility and durability, even if those card readers are only in museums and garages today. One recently sold on eBay for more than $300 to a collector.
Maybe it was the debut of a System 360 mainframe on Mad Men's penultimate season that put punched cards into the minds of its longstanding users. Mark Ranft of Pro3K told a story last month about his first IT job as a System 360 operator in the US Marine Corps -- and how that led to a Nortel assignment with a card reader and paper tapes. "Thankfully they had a Series III [HP 3000]. As an operator, I was bored to death, so I read all the manuals. That's how I got hooked on MPE."
About a month later, former OpenMPE secretary Tracy Johnson started the 3000-L readers down nostalgia lane by pointing to TELTAC: a Teletype tape-to-punched card conversion program. "Was there a Contributed Software Library program for that?" he asked. The MPE CSL was born as a swap tape, during this era of punched card holdouts. Gilles Schipper of GSA associates replied there was no need for a CSL program, because FCOPY has always had that capability.
The memories of cards and punching and the 3000 started to tumble out of the readers of the L. "If I recall correctly," said Terry Simpkins of Measurement Specialties, "when I was with HP's Disc Memory Division in Boise back in the early '80s, we actually had a card reader connected to one of our 3000s. I brought several boxes of cards with me from grad school, and we read them into EBCDIC files. Don't ask why I was carrying boxes of punch cards around the country."
The HP 3000, in its infancy, could use punched cards or paper tape. Those were two computing props not seen in Mad Men this spring. But they're remembered as durable data mediums, even by those of us who dropped a deck or two of them in front of a college computing center on the way to running a program.
In just a matter of about eight hours, Jeff Kell of the University of Tennessee at Chatanooga had chipped in a thorough history of how data was sent to and from the earliest HP 3000s. The story included a speed measurement that used a holiday as comparison. The HP optical mark sense card reader was the tortoise in the data race.
As for the "mark-sense" reader... we had this grand plan to do grades on "mark-sense" cards. The idea was to "print" class cards (one card per student, sorted by instructor by class), and let them pencil-mark the corresponding grade for the student. It was great in theory, but the mark-sense reader had much less than stellar performance and reliability (it sucked!). And having these "printed" cards burst on their perforations to yield the "card" left some rough edges, which the reader really, really hated. And it was slow as Christmas. Heck, it was slower than Leap Year.
We got an HP2000/Access system in Fall of 1975. It not only supported a card reader and printer, but also supported the remote job entry communications with the IBM at the University of Tennessee at Knoxville. So we had a card reader upstairs in the student keypunch lab, as well as a printer, and there was no more waiting for submission. They could just feed their jobs directly to the reader, their printouts came back to the printer, and it was available constantly. Big step forward.
Later we got an HP 3000, and had a copy of MRJE/3000. Now students could enter their programs online via Editor/QEdit/Quad/whatever they prefer, submit their jobs via MRJE, and view their output in SPOOK before actually printing it out. Even better still.
Kell added that his campus kept the card reader for the 3000 for legacy purposes. This is a 3000 customer that only turned off its MPE systems last December.
Card readers for the 3000 lasted through the lifespan of the Series 70, which means into the early 1990s.
"HP reluctantly supported a card reader through Series 70," said Bob Jankowski of Ideal Computer. "It was definitely available with HP-IB interface and required a dedicated GIC and an auto tap switcher for power. I remember working on these a few times -- and one of my current customers still has theirs in the computer room. One of the wearing parts was called a 'picker sector.' Try saying that 10 times fast. The HP 7260A was the optical mark sense reader. I remember it being a serial device used through MPE-V and being picky about what it would read.
Kell's colleague Tony Shepherd recalled the budget-conscious approach that a computing pro of the 1970s had to embrace. Carpentry power tools and rubber stamps were sometimes among the best data tools.
The perforated card edges were a problem. We wound up printing and bursting them, then putting them in card trays (3,000 cards per tray) and sanding the long edges. It took a little explaining to get management to understand why we needed to buy a dual-action orbital sander with integrated vacuum pickup in order to get grades to post. Sears had one for about $50 that did a great job. We had a good incentive to get the OpScan process developed quickly, and it was indeed much better.
In those days we were just staying ahead of the bleeding edge -- we had a small (but dedicated and very smart) staff and no money. Solutions had to be quick and cheap. For example, one office wanted a new system to record sales of parking bumper stickers. We spent some hours "studying" their needs, then presented them with a bound ledger book and a Bates numbering stamp. It fulfilled all their stated requirements.
The physical manifestation of data, cards had personality. "The first card stock we used had problems with curling," Walter Murray reported from his early 3000 days, "but the second stock we tried worked pretty well. Something to do with "long grain" versus "short grain," as I recall. You'd think we were buying rice."
As for the card reader on Murray's Series II HP 3000, Kell described it as hardware that reduced the footprint versus IBM's original designs.
In the HP card reader you loaded cards on the right into a diagonally-slanted tray, pushed the start button, and it had some sort of combination air driven / pick roller thing that swiped the card through the reader into the output stacker on the left. It was pretty darn quick about it too... not up to par with an IBM's speed, but not slouchy at all. And it fit easily on a tabletop, while the IBM version was the size of a chest freezer.
The work was obviously tedious. It might have helped develop an attention to detail in the earliest part of a 3000 pro's career. Kell has the last word on what a keypunch looked like from that era.
If I remember models correctly, there was the 029 (punched cards real-time), the 129 (buffered a card, you could "backspace," it only punched the card once you released it) -- and this service bureau I worked for had some key-to-disk things that "punched" (wrote) data to floppy diskettes. When they were done and verified, you loaded the diskettes into another IBM thing that loaded the diskettes to 9-track tapes that were used as data input on the mainframe.