On board "immune system" lets chips route around damaged components.
Electronic components are delicate. Get them too hot or too cold and they crap out. Hit them with a hammer, and they tend to stop working. Then again, the same thing can be said for people. Much like people, microchips may soon be capable of limited self-repair.
A team of researchers at the California Institute of Technology (Caltech) claims to have developed a system for chips that allows circuits to route around damaged components while maintaining near-optimal efficiency.
In a demo of the tech, they blasted a millimeter-wave power amplifier circuit with a laser, destroying several of its components. Normally, the circuit would become useless if just one of its components were compromised. Instead, an onboard ASIC (Application-Specific Integrated Circuit) monitors the circuit's, voltage, power and temperature. If any of those variables change, it shifts through some 262,144 possible configurations to find an optimal solution that accounts for the damaged components.
"We don't know all of the different things that might go wrong, and we don't need to," said graduate student Steven Bowers of Hajimiri's lab at Caltech. "We have designed the system in a general enough way that it finds the optimum state for all of the actuators in any situation without external intervention."
In the demo, the chip could diagnose and account for a damaged component in 0.8 seconds with the ASIC running at 50MHz. At 200MHz, self-healing took a maximum of 0.2 seconds. Of course, outfitting every circuit with the custom-ASIC would cost a pretty penny, but the benefits would be immense for circuits designed to run for long periods without human servicing (unmanned space exploration is the first potential use that comes to mind).
So, in short; When the robots come to harvest your supple human flesh, hitting them with a hammer probably won't work, and even if it does, they'll just heal straight away anyway. Thanks, science.
First the cloud mind and now this? We're doomed. It'll be pretty cool to see this when i's finally applied to normal stuff. See what little problems it fixes. :D
There sure is a bunch of articles recently detailing how robots will be able to keep going if say...there was some resistance damaging them. Pretty sure we are in trouble and its 100% our own fault.
As far as I can tell, they aren't capable of repairing themselves. They can just figure out how to survive without the damaged parts. The double-tap rule should deal with such advancements.
Well, only one thing to do.PHA+PGltZyBzcmM9Imh0dHA6Ly9pbWFnZXMud2lraWEuY29tL3RoZWFtYXppbmd3b3JsZG9mZ3VtYmFsbC9pbWFnZXMvOC84OS9GdW5ueS1naWYtbWFuLWp1bXAtb3V0LXRoZS13aW5kb3cuZ2lmIiBhbHQ9ImltYWdlIi8+PC9wPg==
Daaaah Whoosh: As far as I can tell, they aren't capable of repairing themselves. They can just figure out how to survive without the damaged parts. The double-tap rule should deal with such advancements.
Also since it's a repair chip monitoring another chip, there's no repair chip to repair the repair chip when it croaks. Unless they stack that sh*t all over the chip.
I dunno... what if it heals incorrectly? Wouldn't it be better to just have your CPU fail altogether instead of possibly healing incorrectly and bricking your motherboard or hard drive?
Guys, if a robot uprising occurred, don't you think they'd invent these technologies for themselves even if we hadn't?
The thing about robots is that they wouldn't be impeded by the kinds of handicaps we have: politicians, pundits, lawyers, marketers, journalists, etc. They'd outperform us technologically, wipe the floor with us after we picked a fight with them, and then go on to do all the stuff that we geeks would already be doing if we were similarly unburdened.
In a way, they'd be the children who get to build a life for themselves better than what we ever could. Isn't that a good thing?
Jamous: First the cloud mind and now this? We're doomed. It'll be pretty cool to see this when i's finally applied to normal stuff. See what little problems it fixes. :D
we are only doomed if we start threating them like pieces of shit.
the opposite of this lead to 60% of all the robot versus humanity media in existence.
McMullen: I dunno... what if it heals incorrectly? Wouldn't it be better to just have your CPU fail altogether instead of possibly healing incorrectly and bricking your motherboard or hard drive?
Guys, if a robot uprising occurred, don't you think they'd invent these technologies for themselves even if we hadn't?
The thing about robots is that they wouldn't be impeded by the kinds of handicaps we have: politicians, pundits, lawyers, marketers, journalists, etc. They'd outperform us technologically, wipe the floor with us after we picked a fight with them, and then go on to do all the stuff that we geeks would already be doing if we were similarly unburdened.
In a way, they'd be the children who get to build a life for themselves better than what we ever could. Isn't that a good thing?
Yeah, this is more like rerouting traffic to avoid a broken bridge. Damn clever though, they do a better job of it than our traffic control systems do, I've yet to see a jam sorted out in 0.8 seconds :P
The thing about robots is that they wouldn't be impeded by the kinds of handicaps we have: politicians, pundits, lawyers, marketers, journalists, etc. They'd outperform us technologically, wipe the floor with us after we picked a fight with them, and then go on to do all the stuff that we geeks would already be doing if we were similarly unburdened.
In a way, they'd be the children who get to build a life for themselves better than what we ever could. Isn't that a good thing?
They are called Cylons and this has already happened. we live on the 13th world, remnants of the failed Caprican civilization. and we do the mistakes all over again.
Groenteman: Just install Windows on all the robots and they will be about as logicaly crippled as humans.
or you could install MAC and then you would need to insert counts to unlock certain parts of its body. that is, if those parts wouldnt have stayed ddorman for 2 weeks beucase you know theres an update and you gota buy new becuase we just made your old one unusable.
Groenteman: Just install Windows on all the robots and they will be about as logicaly crippled as humans.
or you could install MAC and then you would need to insert counts to unlock certain parts of its body. that is, if those parts wouldnt have stayed ddorman for 2 weeks beucase you know theres an update and you gota buy new becuase we just made your old one unusable.
Heheh, thats right, Linux is the only real threat here. I knew that penguin looked fishy!
Genocidicles: So I guess this advance will just be ignored because it'll ruin planned obsolescence.
Correction: Microsoft will balk against it, but then competitors might see it as a golden opportunity to take the market from THEM, and because the things are so powerful, anyone lagging behind will get dragged kicking and screaming into the next generation of computers.
I'm not worried about a robot apocalypse, though. Uhhh...how can I put this in a delicate way? We're not able to produce proper intelligence in a machine for that.
"Self-Healing" Microchips Bring Us Closer to The Robot Apocalypse
On board "immune system" lets chips route around damaged components.
Electronic components are delicate. Get them too hot or too cold and they crap out. Hit them with a hammer, and they tend to stop working. Then again, the same thing can be said for people. Much like people, microchips may soon be capable of limited self-repair.
A team of researchers at the California Institute of Technology (Caltech) claims to have developed a system for chips that allows circuits to route around damaged components while maintaining near-optimal efficiency.
In a demo of the tech, they blasted a millimeter-wave power amplifier circuit with a laser, destroying several of its components. Normally, the circuit would become useless if just one of its components were compromised. Instead, an onboard ASIC (Application-Specific Integrated Circuit) monitors the circuit's, voltage, power and temperature. If any of those variables change, it shifts through some 262,144 possible configurations to find an optimal solution that accounts for the damaged components.
"We don't know all of the different things that might go wrong, and we don't need to," said graduate student Steven Bowers of Hajimiri's lab at Caltech. "We have designed the system in a general enough way that it finds the optimum state for all of the actuators in any situation without external intervention."
In the demo, the chip could diagnose and account for a damaged component in 0.8 seconds with the ASIC running at 50MHz. At 200MHz, self-healing took a maximum of 0.2 seconds. Of course, outfitting every circuit with the custom-ASIC would cost a pretty penny, but the benefits would be immense for circuits designed to run for long periods without human servicing (unmanned space exploration is the first potential use that comes to mind).
So, in short; When the robots come to harvest your supple human flesh, hitting them with a hammer probably won't work, and even if it does, they'll just heal straight away anyway. Thanks, science.
Source: The Register
Permalink