Theonik
Member
The US Air Force used to run farms of PS3s running Linux. There was also the folding@home project for the PS3!
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
PlayStation 1 CPU powers #PlutoFlyBy / NewHorizon probe
- Thread starter Man
- Start date
This is all fine until Cell becomes self aware and appoint Kutaragi as supreme world leader.
BronzeWolf
Banned
I doubt that a decade long no-service can't-fail sortie is not the sort of place to go innovating. Better for it to be super reliable and efficient.
They should send a 3DS for the next probe /stability
They should send a 3DS for the next probe /stability
Teletraan1
Banned
Optical media playback hardware was in its early stages and frankly not up to par. Dig up any CD player from the time the PS1 came out. Does it still work? Probably not.
That's because risc based chips are supposed to be simpler in design than cisc based chips which is why risc chips are generally used in many low power/embedded systems.
RIP AIM alliance.
Hey, the Wii also in space:
http://www.cpushack.com/space-craft-cpu.html
"Dual BAE RAD750 @ 400MIPS (up to 200MHz): Radiation Hardened version of the IBM PowerPC 750 (one configured as a backup, that can take over if the other fails (or a software update disables it). The processor is designed to have no more then one single event upset (a problem requiring intervention from Earth) in its 15 year lifespan."
http://www.cpushack.com/space-craft-cpu.html
"Dual BAE RAD750 @ 400MIPS (up to 200MHz): Radiation Hardened version of the IBM PowerPC 750 (one configured as a backup, that can take over if the other fails (or a software update disables it). The processor is designed to have no more then one single event upset (a problem requiring intervention from Earth) in its 15 year lifespan."
PPC 750 is based on AIM's PPC G3 Apple CPU..so Apple G3 is in space~ I had a G3 and it was a monster....
Stability doesn't seem to be the issue here. Stuff in LEO (low earth orbit) often uses x86 CPUs. Hubble uses a 486 CPU, the ISS a 386 CPU. I assume those chips require either too much power, or too much shielding (or both) to be useful for deep space stuff.
Mihos
Gold Member
Hardening a microscopic system to survive 20+ years of radiation, use less power than your watch, and less heat than an active lizard with less 1 failure requiring a remote intervention is no joke.
Mihos
Gold Member
you can use solar panels near earth, that takes a lot of pressure off power requirements. It also allows you to use active cooling
Never stop fighting the good fight Sidewinder. I'm with you brother.
Paging scientists at NASA, Neogaf user FZZpure thinks you should have put his Pentium 4 chip in the New Horizons probe.
The Sum of Zero
Member
vagabondarts
Member
NewHorizon is Robbit. Jumping Flash in real life!
486 crashes are probably more to do with the software, and/or the associated system (ram, NB/SB/etc..)
The question of 486 vs MIPS3000, it probably came down to power consumption as you ay, but also performance. the MIPS3000 does about 5x's the performance of a 486 DX 66mhz in floating point. I would assume FP is the types of calculations needed for navigational calculations.
Downclocking would also be about reliability and power consumption.
At least they didn't need to buy the proprietary memory card to save their data.
Mihos
Gold Member
course corrections are usually planned from earth months in advance, I would think it's primary focus would be data compression.
DeepEnigma
Gold Member
Had one as well. It was a beast, until you had to reboot almost daily for running out of virtual memory since they lacked the ability to flush the ram.
krypt0nian
Banned
Yeah you're not really in a position to call this silly.
When they started up NH probe, they heard the booming PS1 boot up sequence.
Patrick S.
Banned
Remember, the enemy's gate is down.
Robottiimu2000
Member
I have no idea what is the power requirements for pentium or 486, but they said somewhere the whole craft uses 28W to work. Might be that it's a question of energy consumption.
Edit:
Yup
Source: http://mobile.extremetech.com/lates...s-space-probe-powered-by-playstation?origref=
Can anyone explain why they use this instead of a newer part? It says that they use older parts for reliability, but wouldn't it be possible to make a newer part that's more reliable, more powerful, and more power efficient? Or does it become a matter of cost at that point?
Mr. Wedgewood
Member
So is pluto a "real" planet again? Or they needed to use a Saturn cpu to figure that out?
Well I would guess they want to go for a part that's been tried and tested over the years. How do you demonstrate that a newer part is more reliable than something that's been in use for many years?
A very crucial part is that old processors are known to work.
This doesn't really apply to new, much more complex cpus. Google for Intel cpu errata, Intel is quite open with this. Compilers know often these bugs and create code to circumvent them. Also it is possible to update the microcode for processors since several years.
So it's not only power and power efficiency but also stability and a predictive behavior that makes people choose old hardware.
Robottiimu2000
Member
Designing new parts and testing those parts is a very long process and costs. For this mission they had only 4 years to catch the Jupiter window for much faster approach to pluto.
And when it comes to space exploration, a piece of hardware you know that works and you can rely on, is more than worth it's weight on gold, since the operation costs easily several hundreds of million dollars and possibly decades of work, you really want your vehicle to be as fail safe as possible and to work.
Sir Abacus
Member
The RAD750 is a 10W board. The Mongoose-V is a really lower power board in comparison. When you're away from the sun relying on an ever lowering output RTG every watt counts.
The R3000 instruction set is tiny. 50 instructions in total IIRC and has a very limited number of addressing modes. A tiny number of instructions compared to the 486 which has hundreds plus four addressing modes. More instructions, larger decodes, more powerful decoder, more microcode, larger chip, more power, more bugs.
R3000 programs are easier to code, easier to debug, easy to step through, easy to know what will happen. You don't exactly get multiple shots at this stuff so its kind of imperative you're able to know with some measure of certainty that what you put there is what's going to happen.
The other thing is that VxWorks doesn't support the 486. This RTOS is legendary for being absolutely fucking bulletproof and it's NASA's go to off the shelf OS for that very reason. It doesn't support any earlier than a Pentium. But its been on MIPS for decades.
The R3000 instruction set is tiny. 50 instructions in total IIRC and has a very limited number of addressing modes. A tiny number of instructions compared to the 486 which has hundreds plus four addressing modes. More instructions, larger decodes, more powerful decoder, more microcode, larger chip, more power, more bugs.
R3000 programs are easier to code, easier to debug, easy to step through, easy to know what will happen. You don't exactly get multiple shots at this stuff so its kind of imperative you're able to know with some measure of certainty that what you put there is what's going to happen.
The other thing is that VxWorks doesn't support the 486. This RTOS is legendary for being absolutely fucking bulletproof and it's NASA's go to off the shelf OS for that very reason. It doesn't support any earlier than a Pentium. But its been on MIPS for decades.
They knew that since there is no gravity in space, it would work 100% of the time!
It is a radiation-hardened version of the R3000. So while the basic architecture is old, the part itself is (comparatively) new. The R3000 has the advantage of being comparatively small and simple, requiring little power, and yet delivering decent performance within those constraints.
MIPS is widely used in embedded systems, so there are up-to-date tool chains for development, and it isn't hard to find people who know the architecture.
Oh did you report you trips to Mars and Venus to NASA? they could have used you pentium 4.
https://en.wikipedia.org/wiki/Radiation_hardening
The article should have explained why they used an older CPU. Also the probe was launched 9 years ago, so that guy's P4 was still a somewhat modern CPU, but it would have caused the probe to melt.
Also most space probes have nuclear generators on them.
The article should have explained why they used an older CPU. Also the probe was launched 9 years ago, so that guy's P4 was still a somewhat modern CPU, but it would have caused the probe to melt.
Also most space probes have nuclear generators on them.
PdotMichael
Banned
Why investing billions in a new chip family if you can just buy the spaceproof version of off-the-shelf product family - while knowing that there is a production line avaiable in the near future.
andrerobot
Member
What a beautiful post!
Sidewinder
Member
If Shenmue fans can get what they want, then so can we!!
I've read an old article recently about some Psygnosis dudes, who were talking about Colony Wars and one of them said he repeatedly asked Sony to make a new one, but they told him there was no market for space operas anymore... but I'd say that now things have changed!
I'll search the article.
Edit:
Here it is, very interesting stuff, you should read it if you haven't yet and the last comment on the page too:
http://www.gamasutra.com/view/feature/134865/how_colony_wars_came_to_be.php
This one too:
http://www.gamasutra.com/blogs/IsaiahTaylor/20111001/8562/Colony_Wars_And y_Satterthwhaites_Side.php
RurouniZel
Asks questions so Ezalc doesn't have to
Shoulda used a Saturn processor. ;P
privitsan02
Member
Gotta love internet armchair analysts.
T0KY0ST0MP
Banned
PS1 - Beta Tested in the Future
But p4 lasts a decade! I kid. Should let FZZpure off the hook now.
Nintenyearold
Neo Member
GAF has expanded its range into armchair embedded systems engineering, well done 
Gotta keep us occupied somehow or we'll have to start playing games.GAF has expanded its range into armchair embedded systems engineering, well done
If a component handles the task, "more" isn't necessarily more. And from the sounds of it, the power efficiency wasn't an engineering challenge.
davepoobond
you can't put a price on sparks
It worked, and somehow it is silly?