Re: Shielding
Yes.
Any further questions?
3368 publicly visible posts • joined 26 Jan 2010
Perhaps, but I expect this weapon is mainly aimed at cheap amateur-built drones which could be launched by terrorists. Even if they can navigate by dead reckoning (e.g. inertial, plus optical map correlation), the article says it is "effective against targets that are resistant to electronic warfare" suggesting that it is powerful enough to "fry" some components, but I suspect that applies only if those components are unshielded or very poorly shielded.
If they were trying to take down a military drone such as a Shahed, or a drone that was designed with resistance to RF directed energy weapons in mind, then they would use the laser, or failing that, a missile.
The only effective countermeasure would be to have a drone that can navigate with 100% dead-reckoning (no GNSS) and no remote control or telemetry (and plenty of shielding around wires and electronics, optoisolated components, etc). Such drones would have very few legitimate civilian use cases.
@Jou(Mytzplyck > Unless you operate it with a long optic fibre you cannot shield it.
Half-right, for the wrong reason.
You could use Fibre Optic Gyroscopes, which are much more accurate than MEMS IMUs (and more immune to noise) and would allow for inertial navigation without GNSS, but these sort of things are bulky and expensive, more commonly found in cruise missiles than drones
If he keeps mugging off everyone around him, including his own supporters, civil servants, corrupt officials, gangsters, foreign leaders and powerful businessmen who have more money than he ever will, then he is going to find himself wearing a pair of concrete boots at the bottom of the Potomac before the midterm elections
He's not the Messiah, he's not even the Antichrist, he's just an arrogant, treacherous, very naughty boy who will get what's coming to him eventually.
The sad thing is: While we, a bunch of relatively well-educated nerds, sit here laughing at how utterly frivolous and doomed to fail these hype-train technologies are (Fusion Power, Quantum Computing, AGI), the UK so-called Labour government is spaffing hundreds of millions of taxpayer pounds on each, while slashing welfare, social care, education, etc..
"The last 14 years of Tory government has completely wrecked the UK! .... So we'll just carry on with all of their policies then"
Well indeed, but the core of the Sun is so huge that it has the approximate volumetric power density of a compost heap... About 300W/m3
Which is why Earth-bound fusion is so difficult - we need pressures and temperatures orders of magnitude higher than those found in the core of the Sun, to produce any useful amount of power in a relatively compact space.
I worked at Culham for 5 years, I left just before Boris turned up..
5 years was more than enough to realise that Fusion is never going to be useful for anything more than an interesting science experiment.
Transporting gigawatts from a marginally-stable 150-million-degree plasma across a vacuum by sheer intensity of ionising and transmuting (neutron) radiation hardly seems like a wise, never mind economically feasible means of generating power.
It's not surprising that so much Fusion funding comes from Oil giants - anything to keep those pesky nuclear boffins occupied so that they don't come up with a fancy new Fission design
Well, I don't think we'll ever have "Quantum Power", it being a field of rather low-energy things ...
It depends on your definitions, of course. Both "fusion power" and "quantum computing" are already here by some measure, but neither live up to the traditional definitions of power generation or computational utility.
If you mean quantum computing "able to break AES encryption faster than non-quantum techniques" versus fusion power "able to produce (controlled) power as cheap and plentiful as fission", then my wager is that Quantum is going to win that one. Assuming WWIII doesn't send us back to the dark ages first.
This.
I bought a little Schneider UPS from scamazon, but so far it has had a lower reliability factor than my house's electricity supply (quite an achievement, given my over-sensitive RCD and its propensity to trip if I scorch a naan bread in the toaster).. It is is currently bypassed because it threw a wobbly for no apparent reason yesterday (continuous tone, light flashing, no way to shut it up without shutting everything down - it was only on about 1/3 load).
Curious as to what counts as "Too much information" from your Eaton UPS though? Was it talking about its piles?
It's not just disk space though - having everything statically linked would also come with a serious performance penalty - with shared libraries they are cached in RAM once, whereas static libraries are loaded separately for each executable. Imagine the boot times when you are loading hundreds of copies of every library for every executable that gets called at boot..
> Thought it was for houses? Nah mate, data centres.
No doubt they will still quote 1GW(peak sun) as powering "one million homes", rather than 1 million scams and fake cat-videos per second..
And no doubt with the Trump trade-war, we will see even more dumping of cheap panels and naff inverters (but no cheap grid transformers, natch)
Nothing wrong with a bit of solar, but we can have too much of it (because it is so intermittent), and its place is on rooftops, car-parks, arid deserts, and maybe by the sides of the odd motorway, NOT on farmland
Worse, the datacentres have their own OCGTs and/or Diesel (which are afaik not counted in the overall generation reporting). We may end up in a situation where datacentres act as a "flexible load" by falling back to local gas generation when the grid is short on renewable power.
The trouble with that is, datacentres being so huge, have a very destabilising effect on local distribution networks - they trip over to backup generation at the slightest sniff of a voltage dip or spike, and they do it en-masse, which can cause localised blackouts.
On top of that, they are constant-power loads, which means that if the voltage dips slightly and they stay on the grid, their current draw rises by the same proportion. That would also be bad news for distribution network operators, and it's a pet theory of mine that the Heathrow debacle may have been caused/contributed to by the concentration of datacentres in West London. (a rapidly changing localised load puts a strain on Automatic Voltage Regulation, whereby a control system at the substation moves a tap on an autotransformer to keep the voltage stable, but too much movement of that tapchanger causes arcs which can contaminate the insulating oil and prematurely age the transformer, leading to an explosion if the insulation breaks down completely)
Shurely even Ed Milliwatt realises that you can't power a datacentre with an intermittent energy source.. But then again he is a nitwit..
Ideally, we would build some nukes to power the bullshit barns, then the AI bubble pops and the rest of us can use the cheap energy that follows. But nukes sadly take very a long time to build now, due to some rather excessive regulation which the fossil fuel industries lobbied for, to save themselves from nuclear competition..
What we'll probably end up with is more renewables, and regional pricing, so us plebs have to fork out a fortune for energy while the bullshit barns get it on the cheap if they build in Scotland. Then the incentive to modernise and upgrade the transmission network disappears (why invest money to alleviate grid constraints if you can simply charge more for electricity in constrained regions?), then something goes bang and all the lights go out. Welcome to Britain.
(Ok Crysis 1 has some quirks, but it was always finickety on Windows)
While I agree with the sentiment that COBOL shouldn't be whatsoever involved in any web request, there's certainly a non-zero chance that it is.
Perhaps a COBOL program is required to query an ancient database, and that query is used periodically and/or on request, to update a (more modern) intermediate database as mentioned above. Or worse, the website might have to call COBOL on the fly if it needs to update the core DB.
So, any change to the interface/schema between the website DB and the COBOL DB may require a change in COBOL
A manufacturing line for glass and/or solar panels is a lot more than just a heat source though - you'd need a container for the molten regolith, various gubbins for purifying/separating it into transparent glass and Silicon, something to pour it out onto (you can't float it on water like how we make sheet glass down here on Earth), more gubbins to chop Silicon into wafers and P/N dope it, etc etc. A solar panel manufacturing plant is a big, heavy, complex thing, and usually requires a LOT of water, which usually ends up highly contaminated and dumped in some river..
Something indeed fishy about that..
1, why would the user be "illuminating the screen" with an anglepoise lamp? External light on a CRT or any emissive display makes the picture harder to see..
2, i'm surprised at any significant magnetic field from a filament bulb.. While it is a solenoid coil, it has relatively few turns and low current (50W bulb is 430mA at 115VAC or 208mA at 240V) so not many Ampere-turns and no ferromagnetic core. The field strength drops off sharply with distance too, and the filament can't have been right next to the screen
More likely, i'd guess that the perhaps-ungrounded metalwork of the lamp was acting as an antenna, which was interfering with the TV/monitor's RF input
Am I the only one who has never heard of Peter Kyle, and therefore read that as Jeremy Kyle finally detailed the plans for the bill at length today.
One imagines him sitting the offending CISOs down on his sofa and giving them a dramatic talking-to about their sordid cybersecurity failures, for public spectacle.
Mr. Smith, we have conducted an AI code-authorship analysis, which has indicated that the hideous cyber-vulnerable SAP-to-SCADA connector was in fact YOUR bastard lovechild! <audience gasps> What do you have to say about that??
Unfortunately, when you take a big pile of statistics and try to N-dimensional gradient-descent some pseudorandom noise towards something statistically plausible:
"accurate", "repeatable", "controllable" are not adjectives normally associated with this activity...
"useful" applies only where you do not care about the above, and there are many governments around the world which apparently do not care about the "accuracy", much less "repeatability" of their law enforcement or military decisions, so long as they can be overridden by those in power i.e. themselves.
I assume sarcasm, so I'm not downvoting you, but:
On-prem servers typically don't include the sort of 100+ kW racks that can run full-size LLMs. We can run quantised/pruned models locally on GPUs and CPUs, but they produce lower-quality bullshit than the cloudy bullshit-machines, due to memory requirements.
On-prem servers are (typically) designed for business functions, and are sized according to business needs. CPU frequency/voltage scaling means that even "underutilised" servers don't actually burn that much power when they are sat idle, and they don't spin up to generate useless nonsense, or scrape the internet to find more data to generate useless nonsense from. The owner pays the electricity bill, and isn't going to want to waste money on frivolous shite.
Whereas for cloud servers, often the 'leccy is included in a fixed monthly rental, and the renters feel that they must maximise the usage of whatever they are renting (which may be a 100kW rack) even if that means wasting it on frivolous shite.
A habitual liar on Crack, more like..
If said Investors would put down their own crack-pipes then perhaps they would notice what everyone else can already see..
I wonder how much energy it has cost so far to watch the world's latest bullshit machine fail miserably at playing Pokemon (there is a twitch stream of Claude playing pokemon, I won't link it, it's incredibly dull, but it has been running for several days, and has got (checks) no further than a 5-year-old would have in much less time, and it has been playing continuously, with humans intervening to delete its data cache when it gets stuck..) If someone were to count the energy cost of that effort, i'm sure it would run into hundreds of tonnes of oil
This aint no technological breakdown.. Oh no, this is the road.. to Hell.
fyi, nuke plants -always- have means to dump the load. Their thermal efficiency (steam turbines) is 50% at best, so they are always dumping at least half of the 'load' via cooling towers (as do other thermal power plants). But they will size their cooling towers such that they have well in excess of 100% cooling capacity should there be a loss of load. And it is pretty common in practice for nukes to disconnect themselves from the grid at the slightest sneeze of a problem, given their safety regulations.
But of course you are correct, it would be madness (operationally) not to have a grid connection to be able to export the power when not used locally (or indeed to import it when the on-site nuke reactor is down for maintenance or whatnot). It just wouldn't be a safety issue.
Of course, this is not to say that nuke power -should- be used to power bullshit generators, only that it -could-.
Nobody seems to have noticed this ridiculous quote:
"A datacenter is now 250 megawatts. That's kind of the limit per rack. I think the rest of it is just details," Huang said. "If you said that a datacenter is a gigawatt, and I would say a gigawatt per rack sounds like a good limit."
Shome Mishtake? Shurely he meant a Megawatt per rack? (which is already ludicrous) or a Gigawatt per datacentre?
a Gigawatt per -rack- is just thermodynamically impossible (given that chips operate up to 125C at most)
Not to mention the power requirements.. Gigawatt-scale datacentres are putting massive strain on electricity grids.. Even if his fantasy datacenter contained only 10 such "gigawatt racks", the loss-of-load event caused by a program halting would be enough to cause blackouts.
Huang has really lost the plot with this one.. Watts are not like Bits and Bytes - you can't endlessly have more of them in a chip without consequence
Carry on up the Kyber Rack?
NVidia really are up shit creek if they can't improve performance per Watt, which is what matters
Same happening in China: Shiny new GPU datacentres sitting idle, because the money they are getting from renting GPUs is not worth the energy cost of running them.. Who'd'a thought it
https://www.tomshardware.com/tech-industry/artificial-intelligence/chinas-ai-data-center-boom-goes-bust-rush-leaves-billions-of-dollars-in-idle-infrastructure
Popcorn icon gravely needed
I'm sure "Sam Bell" would do..
Joking aside though, I think the best place for well-encapsulated, vitrified waste is in the sea - water is a very good radiation shield, and provides free cooling. And if you put it in the right place it will eventually be subducted back into the earth's crust from whence it came.
And for those who would still have a problem with that: Well, what do you think happens when a nuclear powered ship or submarine sinks.. And what do you think will happen if we can't solve the energy crisis? Nations will start fighting eachother (see icon) with REAL radiological consequences!
> Perhaps their speakers should have an accident
When noone is around, use an insulation resistance tester (e.g. megger) on the mini-jack plug. This will of course fry the the analogue front-end with no outward signs of damage. For bluetooth speakers, a couple of seconds in the office microwave would do the trick
I see - Page 40 of https://www.heathrow.com/content/dam/heathrow/web/common/documents/company/heathrow-2-0-sustainability/futher-reading/Heathrow_Net%20Zero%20Carbon%20Strategy_v13.pdf
But, this pamphlet is an oversimplification intended for pointy-haired bosses and not engineers; I don't think it's valid to take it as read that they _ONLY_ had generators for "runway lights" - it indeed says "predominantly for runway lights" but almost certainly would include other critical systems such as the ATC control tower, radar, critical IT systems, etc.
My personal theory is that in the hours following the substation failure, this clusterfuck was mainly exacerbated by grid-tie inverters, i.e. power sources such as the CHP, Wind, Solar and Batteries, which in normal conditions would merrily alleviate the load on the grid supply - until such a point as that grid supply is gone, and then they are borked, unable to sync to any upstream grid.
Then while the grid supply tries to come back on, they are unable to contribute any power. This is a microcosm of a major UK-wide issue, i.e. the feasibility of "Black Start"
The electrical engineers technicians would have to isolate _everything_ (including the EV chargers in the car park) until the total connected load was low enough that one single source could power the local grid, and then start up the grid-tied inverters one by one.
> A major energy analyst I had some professional overlap with some years ago and rate very highly, recently got hold of the full dataset for grid frequency.
Would that be Kathryn Porter, perchance?
I rate her very highly, too, and I hope they put her in charge of NESO (and then abolish/renationalise the DNOs)
It's a typical double-circuit mesh substation, it has two 275kV feeds, which may well come from different places, and two supergrid transformers which can be configured to use either incoming feed. Normally I think they would use both, but I don't think they would permit a load to be so big that it would instantly trip if they had to disconnect one circuit / transformer.
The problem yesterday seems to be that a transformer exploded violently, perhaps due to contamination in the oil - it may be 60-70 years old, and it uses oil for both cooling and electrical insulation, as well as paper for insulation. Contaminants can leach out of the paper if it is too hot for a long period of time, or if there are arcs e.g. during tap-changing. And the resistivity of the oil apparently drops with temperature too, so internal arcing can be catastrophic.
Because the fire was so violent and the built-in fire suppression so ineffective, they needed fire crews with hoses and cherry pickers, which meant it was not safe to simply switch over to the second transformer - both 275kV incoming circuits needed to be de-energised until the firefighters had finished, hence the length of the blackout
I've heard (via comments referencing someone speaking to the Daily Mail) that the load was around 106%, (I assume that means both transformers running at 53%?) but if they need to take one off for maintenance then the other one is going to get quite hot..