Just a brief, somewhat geeky note, apparently SpaceX is using PowerPC microcontrollers in their Dragon capsule.
As mentioned here:https://www.youtube.com/watch?v=9lyB9ScW3Y8
Starting at about 7:50.

Why Power? Probably because of a long history of embedded designs (in automobiles and industrial applications).

Anyway, not ARM (and certainly not X86/64), Power.

> apparently SpaceX is using PowerPC microcontrollers in their Dragon capsule.

To be more precise, it'a combination of three x86 microcontrollers (judges) and a PowerPC microcontroller (actor).

> https://www.youtube.com/watch?v=9lyB9ScW3Y8

What he utters at 8:55 sounds pretty clueless, though ;-)

> Why Power? Probably because of a long history of embedded designs
> (in automobiles and industrial applications).

Primarily, because of its long history in astronautics :-)

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=8671&forum=3

> certainly not X86

Please relisten at 7:17 :-)

>> certainly not X86

>Please relisten at 7:17
>To be more precise, it'a combination of three x86 microcontrollers (judges) and a PowerPC microcontroller (actor).

You've pointed out a crucial error on my part. BTW - I like the judges/actor terms.

I'm still confused about the level of redundancy. Which processor makes the comparison of the results of the first three and passes the data to the PPC?

>What he utters at 8:55 sounds pretty clueless, though

It does descend into freeform babble at that point.

> I'm still confused about the level of redundancy. Which processor makes the
> comparison of the results of the first three and passes the data to the PPC?

As I understand it, for each of the three dual-core x86 CPUs, the two results of the two cores are compared by the x86 CPU itself. If the two results from both cores match, it is sent to the PPC CPU. The PPC CPU then compares the three results coming from the three x86 CPUs, and will go with the result that matches between at least two of the three x86 CPUs.

>>> What he utters at 8:55 sounds pretty clueless, though ;-)

>> It does descend into freeform babble at that point.

I find the notion that such radiation-hardened parts (which cost 6 to 7 digit figures per part) could be bought "years ago" at Amazon, Newegg or BestBuy, or could be simply bought used today, nothing short of ridiculous.

>I find the notion that such radiation-hardened parts (which cost 6 to 7 digit figures per part) could be bought "years ago" at Amazon, Newegg or BestBuy, or could be simply bought used today, nothing short of ridiculous.

Buying electronic components in general from Amazon is rediculous. Even Ebay would be a better source.

Quote:

Indeed. I get Doommaster flashbacks.

I'm not even sure if "radiation-hardening" electronics, as it was back during the cold war, is even that much of an important thing any more. A single transistor (or track) inside a chip is a million times smaller than in the late 80s, and would presumably so be subjected to much less inducted current when exposed to a powerful EM flux.

> I'm not even sure if "radiation-hardening" electronics [...] is even that much
> of an important thing any more. A single transistor (or track) inside a chip
> is a million times smaller than in the late 80s, and would presumably so be
> subjected to much less inducted current when exposed to a powerful EM flux.

Shrinking process node decreases vulnerability to total-ionizing-dose radiation but increases vulnerability to other kinds of radiation effects like single-event upset or single-event latchup. Thus, radiation hardening is still a necessary thing for space microelectronics and probably always will be. The specific hardening methods have changed over time with the ever-shrinking process nodes, though.

https://en.wikipedia.org/wiki/Radiation_hardening#Digital_damage:_SEE

Quote:

Ah yes. That's ionising radiation. I was thinking EM radiation.