return to main page

Why a (new, incompatible) 1401, not an IBM 650 compatible upgrade?

Robert Garner and others were e-mailing - and the question arose
"Why not upgrade the popular IBM 650 with transistors and core for the new business machine?"

Already discussed was that the 650 had the address of the next instruction in every instruction - permitting optimal placing of instructions and data on the rotating memory drum. These extra bits were expensive and not useful in a random access core memory.

One of the remarks (August 28, 2011) was
  1. The 650 memory was digit, not character. Storing an alphabetic character required two digit positions. ... Thus one position needs only 4 bits, ...

  2. ... Even "alphabetic" was a hack (2 digit, the card reader could only read 30 alphabetic columns) - the basic machine was numeric! Put yourself in that time frame; the then current IBM technology? the then current competing products? ... Building a transistorized 650 would have made no sense - you wouldn't even have gotten as far as cost considerations.

And from Robert Garner (Aug 30, 2011)
>> Why a transistorized 650 was nonsensical
> That question has, I think, long interested Robert

Well, I clearly understand the rationale by now. ;-) Given that core memory was so expensive, I can see how the 650 team must have struggled to figure out how to transition to transistors and core memory.

Because the 1401 was able to usefully offer core memory with as few as 1400 character locations, it's clear that SPACE [development name for the system later called IBM 1401] would have killed something like a 660 (just as SPACE pushed aside the attempted 305 follow-ons too.)

Given that the ID guys (in Poughkeepsie) in 1957 weren't aware of SPACE isn't surprising since that's the same year that Fran [Underwood] started to work on it.

And a previous e-mail from Robert Garner ( added here Aug 30, 2011)
Dag,

Thx for forwarding the link to the IBM Elliot Noyes industrial design (ID) movie: http://www.ibm.com/ibm100/us/en/icons/gooddesign/

It reminded me of the still-true-to-this-day delicate relationship between ID and engineering: ID/marketing folks striving to impress before engineering is ready, while the engineers are trying to get the damn box to work by hook or by crook. ;-) Nevertheless,since mockups can be assembled long before the actual engineering, ID folks are typically privy to breaking product ideas. This was especially born out in this movie, where it was interesting to see:

  1. A rendering for an entire "660"(an assumed transistorized version of the 650) and a mockup of its console,but still using a drum. Drums certainly died with the advent of magnetic core memory, but nevertheless I've been interested in why the 650, then the world's most popular computer (with lots of software applications others would want to borrow/share) did not survive the transition to magnetic core memory and transistors. The bottom line seems to be that drum instruction sets -- with their 'next-address-to-fetch-from' field --were cost incompatible with magnetic core memory that clearly did not require a next-address field (except branch instructions). The 650 had 10-character words (plus sign bit), with a 4-character next address field. At 2.5 cents per month per magnetic-core-memory bit in 1950s currency, 2,000words of 650-equivalent memory would have cost the user $0.025 * 2000 *(10*(5 bits/char) +1 sign) = $2,550 per month rental. The otherwise unnecessary drum next-instruction field itself would have cost $1,000 per month! Apparently designers/product marketers weren't willing to carry the burden of a 40% memory cost for compatibility. ;-) The follow-on to the 650 became the instruction set incompatible 7070, of which only several hundred were delivered.

  2. A mockup of the SMS "cube"system package that the 1401 used, but mentioned in the movie for the 608calculator. The 608 actually did not use SMS cube packaging. One of the pictures behind Elliot is of Joe Logue's 'transistorized 604' prototype and the tape reveals that Elliot did not approve of the old uninspiring engineering-driven packaging. However, Elliot did note that the SMS cube system was being co-designed by engineers and ID folks working together. The SMS cube system packaging was innovative and actually worked well and (as born out by awards and 15,000 reliable 1401 installs).

  3. Elliot's concept for 7000 series packaging was that electronics would be hidden out of sight in a back room with open racks for easier serviceability, a la AT&Ts phone line switching racks. The final 7000 series packaging, called Polygon, was in tall nondescript racks, with large sub-panels of SMS cards on wheels, whose design was likely optimized for cooling and low-latency interconnect).Elliot's main objective here appears to have been to impress, with a wide wall of tape drives and an up front central operator console for "much greater dramatization and expressiveness of the operator's position in the control room." Kind of like a missile launch control room. ;-)

  4. A mockup of the 729 tape unit for"750" and 7000 series, where Elliot was striving for "a good quality look that you'd find in high-performance camera or microscope,a kind of classic design that one wouldn't get tired of over 10 years,and would work as an individual unit and in banks." He was very successful there(!), as the 729 became an icon for 1960s computers (even appearing on the cover of Jacques Ellul's 'technology beware' book called"The Technological Society:")

  5. It was also interesting to see the mockup of what became the 088 collator and then our 1402 reader/punch unit.

Elliot states that his overall philosophy was "minimum obsolescence and maximum quality" and "maximum expressiveness and a dramatization of the machines and what they can do." Sounds good.

> I thought it might inform how we could set up the 1401 Experience Room

Certainly for "maximal expressiveness and dramatization"! We don't have the 1407 operator console unit, but we are partly along on re-creating one (have wired up a electronic interface to a old IBM typewriter, using the home brew serial port interface that Buzz designed in the 70s).

New thought: perhaps we'll want to actually build out the whole console table plus operator chair??

> (not including the portholes, of course)

Because of its "dramatization"effect, I think Elliot would approve of outside-of-building-looking-in portholes!

- Robert


Page started August 29, 2011
Updated through Aug 30, 2011

return to main page