Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Cosmic rays causing 30k network malfunctions in Japan each year

mainichi.jp

Post reply on HN
Print view View original

61–70 of 120 posts

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#61

Post by kevin_nisbet »

belter wrote: post #44
kevin_nisbet wrote: post #21

Earlier quoted context omitted.

> 1) Use ECC memory Not exactly. When I was in telco, where I had this problem was in FPGA's, we had all ECC memory and I never linked any problems to bit flips in RAM. But as I remember, the FPGA's we had were using a type of SRAM cell, but because it's not a memory module the FPGA programming could bit flip. So the product had a checksum function, that read back the program on a cycle and reset itself if the progra…

Interesting and makes sense. Do you have any additional references you would suggest and particularly in the context of FPGAs ? Would you say this quick reference is a good overview ? https://www.intel.com/content/dam/www/programmable/us/en/pdf...

Sorry, I should have mentioned this was quite a few years ago, so I'm very out of date. So I don't have any known good references that are handy. That link you shared seems pretty good on a quick scan through and inline with what I remember, I'm pretty sure I dug up similar resources for other vendors, including one I think was looking at satellite hardware.

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#62

Post by shiftpgdn »

oconnor663 wrote: post #43
digikata wrote: post #38

Earlier quoted context omitted.

A reduction in floor and building vibration might have contributed too?

Or for example, old cables and improperly installed cooling fans might tend to get fixed during a move. Hard to know for sure.

Well yeah, that's why it's presented as an anecdote. :)

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#63

Post by bostonsre »

andi999 wrote: post #54
typon wrote: post #50

Earlier quoted context omitted.

Humans going to Mars should expect to never come back

But at least they should reasonably expect to arrive.

Anyone know what expectations Columbus or other expeditions had about survival/risk?

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#64

Post by roamingryan »

kevin_nisbet wrote: post #21
belter wrote: post #10

They have to do two things :-) 1) Use ECC memory 2) Go underground "One experiment measured the soft error rate at the sea level to be 5,950 failures in time (FIT = failures per billion hours) per DRAM chip. When the same test setup was moved to an underground vault, shielded by over 50 feet (15 m) of rock that effectively eliminated all cosmic rays, zero soft errors were recorded.[6] In this test, all other causes o…

> 1) Use ECC memory Not exactly. When I was in telco, where I had this problem was in FPGA's, we had all ECC memory and I never linked any problems to bit flips in RAM. But as I remember, the FPGA's we had were using a type of SRAM cell, but because it's not a memory module the FPGA programming could bit flip. So the product had a checksum function, that read back the program on a cycle and reset itself if the progra…

Many modern FPGAs now include dedicated logic for config SRAM "scrubbing." This logic continuously checks config frame checksums to identify upsets. These can then be fixed in real-time either using the error correction properties of the checksum technique, or from the non-volatile config memory (typically NOR flash). It's also important to note that only a subset of the SRAM config bits are critical for a given application. Usually this is a small percentage of the overall array.

https://www.xilinx.com/support/documentation/application_not...

If even higher levels of reliability are needed, there are rad-hard-by-design FPGA families (e.g. Xilinx Virtex 5QV). These have a special config SRAM cell that has more charge storage sites than a conventional SRAM cell. It is less area efficient than a conventional SRAM cell, but geometry of the charge storage sites ensures that a single cosmic ray can't flip the state of a majority of them. Essentially the cell can self-correct, no scrubbing required.

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#65

Post by moistbar »

deepsun wrote: post #13
stuartcw wrote: post #4

I have always wondered whether new ideas are seeded by cosmic rays disturbing neurons in the brain..

That'd mean people underground (or under thick layer of concrete, like in skyscrapers) are not creative.

I guess you could say they'd be...

Living under a rock.

Yeah, yeah, I'm leaving. Put the weapons away...

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#66

Post by Sanzig »

andy_ppp wrote: post #35
notum wrote: post #28

Earlier quoted context omitted.

Rays hit our planet regardless of Sun, if anything it might block a few. They are caused by ultra-energetic cosmic events, such as star collisions, black hole feeding cycles. Cosmic rays are just very potent photons, capable of knocking out electrons from an atom (meaning they are ionizing) causing havoc in precision electronics and, well, our DNA.

What is the tolerance for cosmic rays for a human... I ask because on the journey to Mars you’ll get quite a few I guess and also on the planet’s surface after you’re there with such a thin atmosphere?

That's actually a fascinating question! It turns out we don't know enough to answer one way or another.

The model commonly used in radiation protection to assess cancer risk with respect to radiation dose is called the Linear No-Threshold (LNT) model [1]. The model critically assumes that (1) total radiation dose is the only predictor of cancer risk, and (2) any radiation exposure results in an increased cancer risk.

This model works at high absorbed doses, however, its applicability is highly controversial when used with low abosorbed doses or with relatively high absorbed doses that occur over a long period of time (ie: low dose rate).

The thing is, the human body has built-in defense mechanisms against cancer such as DNA repair. There is a good body of evidence that small doses and low-rate exposures do not result in cancer risk (ie: there is a threshold absorbed dose and probably also a threshold dose rate), but the model does not account for this.

This is particularly problematic when trying to assess excess mortality from things such as radiological accidents: when you multiply the small LNT-predicted risk for a low dose times a very large population, you end up with a lot of cancers. This is one of the reasons you'll see estimates for deaths from the Chernobyl accident vary by orders of magnitude.

It's also problematic when assessing something like a Mars mission: yes, the astronauts would get large cumulative doses, but the dose rate is pretty low over most of the mission (other than during high dose rate solar events where they would need radiation shielding). How much of an elevated cancer risk is it actually? Nobody is quite sure.

[1] https://en.wikipedia.org/wiki/Linear_no-threshold_model

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#67

Post by tinco »

bostonsre wrote: post #63
andi999 wrote: post #54

Earlier quoted context omitted.

But at least they should reasonably expect to arrive.

Anyone know what expectations Columbus or other expeditions had about survival/risk?

A lot less than an astronaut has when going to Mars for sure. We at least know how far away Mars is exactly, and what obstacles lie in the way, and what inhabitants it has. They just got on a boat, and pray for fair wind and good weather so they might reach india, and thank god there was a whole other continent in between.

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#69

Post by iso1631 »

bostonsre wrote: post #63
andi999 wrote: post #54

Earlier quoted context omitted.

But at least they should reasonably expect to arrive.

Anyone know what expectations Columbus or other expeditions had about survival/risk?

For a normal NASA flight

"NASA told Business Insider it estimates there’s a 1-in-276 chance the flight could be fatal and a 1-in-60 chance that some problem would cause the mission to fail (but not kill the crew)"

That's fairly low - feels like the age of sail would be far higher -- Columbus's second voyage alone had about 25 deaths just from scurvy. His first voyage was only about 90 crew.

But remember that life wasn't exactly easy on land either.

Re: Cosmic rays causing 30k network malfunctions in Japan each year

#70

Post by woodruffw »

notum wrote: post #16

Cosmic rays are, or rather were, the reason why manufactures avoid shipping high end cameras by plane, cosmic rays kill pixels in imaging sensors: https://youtu.be/98FZ8C6HneE?t=479 It's also the reason why all ISS footage has dead pixels on it.

This is a fascinating claim (and video), but I can't find a ton of corroborating evidence for it: searches for "digital camera cosmic rays" bring up (1) considerations for cameras in space (much higher background radiation levels than a commercial flight), and (2) consumer photography forums where people are discussing that 2011 talk. Some of the latter seem to think that Kodak has/had a vested interest in spreading fear around digital photography, since their analog division was imploding at the time.
Post reply on HN