IBM 7030 Stretch
en.wikipedia.org
en.wikipedia.org
"Of course, Stretch came originally with an oil-cooled core memory system.
Jack Worlton of the Los Alamos Scientific Laboratory (LASL) recalls a problem experienced during the acceptance test phase, when there was a transient memory error in the cores. The engineers worked for days to remedy the error but were unable to figure out what was going wrong. They finally hit upon the problem. As Worlton explains it, there was a piece of solder loose in the oil bath, and because the oil was constantly in circulation the solder would move and attach itself onto a core and cause an error. Then it would move and lodge onto a different core and cause an error there.
It was the only error I know of that was corrected in a machine by giving it an oil change, Worlton quips."
from : http://www.chilton-computing.org.uk/acl/literature/reports/p...
So yeah, it's "a bit" like yield, but only in the sense that a Cessna and an F-22 are "a bit" alike. I've done low-level work on multiple RISC processors that had all sorts of exposed-pipeline and delegated-to-software functionality (e.g. TLB handling on MIPS) in the name of simplicity, and I've even enjoyed it somewhat, but I'm still glad I never had to work on anything as "raw" as Stretch.
Weight: 70,000 pounds (35 short tons; 32 t)
Power: 100 kW[1] @ 110 V
It had incredibly cool-looking DUAL circular displays and a minimalist console keyboard built into the table...
https://en.wikipedia.org/wiki/CDC_6600#/media/File:CDC_6600....
The most popular parts of the 6600, as souvenirs, were the smoked-glass cabinet doors.
He has some good stories. Says he learned how to create operating systems by hours of watching the data flow on those tubes.
"π was computed to 250,000 decimal places on an IBM 7030 at the Commissariat à l’Energie Atomique in Paris in February 1966, and a year later, in February 1967, a CDC 6600 was programmed by J. Gilloud and J. Filliatre, at the same institution, to yield 500,000 decimal places. The program was again based on Störmer’s formula (2) and the Shanks-Wrench method for checking the digits; the running time was 28 hours and 10 minutes (of which 1 hour and 35 minutes were used for conversion), and an additional 16 hours and 35 minutes were needed for the check. These quarter- and half-million digit values of π were published in reports of the Commissariat à l’Energie Atomique in Paris."
Well worth a read if you can find it -- the Wikipedia summary is rather spoiler-y and doesn't capture the charm of the story. Mr. Bester worked in broadcast television at the time he wrote it and there is some amusing commentary about ad trends of the day.
For example, there's a comment about having to lower the clock speeds but it doesn't say why this decision had to be made.
Cray at CDC used the newer and faster silicon transistors in the 6600. So did IBM's later System/360.
In fact Cray funded the development of the fast silicon transistors at Fairchild with a $500k contract.
https://www.computerhistory.org/siliconengine/silicon-transi...
IBM made the classic corporate mistake - sales over-promised and then told engineering to build something that couldn't be built. So Stretch became an involuntary R&D project instead of a commercial product.
This is sometimes interpreted as indicating NSA interest in the machine.
I wonder if anyone has made an emulator for it yet?
But I suspect it was so much “we want to address petabytes of ram” but rather “we are doing numerics that require 64 bits of accuracy”
Apparently it isn't an easy system to emulate.