Re: test (was: [f-cpu] No latches, please !)

[Yann Guidon <whygee@f-cpu.org>]

hi !

so here goes the Science Fiction again :-DDD
I respect you all but sometimes, you look more
like children than me :-P

Ben Franchuk wrote:
> Juergen Goeritz wrote:
> > Other example: imagine a 500 processor cluster in an orbit
> > around Jupiter at high rad doses (I love these examples :)
hmmm i didn't know Jupiter was made of uranium/plutonium...

> > 1/2 of your cluster sleeps to recover from radiation, the
> > other half is operating. Before sleep you test, after sleep
> > you test and during operation you test at regular intervals.

hey ! in my example i did not mean a multi-cpu system.
In this case, of course you wouldn't forward the hard reset
signal to all the other cpus. i guess it was discussed (is it
the word ?...) months ago...

by the way, how do you want to verify the system's integrity
without reset ? Because its seems obvious to me that the BIST
will erase the memory it wants to test : you will loose all the
program's current state.

> Nah... I would have the 500 cluster in orbit around the sun
> running a anti-matter production plant using solar energy.
Pffff...

btw, it remembers me a talk that was recently given in my university :
they want to send clusters of micro-satellites in space, in gravitational-neutral
"holes" that would allow the stuff to stay there without propulsion.
I have no idea what the goal is. it looks completely dumb to me, very expensive,
useless and unrepairable. ask the NASA's JPL for details. at least,
sending people to the moon made more sense to me.

> Anti-matter
> takes a hell a lot of energy to produce but could make inter-planetary
> travel possible. The problem is that transistor size of modern logic is
> it too small to not glitch or fry from radiation. A machine with a 250 ns
> memory cycle and 1 Meg of ram could be a practical size and speed limit
> for hard radiation. While I have not looked a 'Leon' I would expect to have
> error checking of some kind in the alu and memory circuits. That was one
> advantage the old decimal machines had, error checking codes.A parity bit
> may be a good feature to consider with transistor sizes getting smaller
> and the mean time between soft errors goes down.

From the leon website, the radiation tests were satisfying.
they use specific hardened cells as well as a self-recovering architecture
but i did not care. have a look !

> Ben Franchuk - Dawn * 12/24 bit cpu *
WHYGEE
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/