Posts by cyberdemon 1895 publicly visible posts • joined 26 Jan 2010
> What mechanism other than a ban would mean "99% of websites were gone tomorrow"
Oh, I don't know, Global Thermonuclear War?
Far more likely than any effective ban on datacentres!
At this point, a "political accident", "cuban missile crisis alt. ending" style, is looking a lot more likely than any civil nuclear accident
*...and yes, I know Nuclear power plants don't do that as a rule, but for the sake of this example it's a thing that people are afraid that they will.
But non-nuclear industrial plants can and do. Yet we don't have the same level of hysteria about those.
Even a Carbon Capture and Storage plant has the potential to asphyxiate everyone in a half-mile radius if it goes horribly wrong in calm weather.
The specific hysteria about Nuclear is so stark that i think it has to have been pushed by a combination of CND types and Big Oil in the background. That and The Simpsons hasn't helped.
The ONR (UK office for nuclear regulation) is staffed by a lot of CND types, as well as busybodies and jobsworths who love making work for themselves and other people.
This is a symbiotic relationship as the CND anti-nuke types can use the busybodies to push up the cost of anything 'nuclear' to astronomical proportions, while the busybodies use the CND types to justify their own miserable existence.
Meanwhile there are few people left at the ONR who actually have a clue about nuclear engineering. I refer you to Page 39 of the latest Private Eye (#1615) for a column by 'Old Sparky' on the current exodus of staff from the ONR.
This is why Sizewell and Hinkley are both billions over budget and a decade late - because they are both on the hook for thousands of engineering changes ordered at the last minute by the ONR, which itself has no accountability against being obstructive
So I agree with Mr Semicolon's sentiment. Also I think you needed to explain the meaning of "Do one" in the modern British vernacular to our left-pondian friends.
> Thorium reactors may be an option in a decade, but is clearly not available now.
For molten salt Thorium I tend to agree, they have a big issue with contamination, and the materials involved are really nasty.
The tech that 'excites' me the most though is the Accelerator-driven subcritical reactor. I wonder what happened to Aker's ADS reactors?
https://en.m.wikipedia.org/wiki/Accelerator-driven_subcritical_reactor#Rubbia_design
Security in a computer science context should be things like memory safety, race conditions, process isolation. That stuff IS covered by CompSci degrees (certainly the one I took anyway) despite the alarmist headline.
What should NOT be part of a CS degree are things relating to specific software technologies. Managing user behaviours etc. My CS course had a module on operating systems which included how to set up a generic UNIX-like system, but it did NOT go into the detail required to actually set up a robust UNIX system, that's for the IT course.
Similarly, Firewalls SHOULD and are covered, but how to properly configure a Cisco one should NOT. That's for the Cisco-sponsored IT course.
At this point they need bullshit talkers to massage Mr Trump's arse so that he might change some rules in their favour, otherwise this is unlikely to happen, no matter how many engineers and technicians they had.
Also, investors are prone to believing in bullshit press releases, because they are too busy snorting cocaine to listen to anybody technical
Pretty sure the only people allowed to operate the reactors on navy subs are.. The navy.
That and any civil use of the thing would come under different rules.
But then again, in a few months we could have Trump back in charge, so laws, rules etc are for sissies and losers, so you could be right. I'm sure Zuck wants one too for his bunker
> with fusion each joule puts >2J into the environment,; with wind/solar etc the overall change is zero.
You seem to be arguing about the heat produced by human energy production/consumption, in the context of global warming?
This is negligible. Total energy use (including renewables) is about 30TW. Global insolation, (energy input from the sun), is 175PW. So you're arguing about 1 part in 6000, and we haven't even factored geothermal energy yet.
And even your assertion itself is wrong: If you were to take a hundred hectares of field or desert and cover it with solar panels, have you not changed its albedo?
Well, I'm sure OpenAI et al would tell you that a machine brain is many times more useful than a human brain of the same intelligence - because the machine brain, once trained, can be used to make millions of decisions simultaneously. So it only needs that 936 MWh once, to replace a million human brains that needed 3.5 TWh between them to train, and those weren't even consistent.
They would also tell you that it is not necessary for the AGI to be a superbrain or to understand the physical constraints of our world. It just needs to be omnipresent and people will worship it like a God. And it would be a God that they control.
Where I disagree with your post though is where you say "What is needed is ... ". I don't believe that AGI is "needed" at all. If it does ever come along, it will be one hell of a threat.
When I first played Deus Ex 23 years ago, I picked the ending where JC Denton merges with the Helios AI and becomes God.. Now in my late 30s, I would pick the ending where JC Denton blows up the whole complex and shuts down the Internet for everyone on earth. Let's go back to living in villages.
Maybe when I am older still, I will be grumpy and cynical enough to prefer the ending where we just kill Elon Zuckerborg Bob Page, and let the old secret government keep the AI for itself and do what it has always done.
> it is taking energy locked up shortly after the Big Bang and dumping it into our environment.
Isn't that exactly what the Sun is doing? :P It won't last forever either.
Fusion has the potential to make its own fuel, to turn Lithium into Tritium + some other stuff. There is a mind-bogglingly huge amount of energy locked up in matter (see Einstein's most famous little equation) so we would never, ever run out if we were to make energy that way. Fusion is as renewable as the sun is. The sun itself is fusion after all.
However, it is not free of chemical or radioactive pollution. There would be lots of that from fusion power. More by weight and volume than fission waste, but fewer active materials, not as long-lived, not as toxic.
It's also quite impractical to get any useful energy out. Unlike fission where we can just run a coolant loop straight through the reactor, fusion has to happen at millions of degrees, and no material can survive that, so it is done in a vacuum. How do we transfer Gigawatts across a vacuum? It's tricky, especially when so much of it comes as neutron radiation.
In any case, it's looking very likely that we will have blown ourselves up with within the next few years, so it's all a bit academic.
Installing your own meter downstream of, or even inline (with a current clamp) is not "messing with" the supplier's meter, it is just checking its accuracy.
If/when they tell me they are changing my meter, I will tell them that i disagree with the smart metering terms, so if they want to put in a new meter, it will have to be a dumb one.
Well, it's not just for load-shedding. It's also so that they can remotely flick customers' meters into Pre-Payment Mode, because it's easier than obtaining an increasingly-unpopular bulk court warrant to forcibly enter hundreds of properties to install pre-payment meters.
You say that you were given no choice but to have one. Does that mean they obtained a warrant to forcibly install it, or did they just convince you over the phone that you had no choice?
There are also issues with smart meters over-reading. I think it will get interesting if the UK does change the law in light of the Post Office scandal that computer systems are not automatically trusted.
There are a lot of people who have been getting spuriously high bills after having a smart meter, e.g. a pub near me had to close after receiving an electricity bill £300k higher than it should have been They queried it and got an electrician to install their own meter, which showed the smart meter was overcharging by more than a factor of 3. They are now disputing the claim with the supplier and have reopened for now.
They called it a "Programming Error", however My theory is this: The pub in question says that it has no electric heating, but it has a large quantity of big-screen TVs and LED lighting. These loads have bridge-rectifiers on the front-end and no active-PFC, which means they only take a pulse of energy at the peak of the voltage waveform (when the meter is measuring) and no energy anywhere else. The meter assumes a sinusoidal load current, and if it is not, then it will over-read.
Because of this, smart meters have been seen over-reading by as much as a factor of 6 for certain LED lights, for example.
That said, I would hope that more modern meters have resolved this er, "programming error" by sampling continuously, not just at the peak, and integrating hundreds of samples over each mains cycle.
Doesn't need a YouTuber to tell you that remote-disconnect contactors are a feature of all smart meters, it's right there in the spec.
5.6.3.11 Disable Supply
A Command to establish a Locked state whereby the Supply is Disabled and can only be Enabled or Armed in response to a Command to Arm the Supply (as described in 5.6.3.7) or Enable the Supply (as described in 5.6.3.12)
In executing the Command ESME shall be capable of setting the Supply State(5.7.5.32) to Disabled.
The current market is fueled by unsustainable subsidies at both ends: China dumping loss-leaders, plus Western governments giving out subsidies and tax breaks for Net Zero.
But when that system of subsidies inevitably ends, EVs will be the only option, and so only the rich will drive.
Quality will come down for a while, as western EV producers are put out of business by the loss-making Chinese ones, then the prices will go back up, but China has to recoup its loss somehow, so the quality won't improve.
Here in the UK, 400kV transmission losses are about 5%, and distribution losses average 15%, so 20% overall. But that will get much higher as we increase the load and then try to "fix it in software" by throttling chargers to avoid overloading substations, instead of building more substations. Because that means lines and transformers will spend much closer to 100% of their time at max load, and losses go with the square of load
I thought that in America, the 110V lines were very short, i.e. from a pole mounted transformer outside each property.
Making them literally a mile long would be very silly indeed.
Not sure if you understood what I meant or not, but that's useful info
A Diesel genset is around 28% efficient.
A lithium battery is ~90% efficient depending on charging speed etc, and an EV traction motor can be as much as 95% efficient.
If a charging station is using a diesel to run the charging points (because it can't get enough power from the grid, as JE claims happens fairly regularly), then it has 0.28*0.9*0.95=24% overall efficiency, compared to your 33% for driving a Diesel car directly.
Therefore in this (perverse) case, the EVs have significantly more direct CO2 emissions than the Diesel cars filling up at the same station
24% vs 33% would be a factor of 0.27, i.e. even worse than my original guess. It depends on the real efficiency of the genset vs the car of course, which will be variable, but it confirms what should have been obvious: using a generator to charge an EV is utterly backwards, and I really hope it doesn't happen as often as JE says it does, but without the infrastructure, that's the only way to get "superchargers everywhere"
And the efficiency of charging an EV from a Diesel genset, as compared to driving a Diesel car is..?
I would guess at least 20% loss, for the combined (in)efficiency of an extra generator, transformer, charger, battery, and motor.
More of course, if the genset is still sat idling away when no EVs are charging.
Yeah, it is ridiculous that you need an App. You should be able to tap a card to authorise up to £X.
But then you'd probably have to get a parking ticket if your £X ran out and your car was still sat there blocking the charging bay.
Not sure how an App helps that. But hey, they can foist anything they like onto Early Adopters.
Ah, road charging.
Yeah, thought they might do that.
And of course, everything happens first in Australia, then in the UK, then in the US and Europe. It's like Oz and Blighty are the world's guinea pigs for the authoritarian invisible hand..
But it's fine, cos nobody will need to drive more than 15 minutes, right?
> If EV's start to have 400mi range as standard, and Superchargers everywhere (I'd bet real money on the latter happening)
I'd bet real money on those superchargers never appearing or never being connected to the electrical grid, because a supercharger station needs at least a 1MVA electric substation, HV pylons, that need to be connected to a bigger substation, and so forth. The infrastructure to support this at a wide scale does NOT exist and is ridiculously expensive to build. It's also unreliable and vulnerable to disruption, especially as more renewables enter the mix.
Sorry, but that's utter bollocks
The power plant has an efficiency of 30-40% if it's gas, less if it's coal. The electric distribution grid takes another 10-15%. The battery another 5-10%.
It's not a fair comparison to compare the electrical kWh in a battery to the thermal kWh in a fuel tank. EVs shift their emissions and losses elsewhere.
The report said that the chargers weren't working, not the car. I would expect that the charger would be able to power the battery heater if it was working. But if the charger had temperature sensors in its electronics that went into fault at -40C then it would explain what is going on.
(e.g. MAX30210 temperature sensors will not read below -40 C and will have an error flag instead, who knows what temperature sensors are inside any given EV charger, but they will be built for a certain temperature range)
Somewhat surprising that the charger couldn't power a resistive heating element in the battery coolant circuit, if that were truly the case that the battery is too cold and is refusing to charge.
It's also possible that the charger itself has an "error" with its temperature sensors. i.e. "sensor out-of-range, assume faulty".
Your EV already has said instruments (coulomb-counting is necessary to estimate the battery's state of charge etc.) and the means to report the data (5G connectivity in every car). The rest is just software and legislation.
EVs cause a lot of potholes and represent a huge hole in tax revenues compared to the fuel duty and vehicle excise duty that are levied on ICE vehicles. Governments WILL start taxing them when they have finished stamping out ICE vehicle production.
Also, governments just LOVE setting up bureaucracies for anything and everything. What better way to reward party donors than giving them a cushy job as a director of a new quango!
> I heard stories like this from Siberia, where they make a fire – a real fire – under a truck's engine to warm it up before starting in the morning.
Here's the Electric equivalent to that!
https://www.businessinsider.com/electric-vehicle-owner-denmark-toaster-warm-car-battery-starts-fire-2023-12
All automotive heat pumps are "inverter" heat pumps, because the EV battery is DC, and all motors (except brushed ones which nobody uses for more than a few watts) use AC.
The thing that makes a heat pump reversible or not, isn't the inverter though.
I'd be very surprised if any still-in-production EV used resistive heating for the cabin except when it is too cold to use a heat-pump.
But, no heat pump is going to be 400% efficient when it is sub-zero outside. The 400% figure is useless marketing blather, because it's the max efficiency and only applies when the temperatures inside and outside are the same.
More likely, the heat pump sucks heat out of the electronics & battery coolant, but when that coolant is too cold, they need to switch to resistive heating, because running the heat-pump and chilling the battery even further would be counter-productive.
> if you can run your heat pump using the output of a data centre then that's easily scavengable heat to convert into medium grade heat - i.e. a substantial cost saving operationally.
Err, no.
You can't even use a "geothermal source" with a heat-pump to produce economically-viable heat for a swimming pool, apparently!
https://www.energylivenews.com/2024/01/15/geothermal-lido-shuts-for-winter-due-to-high-energy-costs/
Electricity is fucking expensive in the UK, because there are so many snouts in the trough i.e. special markets for subsidising various well-connected people. So much so that a "geothermal pool" (glorified GSHP) isn't very cheap to run.
From another couple of articles on the subject: https://www.energylivenews.com/2024/01/15/octopus-dives-into-data-heat/ and https://www.businessgreen.com/news/4163570/octopus-energy-invest-gbp200m-waste-heat-innovator-deep-green (paywalled, but has picture)
It would appear that this is not really a datacentre, but a small cluster of GPUs submerged in mineral oil in a "fridge-sized" unit, presumably with swimming-pool water pipes running through the oil.
To heat the pool to 30C the oil would need to be quite warm.. Somewhere in the region of 60C, i.e. die temperatures of 80C or above. (just a rough guess on those figures)
So it could work, but it sounds quite expensive for a swimming pool heater. Hard to keep it secure, too. Wouldn't be surprised if someone cracked it open with a crowbar and ran off with a couple of (oil-soaked) £20k high-end GPUs
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
