On Sunday, August 15, 2021 at 5:44:41 PM UTC-5, Quadibloc wrote:
> On Sunday, August 15, 2021 at 3:06:43 PM UTC-6, MitchAlsup wrote:
> > A particular data pattern would cause worse
> > case power supply consumption at the multiplier board--setting up standing
> > waves between the multiplier and the power supply of that sub-rack.
> It was not designed so that the power supply had a capacity in excess
> of that required for all possible operations and states of the machine?
>
> What on Earth were they _thinking_?
>
> Of course, even IBM designed some of their line printers so that they
> were unable to fire all the print hammers at once should that be
> required, so Control Data isn't the only company with over-optimistic
> engineers.
<
We found out what the hammer timing parameters were and constructed
comment cards so the when out output was being printed the printer
would make distinct noises and we could walk over and pick up the printout
without standing in the line.
<
>
> John Savard

According to Quadibloc <jsavard@ecn.ab.ca>:
>On Sunday, August 15, 2021 at 3:06:43 PM UTC-6, MitchAlsup wrote:
>> A particular data pattern would cause worse
>> case power supply consumption at the multiplier board--setting up standing
>> waves between the multiplier and the power supply of that sub-rack.
>
>It was not designed so that the power supply had a capacity in excess
>of that required for all possible operations and states of the machine?
>
>What on Earth were they _thinking_?

They were probably thinking that the 6600 used so much power that it circulated freon
through the processor bays to carry the heat to the chilled water system, so that's
all they could provide without the thing catching fire.

--
Regards,
John Levine, johnl@taugh.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly

On Sunday, August 15, 2021 at 5:44:41 PM UTC-5, Quadibloc wrote:
> On Sunday, August 15, 2021 at 3:06:43 PM UTC-6, MitchAlsup wrote:
> > A particular data pattern would cause worse
> > case power supply consumption at the multiplier board--setting up standing
> > waves between the multiplier and the power supply of that sub-rack.
<
> It was not designed so that the power supply had a capacity in excess
> of that required for all possible operations and states of the machine?
<
The crashes were data specific--and it took a long time to connect the
crashes to the power supply.
<
Side note: The CRAY 1 was designed so as to present a DC load to the
power system at all times. Each ECL gate had a True and Complement
Output lines and each were terminated with 50 ohms, the used signal
at the receiving end, the unused signal at the driving gate. Each gate
simply routed its in current out one or the other output.
>
> What on Earth were they _thinking_?
<
They had never run into such a problem before and were caught off guard.
>
> Of course, even IBM designed some of their line printers so that they
> were unable to fire all the print hammers at once should that be
> required, so Control Data isn't the only company with over-optimistic
> engineers.
>
> John Savard