[BC] Re: Kevlar cored wire? (PeterH5322)

[PeterH5322]

>> At Amdahl Corp, we had our coax made to our spec by W.L. Gore Company 
>> (Gor-Tex?).
>
>Their first product was  Teflon insulated wire - made in the Gore family 
>basement and back yard. Some customers were surprised at the grass 
>clippings in the packages. They've come a long way.
>
>http://www.marketbusting.com/casestudies/WL%20Gore.pdf

Gore made one batch of our special coax which had the leads reversed.

The ECL 10K signals were driving the shield, and the center lead was the 
ground.

The Amdahl machines, which then were the fastest machines on the planet, 
and which were classified as "munitions" and could not not be exported, 
didn't work worth a damn, and all coaxes had to be replaced.

The first 25 machines were made with a Japanese-manufactured coax, and 
these were brittle and sometimes failed.

It was after machine number 25 that we changed to Gor-tex.

NASA received machine number 1, and upon return of that machine from its 
lease, it was remanufactured equal to new in every respect, including 
replacement of all coaxes, and was donated to the Smithsonian Institution.

For a 5 MIP machine, the 470/V6 was revolutionary.

Today, the Hercules S/370 emulator running on a very fast Pentium can 
successfully emulate a 1 MIP machine.

No emulator can equal the throughput capacity of the Amdahl's I/O channel 
unit, which was provided with sixteen S/370 I/O channels, each of which 
could "burst" at 4.5 MB/sec, simultaneously, and which exceeded the 
competitive IBM CPUs by a significant margin.

The early machines had a cycle time of 29 nsec; later machines had a 
cycle time of 26 nsec. Many machine instructions could be executed in one 
cycle. Division was accomplished by multiplying by the reciprocal, an 
optional high-speed arithmetic feature which was licensed to our 
competitors.

Machine timing was to within +/- a few hundred psec, and was adjusted by 
changing lengths of coax. Obviously, this was a factory adjustment.

Every machine was operated for two weeks without error before being 
released for shipment. Should any error occur before completion of 
testing, the test cycle was restarted from the beginning.