Re: $14 Billion buys you and awful lot of oil
At the current unusually low prices, it buys you five hundred million barrels of oil, which is a little more than the amount China has so far burned in 2016.
86 posts • joined 7 Jun 2008
At the current unusually low prices, it buys you five hundred million barrels of oil, which is a little more than the amount China has so far burned in 2016.
Yes, really high-end tech skills have a substantial market value.
But this means that, if you train people up in-house, they will notice that they now have a substantial market value and they will leave. Until someone developing a billion-item million-transaction-per-second system for HMRC gets paid the same as someone developing such a thing for Tesco, there will be a flow of people from HMRC to Tesco.
Yes; you need to devote a modern computer to being a filter between the scanner and the world. Scanners are staggeringly expensive and modern computers very cheap, so this isn't a terrible problem, but IT organisations seem very resistant to models which treat the symptom when they think that merely by wasting vast amounts of development effort they can treat the cause.
The powers-that-be appear to worry much more about the confidentiality of the data than about its availability.
Frankly I don't care if the consequence of any ambulance-operative in the country being able to discover instantly that I have blood type B is that every hacker in the world also knows my blood type.
And that's why I'm proposing putting the USO on Openreach. They're big enough to absorb the cost of the USO and efficient enough that they can keep charging the same price as Virgin while doing so; and having satellite broadband available as an option to satisfy the USO means the last 0.1% cost Openreach no more than the £1800 per year each to settle their satcom bill (obviously in reality less, since Openreach will have a bulk arrangement with the satcom provider - sadly I suspect the economics doesn't work out for Openreach buying and operating its own satellite)
What is the point of a subsidy which covers the cheap part of the installation and not the expensive part? Give Openreach a full-service obligation without differential pricing, so they have to absorb the price of the ground station and the satellite bandwidth for any place that they're not willing to run wire to, and look how suddenly they will become more willing to run wire.
Planned maintenance would be the problem there; guaranteed breakage, whilst most kit will just keep on keeping on without worrying about maintenance. The servers at a previous place of work went down mostly because of UPS preventative-maintenance or bugs in the device that attempted to determine whether the chiller cabinet chiller had stopped working.
The mains is much more reliable than your average UPS.
Is there anywhere still on a three-year refresh for desktop computers in general? It's just about worth doing for software developers, but for general call-centre users five years ought to be fine and I imagine people are heading towards seven. Laptops wear out more quickly and the good-enough moment for laptops was more recent.
"Fuel poverty" means you have to spend more than 10% of your disposable income on energy to achieve an inhabitable house. It usually means 'insulation poverty'.
(IE it's plain old poverty, but specifically the kind caused by having to spend a lot on fuel just to keep warm; mostly affects poor pensioners who want to keep living in the same house they haven't been able to add insulation to since 1953, or poor single mothers of small children who have to keep the house warm for the baby)
A "deep-learning operation" is half a 16-bit-FP multiply-add operation.
The core of neural-network implementations is multiplication of short wide matrices by tall skinny matrices, in low precision; "machine learning extensions" is sexier marketing-speak for "supports half-precision FP"
Not especially tiny satellites (172kg, 100cm x 100cm x 50cm) - midway on an exponential scale between a Cubesat and something like Hubble.
Is it really wise to point out in public the faults in the encryption technology being used by your adversaries while the war is still on? OK, with luck if they move away they will move away onto even worse home-brewed encryption, but it seems somehow better to leave them in the swamp they're happy with.
(can we have a Snowden icon?)
" $2,495-$3,318.90 " seems renarkably high for a fitness band, and indeed remarkably high for an Apple Watch; it's listed at £199.99 from www.microsoftstore.com
Home of sock factories with sidelines in coal-mining and real estate trading; sticking to making electronic hardware is a reasonably tight focus by the standards of huge Chinese conglomerates.
There are two ways to convert an appreciating asset into money: you can sell it, or you can borrow against it. If you sell it the over-valuation problem passes on to the person who bought it; if you borrow against it the over-valuation problem passes on to the banking system. Since often the person who bought the over-valued asset did so with borrowed money, the problem ends up in the banking system.
So the end cost of feeling wealthy because you have over-valued assets is that you feel squeezed by the taxation process required to recapitalise the banking system - unless, of course, your Government decides that it should recapitalise the banks by cutting government services that people actually have difficulty living without, rather than by sticking up taxes.
It is remarkably politically unpopular to forbid buying over-valued assets with borrowed money, even by trivial changes like requiring the valuation of a property for a mortgage to be no more than the minimum amount that a comparable property has sold for over a period going back in time by the term of the mortgage.
Why do you think the marginal cost to provide a gigabyte of mobile data, in an environment with other customers, is as little as $1 ?
Charge one cent per megabyte, provide throttling at a customer-set level as a free customer service option, and they will line up and beg to be throttled the first time they get even a $250 monthly bandwidth bill.
'The nice old man’s family now want to sue you for compensation but you are uninsured. You had looked into the cost of insurance but it would have pushed up your costs, so didn’t bother. Instead, you simply declare personal bankruptcy and avoid paying the family a penny.'
Wouldn't declaring personal bankruptcy mean you paid the nice old man's family *all your money*, rather than 'not a penny'? You're declaring bankruptcy because you can't pay your creditor, namely the nice old man's family, and so the liquidator gets to give all your money to the creditor.
I'm not sure it's true that 2012 boxes will need replacing as soon as 2017; 2012 is late enough that Intel had decent power-saving-on-idle implemented, and is after the death-of-Moore's-Law point which meant new processors were not significantly better than old processors except on vectorised HPC apps.
If you can consolidate down to one Xeon-D in a breadbox, it's probably worth doing that; but replacing one old computer with one new computer costs at least £600 and saves at most £100 of electricity a year.
Which OS were you upgrading from? I have a two-out-of-two failure rate going from iOS 7 direct to 9.0
Diclofenac use is hardly unusual in the prone-to-die elderly - it's the normal prescription for osteoarthritis - and is likely to be no less persistent in human corpses than in cattle.
The launch cost was about $140 million (NASA procured the last three Delta II launches for $412 million in 2012) http://www.spaceflightnow.com/news/n1207/16delta2/), so there's still a very substantial saving if the backup craft costs half as much as the original to build.
"Lasso snaps" is at best a misleading heading - as far as anyone knows the antenna is still spinning happily, the radar's not working because the transmitter burned out.
They've had mainstream processors with eight cores and sixteen threads for three and a half years, and with DDR4 support for nearly a year now. They say 'Xeon' on the box and the motherboard's a bit more expensive.
"Please, Intel, sell me your fanciest chips for a price closer to that of your cheapest chips" is not a plea likely to achieve success.
Thanks for this interesting description of an interesting problem!
If you google-images 'Uranus Keck' there are some pictures of outbreaks of large white spots on the planet in August 2014; it was in a boring state when Voyager flew past it, but occasionally it wakes up.
If you've got thirty years to work with, gravity tractors (stick a heavy spacecraft with an ion drive hovering near the asteroid) will probably work, as would chroming one side of the asteroid; and you can measure if it's worked.
Five years is cutting it a bit close, but an object that would hit us in five years and that we haven't discovered already is probably small enough that it turns into an evacuation exercise. Which would be somewhat expensive, and would cause quite a lot of stress-related illness, but nothing an even fairly incompetent government couldn't handle.
Coping with a well-predicted medium-sized tsunami, which is the expected result from a hundred-metre impactor, is a harder but not an impossible coordination problem
The price is what it costs; if you want to collect it when people buy TVs, people buy a telly once a decade so it'd be £1500 on the price of every TV sold, which I don't think is practical.
The issue isn't "shipping Xeons to China", it's shipping anything to the four supercomputer centres (NUDT in Changsha, and NSC-Changsha, NSC-Guangzhou and NSC-Tianjin) that have been declared to be using their machines to work on atom bombs. It's not a full ban, you have to apply for an export licence and there's no guarantee that the licence will be refused, but the particular licence for building Tianhe-2a was refused. IBM are no more permitted to ship them Power chips than Intel is to ship them Xeons.
That's one of the main points of espionage.
If the Ruritanians have learned something that would be of great interest to the Potslavian opposition, and that would derail Potslavian plans in a way to the advantage of the Ruritanians were the opposition to learn it, of course they should inform the Potslavian opposition - you can't do this directly, but you might well be able toinform the Wallachian allies of the Potslavian opposition, who would cause the Potslavians to know the information through a reasonably deniable conduit.
That's exactly how diplomacy is supposed to work! Derailing the internal politics of other countries in ways that cause advantage to your own is what foreign offices are *for*.
Oxygen-free-copper cabling designed for the audiophool market turned out to be just what you needed to make practical proton precession magnetometers, and the process accidentally removed radioactive impurities so OFC shielding is used in a lot of sensitive particle physics work; there may still be some weird quantum-physics application in which extremely smooth silver cylinders are useful, though you'd feel a bit silly getting this cable and then dissolving the insulation to get to the silver.
This project cost £2.5 million.
Kepler answered 'how common are exoplanets'. This project will answer 'what are the easiest largish exoplanets to observe'. The TESS satellite, launching in summer 2017, which is much the same as this project but done from space at about a hundred times the cost, will produce a complete catalogue of exoplanets around bright stars, down to half the size of the ones found by this project.
The problem with Kepler is that it looked at such faint stars that it took inordinate effort to follow up the results; if you find planets around significantly brighter stars, you can do radial-velocity follow-up with moderate telescopes (which confirms that you've found a planet rather than a sunspot of unusual size, and tells you whether there are other planets in the system), there's the possibility of doing differential spectroscopy from space to find out what their atmospheres are made of, and if the stars are bright because they're close there's the possibility of directly imaging the planets with things like the Gemini Planet Imager.
A 20% price hike seems exactly right for a no-VAT price in the US and a with-VAT price in Europe
I think it's worth mentioning that this is a silicon-packaging facility (wafers go in, little square things with lots of pads on the bottom come out) rather than a chip fab.
The Apples were precisely designed to be incredibly cheap, in a world where the cheapest 'real computer' was a PDP-11/34 which cost $10,000 and was a metal box about 8U tall. The Sinclairs and Commodores and Acorn Electrons were cost-reduced versions of things that had already been significantly cost-reduced.
You wait. They'll come out in your country eventually.
Simultaneous with the release of iOS 8 were reviews demonstrating that it slowed down iPhone 4S or original iPad Mini hardware by about 30%; at which point my desire to upgrade rather went away.
Whilst Vadadora and Gandhinagar are quite close to Ahmedabad, Ghaziabad is a thousand kilometres away. Ghaziabad is, however, adjacent to New Delhi.
Thane is the one next to Mumbai; Nashik is a hundred miles north-east of Mumbai and Aurungabad a further hundred miles inland.
Nagpur is pretty much exactly in the middle of India, I would be interested to hear DEITY's advice about its infrastructure issues.
On the other hand, it's nice to have started hearing about India's tier-two cities.
The UK does not import gas from Russia.
We import gas from Norway, Qatar and the Netherlands.
We do import a fair amount of coal from Russia.
It's not 32 devices stacked vertically.
It's a single silicon chip with 32 layers; the flash memory cells are on the sides of U-shaped things pointing downwards into the silicon. It's a ridiculously impressive piece of engineering.
But each of the layers is made with a lower-precision (and therefore cheaper) process - something like 100nm rather than 20nm - so the 3D chip doesn't have the same capacity as 32 current-technology flash chips; it has about the capacity of one current-technology flash chip at a lower manufacturing cost, and a higher reliability because the memory cells are bigger.
I'm not sure you've made entirely clear the connection between Jodrell Bank, which is an astronomy centre in Manchester, and the ESS, which is a solid-state-physics facility to be built in Lund, Sweden.
Yes: the high speed of an ASIC hash engine comes because it's a pipelined implementation of SHA with the try-next-nonce code baked in, whilst a password-cracking engine needs a much more complicated next-trial-password unit
The Titan does not have crippled floating-point performance; the GTX 780 range of cards does
You don't need floating-point arithmetic to crack codes
The impression I have is that the motors for folding up the solar panels into the heated part of the vehicle are broken; so they will freeze up during the night and stop working.
It's only illegal for the signatories to the Antarctic Treaty; the nearest sufficiently-rich non-signatory appears to be Indonesia, who aren't that bad at mining. Other rich non-signatories with mining industries are Mexico, Kazakhstan and Saudi Arabia.
GPGPU and many-core are in very much the same regime; Intel gives you 456 1.1GHz double-precision FP units in 300W, and at a similar price nVidia gives you 832 706MHz double-precision FP units in 225 watts.
Are you sure about the '34 millimetres by 28 millimetres in size' part? That would make the chips five times the size of Haswell, and larger than your average DSLR sensor.
You mention '40ns, 9.6GB/s' for the core-to-memory-controller link, and then say that a system with eight controller chips can have 230GB/s CPU-to-memory bandwidth; which of those figures is right?
Juno has a remarkably pathetic camera (2-megapixel, 7.4-micron pixels, 11mm focal length, 58-degree FOV - better than most smartphone back-cameras but not by much); it's specced for 15-kilometre resolution at the 4300km closest approach of Juno to Jupiter, and has not the slightest chance of getting near enough to Europa to achieve six-metre resolution,
The Galileo probe is currently a thin spread of titanium vapour lightly sprinkled over the hundred-kilometre cloud deck of Jupiter; is the article very old, or should it be referring to ESA's JUICE mission?
At least burning PLA has a pleasant milky stench.
The problem with PLA for household goods is that it melts at about 50C; if you stick a 3D-printed thing in the dishwasher it comes out a bit Dali, if you try to 3D-print coasters you find you have made expensive and attractive stick-on bottoms for your coffee cups.
As far as I can see from the paper, it's offering a technique for getting 3N capacity out of five capacity-N discs, with protection against one disc failure in conjunction with two badly-located unreadable sectors.
(whereas RAID6 gives you protection against two whole-disc failures, but if you lose one disc and have unreadable sectors in the same place on two of the others then you've lost that sector)
It seems to involve fifteen reads and five writes per sector write, because it works by looking at groups of sectors on each disc, whilst RAID6 requires three reads (the sector you're overwriting and the two parity sectors) and three writes, so there's a lot more bandwidth used.
Basically this is a paper which has discovered a pretty mathematical pattern, with a dubious justification that it might be relevant for data recovery. It doesn't make sense in a world in which discs tend to fail mechanically rather than to develop individual bad sectors.