Posts by Francis Vaughan 397 publicly visible posts • joined 28 Apr 2007
See my post above. I though this too for a while, but a little digging and you find that "iPod Out" is not the analog audio output. It is the special iPod emulation mode iPhones have. I'm reasonably sure that iPod Out requires video out - this is how it displays a virtual iPod on a car's touch screen. (Which a car that supports iPod Out has. iPod Out having been developed in conjunction with BMW.) Apple really should be much more clear on their web page.
It looks a bit more complex that Apple have actually said, and Apple have not been smart in dispelling the confusion.
There is a lot of talk about the loss of audio line level out, and the Apple web page says that iPod out is not supported on the adaptors, leading many to assume they mean no line level audio out. Which it seems isn't the case. IPod out is a mode where an iPhone or iPod Touch will emulate an iPod in a manner that allows really nice integration into car audio systems, where it actually displays an iPod control screen, complete with album art on the in-car system. It is this that doesn't work. Assuming they are actually supporting line level audio out in the adaptor, the adaptors at least include a DAC, so it isn't just a connector. The adaptor probably contains more than this too.
The iDevices have never supported S/PDIF, and I very much doubt they will start now. Those docks that do support it have licensed a special USB chip from Apple that allows access to the internal digital audio stream. I doubt Apple will be giving up that control.
I suspect we are going to see some later technical descriptions abut the Lightning interface, but Apple have let slip a few things, and a look at some of the issues with USB make these make more sense. Apple say it is an 8 signal interface. Which is already interesting. USB 2 is two signal (+ and - signal) and USB 3 adds four more (superspeed TX +/- and RX +/-). The remaining two signals may be Apple simply keeping the old serial interface, or they may have done something much more interesting and the Lightning interface may not be USB at all, and the adaptor contains a USB interface chip as well as a DAC.
The plug is double sided, and I think everyone has assumed that because it can be inserted either way up it means that although it has16 physical pins, they are simply 8 electrical pins duplicated. This may not be true. If the socket has only 8 pins, sure, but if the socket has 16 pins we may see some slick use of differential signalling and symmetry allowing four pairs of differential signalling pins, plus the power, ground, and maybe power output for accessories. Apple have explicitly said 8 signal pins - so the question of where power and ground come from needs answering anyway.
Apple will want to future proof this for some time, so a range of things are possible. In a decade's time our expectations of what can be done on the connection interface, and indeed what we expect from our smart pocket device may be significantly more extended than we imagine now. Indeed, have a look at Thunderbolt. Cut out the two low speed signalling lines and a few redundant ground pins and it would fit. Who knows? The name is tantalising.
Yes, I expect that the vast majority of commentators on this will make the mistake of claiming that the uncertainly principle is in doubt, and totally miss what has actually been claimed.
From the first linked article:
"It is often assumed that Heisenberg's uncertainty principle applies to both the intrinsic uncertainty that a quantum system must possess, as well as to measurements. These results show that this is not the case and demonstrate the degree of precision that can be achieved with weak-measurement techniques."
The experiment addresses the phrase: " as well as to measurements." The intrinsic uncertainty remains.
Samsung are a seriously big company. The executives that sell fab services, those that sell memory, and those that run the phone division probably never see one another from one month to the next. Their individual jobs are to make money with their divisions. If one of the other divisions is lawyering up with one of their customers, all they will care is that that customer continues to buy their product or service. This is simply how large companies operate. Nothing is ever personal, it is always just business.
In the past iPods used really quite good (Wolfson) DACs, and the sound quality from the line level output was significantly better than that from the 3.5mm jack (which passed through a necessarily mediocre quality headphone driver amplifier). So loss of the line level outputs is potentially quite annoying.
On the other hand, Apple introduced access to the digital audio stream via USB - so long as you licensed it - and thus got access to the restricted interface chip needed. Which is how the high-end docks provide S/PDIF output. So Apple may have decided that mediocre audio via the 3.5mm headphone jack is OK for those that use cheap docks, and if you want any sort of quality, you have to buy a USB audio enabled dock. Which seem to start at about $150. They have not been using the same quality of internal DACs as before, so perhaps it doesn't make the difference it once did.
I will be interesting to see what they do use the available pins for. They may decide to output video, although analog video would make less sense now than ever. HDMI needs many more pins than they have, and by the time you get here Airplay in indeed the right answer. An Apple TV is cheaper than a USB enabled dock. It has S/PDIF output too.
Go have a look at the pinout for the 30 pin connector.
There are 8 pins devoted to the now deleted Firewire connection.
There are four more ground pins - so up to three are redundant.
There are three pins for video (composite and s-video) output
There are two pins of either reserved or unknown function
There are two pins for audio input
There are two pins for serial IO
There are two pins used for primitive control of iPod functions (one to click sound output, one to control charging)
Pretty much none of the above are needed. USB control subsumes the serial IO, Firewire is dead (sadly). Noone is recording with their iDevice. That is 23 pins that could be deleted without affecting the provision of audio output.
We only need to find 21.
Pins that will be needed are USB, signal and power (4 pins), 3.3 volt power output pin, to power things like the camera adaptor, maybe a separate ground from the USB one - which is six pins, so add two for audio out, and you still have a pin left. Whihc I suspect will be a new in-dock mode control pin that signals the iDevice what sort of dock it is connect to. It may be actually be a upgrade of the accessory indicator (pin 21) function.
Probably a number of reasons.
The Swedes cannot provide a blanket exemption against extradition. It is illegal for them to do so. A lot of people forget that in most countries there is an explicit divide between the government and law enforcement. It is not possible for the government of either Sweden or the UK to step into the legal process and pervert it. This is for very good reason. Neither the Swedish or UK governments (as in the elected governments) are involved in the Assange legal process. The law courts are, but the whole point of the law courts standing is that they are not under the influence of the elected government. They enforce the law.
The Swedes could make a statement that there is currently no warrant or request for extradition from the US for Assange. Doing so of course would be worthless. No-one who believes the conspiracy theory would believe them. To make a statement that the Swedish government would ignore any such extradition request from the US - before it had been made - would probably put them in breach of a number of treaties.
If any such extradition was requested by the US from Sweden, the UK must also give permission. So any idea that this is some weird conspiracy to allow the UK government to claim clean hands on the deal doesn't wash. Since the UK must give permission anyway, the obvious point is simply this: why didn't the US simply ask the UK to hand him over? If you wanted pick the country in Europe most likely to kow tow to some notional US imperialist line, and ship Assange over, you would put the UK at the head of the list, and Sweden pretty close to the back.
The entire furore is over a total minnow in the grand scheme of things. It certainly feeds Assange's ego. He clearly is someone of international importance, a man who can bring entire superpowers to their knees. He isn't. He is a footnote. People forget. It wasn't him who leaked anything. He simply provided a forum for leaks. And he didn't do it alone, without him Wikileaks continues anyway. Assange has a very heightened idea of his own importance, and seems to act on this ego.
The most likely answer is that he will eventually get shipped back to Sweden, where the charges will either be dropped, or fail in court, probably because the women that complained get cold feet due to the publicity. He will then go free. End of story. Now that probably strikes more fear into Assange's heart than anything else. For it will prove that he actually isn't important enough.
Diplomatic bags are exempt from any search. X-Raying them probably comes under this. What is a bit more interesting is that the treaty is explicit in that the bags contain documents. A literal reading would suggest that if, for some reason it was obvious, without search, that the container contained a person, that that container was, by definition, not a diplomatic bag.
Personally I would simply load the "documents only" bag into the unpressurised hold of the next plane to Ecuador.
"What charges?"
You can't be charged unless it is done in person. The entire point of extraditing him to Sweden is to allow charges to be laid. So long as he stays out he cannot be legally charged. He knows this. The European Arrest Warrant was issued by Sweden in order to get him back there to allow him to be charged. Arguing that because he hasn't been charged he must be innocent is simple ignorance about the manner in which the process happens.
The way Swedish law works is different to the UK. Once he is charged he is required to face trial within two weeks. This is one reason why the charging process happens later than you might be used to. In Sweden, the process requires a "second interview" during which charges are laid. It is for this interview that Assange's arrest warrant was issued. There is a lot of misinformation about the process, which seems to be wilful ignoring the nature of the legal process in Sweden and trying to re-interpret the names used for the stages (which will be in Swedish) in a manner that suggests a far less serious level of intent.
Everything that all the Assange supporters complain about hinges on one wild assertion. That that bastion of conservative politics, the well know lapdog of US imperialism, Sweden, has already agreed with the US to ship him over to the US once he lands in Sweden. This isn't credible. If you drop that one assertion, the rest falls apart.
The popularity of BBM with the UK kids seems assured. But what to the kids in the rest of the world like to use? Being one of those that has no idea what the yoof of today are up to, I have no idea about BBMs ubiquity, but do feel a little sceptical that RIM have the market sown up in every country that their rivals operate.
Like all comedy, the answer is in the timing. Much better to use one that is three times more powerful than mess about with worrying that the three don't ignite exactly at the same time (or worse, one fails to ignite) and then have to cope with asymmetric thrust just as it tries to launch. Failure of one SRB to ignite on the Shuttle was one of the unsurvivable accidents (and possibly most spectacular) possible.
Following up from above, the problem with the test chamber pressure probably remains. Even if the igniter doesn't create much gas, the motor grain is designed to create much gas, so a sputtering grain that is doomed to fail in a real launch might manage to create enough pressure in the REHAB chamber to cause itself to light up. This presents a difficult problem. It suggests that a completely valid test does need a large enough vacuum chamber to cope with a significant amount of gas production.
One is reminded that some tests of real rocket motors use basically a very large water jet ejector that can sustain a vacuum even after the motor starts running. I wonder if something designed to fit on a fire hose would work? (Only partly in jest here.) An industrial size water jet ejector would probably work if you could find one.
A bit of searching around brings a few facts up.
PIC uses Lead Dioxide and Silicon, and yields Lead and Silicon Dioxide as reaction products, neither of which are gasses at low temperatures. That and any vaporised plastic. The electric match uses Antimony TriSulphide and Potasium Chlorate, which yields only some Sulphur Dioxide as a gaseous product, the Potasium Chloride and Antimony Oxide not mattering. Thus the igniter may well not be reducing the pressure in REHAB all that much. Clearly the easy test is as suggested above - just ignite one inside the test chamber and see what happens to the pressure. That alone should set to rest any issue of whether any additional hardware is needed. There are two places where the pressure matters. In the test vessel (REHAB) where you want it to remain low, and in the motor chamber, where you simply need it to be accurate. In the motor chamber the temperatures may be much higher, and the reaction products may stay gaseous for a tny bit longer, but this is what they will do in the actual launch, so this remains accurate. Outside the motor chamber everything will be cold, and I suspect that little of the reaction products will remain gaseous for any meaningful time except the the Sulphur Dioxide.
Next, the equation of burn rate of the motor is Rate = constant x pressure^n. Where n depends upon the propellent composition, and seems to vary between 0.2 and 0.5 Basically the burn rate only depends upon the chamber pressure. Whilst this was known in the abstract, just how critical it is is perhaps a surprise.
The point of the PIC is to slam heavy hot particiels into the rocket grain, obviating the need for heat transfer by conduction though hot gas.. But the grain won't stay burning unless it is subject to sufficient pressure. If it doesn't start burning fast enough to build chamber pressure it seems it fissles out. Probably as soon as the PIC is depleted. So it may be argued that a problem with the PIC is that it may not be producing enough hot gas. Given the above this is perhaps not a surprise. A reliable ignition might be achieved by a modification that simply adds something that produces hot gas as well as the hot sparks in the motor chamber. This could be much more reliable and less subject to catastrophe than a burst plug. However it may simply be that the difference in composition between the different manufacturer's motor grains may be enough to bridge the gap. The burn time of the motors might provide some clue as to this since the motors are much the same weight.
Exactly. It is difficult to imagine how dedupe would make anything more than a trivial difference to a personal computer's persistent storage use. Personal computer file system use is dominated by pictures, audio, movies. All three of these are already compressed. Dedupe makes little to no sense. Indeed it would probably just slow everything down and wear out the flash faster.
This really is a big thing. I remember arguing for this access 15 years ago, but at the time the government agencies were required to make a token profit, and charged for anything they did that was actually useful. This initiative will make a huge difference as it becomes possible to perform useful analysis on data, and build a business on doing this analysis without the impost of what were what significant charges for access to the data. Access for fundamental research will be similarly enhanced.
I'm going to bet that the owner and staff had concluded that he was engaged in creating a clandestine video of goings on at their restaurant. This suggests that that they actually had something to hide, and that they were worried. So rather than just address the attack, either McDonalds, or as likely, the French tax office, might like to give the place's operations a thorough audit.
Sony seem to have simply got the thing wrong. It is cheap, looks naf, and isn't waterproof. It clearly has never had anyone who has designed a watch before near it. The software will probably be rubbish too.
What it should have been is more expensive. And better.
Sony should have swallowed some of their usual not-invented-here pride and gone to talk to Seiko about the entire project. Sony can supply the internals, and a proper watch maker that understands how to make things that actually work, look good, and stay working on ones wrist should make the case. And make more than one case.
As above, when talking about nice watches - they are jewellery. People like to have nice things, and depending upon taste, clearly very nice, and very expensive things, on their wrists. Not something that looks like it was designed as a high school project.
It needs to be something that people will desire. Heck, make it £500 and as lovely as anything from the Swiss. Make it beautiful, and make it work really well. Sell it with a contactless charging pad (just leave it on the pad overnight and it charges). Make it in a range of styles, finishes, bands. Ensure that the buttons are the sort of thing one expects on an expensive watch, not a bit of cheap consumer electronics tat. And so on. Do that and they would sell more, even at £500, than they will sell of this bit of tat at £100 odd.
This will sell to the odd geek, who will later discard it. Sony need to get back a bit of the mojo of old, and make something that people actually desire to own. Once they sell the £500 one for awhile they can introduce the diffusion brand version for the ordinary folk. Oh, and give it a recognisable name, so that people can ask "oooh is that a Sony XXX?" (Where XXX does not equal "SmartWatch Andoid". XXX could have equalled iWatch - but that one is taken.)
Still proves the point, no matter how duplicitously obtained. If a pair of convicted drug dealers can get the "exercise the Royal Prerogative of mercy" call, and be pardoned of a crime, there is no logical argument the British government can reasonably stand on to deny Turing. The process clearly exists, and has been used recently.
There is nothing stopping you writing a state control table that implements a program that can use the tape to hold both data and program code for a new automata implemented by the Turing machine. That is after all just unified code and data. And a nice exercise in showing how a Turning machine encompasses all other automata. In principle you could code an x86 interpreter on a Turning machine. All the CPU state still lives on the tape, an you could put x86 code and ordinary data elsewhere on the tape. What matters is that the ONLY mutable data lives on the tape. The table is fixed.
Yup. It isn't a Turning machine at all. Sorry, but it isn't.
A Turing machine only holds machine state on the tape. The programme, that is the state transition table is fixed and cannot hold mutable state. That is the definition. This model device only used the tape to provide I/O. There was mutable state inside the simulating engine - the engine internally did the calculation 2+2=4. It was not calculated on the tape, which is what a real Turning machine must do.
So, sorry, nice bit of Lego, but it is NOT a Turing Machine.
Metadata. Ah yes. Good point about the metadata. That is going to be one place where flash might help. However it may start to be better to just cache it directly in system memory. There might simply not be enough metadata to warrant any optimised secondary storage for it at all. The breakpoint is likely a moving target. Reliability questions making things a little less clear cut.
HPC has always been thus. I/O isn't random access to lots of little bits of data, it is massive broadside access to very very large lumps of ordered data. Latency of access is swamped by the transmission time. Optimising for access is simply missing Amdhal's Law. Most caching strategies don't apply. Data is very often only read once. Optimising data layout, order, prefetch, matching bandwidths, this is where you win.
Same for deduplication. Science data is inherently not internally correlated. Except for the case where finding hidden correlations is the entire point of the computation in the first place. Enterprise level dedupe doesn't get any traction at all, just slows things down and costs money.
Just the nature of the game.
A couple of hundred million isn't all that much in the grand scheme of things. Consider that in principle LiquidMetal can be formed more like plastic or sheet metal, yet produces a result probably superior to the CNC machined unibody cases, and $200 million isn't a huge investment. The reduction in unit cost of the product, and yet a serious improvement in that product would, with Apple's volumes, mean the $200M could be amortised almost instantly. From the linked to article:
"Therefore, I expect Apple to use this technology in a breakthrough product. Such product will likely bring an innovative user interface and industrial design together, and will also be very difficult to copy or duplicate with other material technologies."
Think unibody design, but on steroids, and an exclusive right to produce them. Very Apple. Also very forward thinking, with a long game plan. Something else very Apple.
Note that Apple is a licensee, they don't actually sue other users of LiquidMetal, the LiquidMetal company does. If LiquidMetal decide to breach the contract with Apple, and give the tech to someone else, then LiquidMetal get sued by Apple.
Late reply.
"No version of Mach is "based on a Unix code base"." This is probably simply not knowing the history. Prior to Mach 3 the message passing system API was grafted into what was a Unix kernel. This included the first external pager, and other services. I know, I have the source code. We are talking a long time ago. I visited CMU and the Mach team in 1989. Mach 3 was under development at this time, but 2.0 and 2.5 had been deployed for a while. Mach 2.6 was shot though with Sun code, and core kernel components (such as the process switch code) were lifted verbatim. Mach 3 changed and extended the APIs quite a bit.
The term Userland is much more recent than much of this technology. Sure the Unix emulation ran in a user mode process, so in more modern terminology is a Userland. The microkernel guys were doing this all over the place. Back in the late 80's it was all the rage. Chorus was doing very similar things. The terminology doesn't make the idea different.
Care to explain what part of Mach 3 is shared with BSD? The answer is nothing. The BSD emulation layer is simply grafted onto the side of the operating system. Mach was written at Carnegie Mellon University, and the leading light of the team was Avie Trevanian, who became NeXT's and then Apple's main technology guy. Mach 2.5 was heavily based upon a Unix code base, built to show the value operating system API, whilst keeping the useful bits of an existing operating system. Mach 3 was a scratch rewrite. In Darwin some of the Unix emulation (which was mostly in user mode server processes) was migrated back into the kernel for speed.
This continual background buzz that OSX is just BSD is simply an annoying lack of knowledge of the technical history, and the current technical structure of OSX. The entire kernel is new code, the process model is different, the device driver model is different, the system APIs are different. OSX includes the Quartz graphics layer, and the list goes on. There is however, very usefully, a BSD compatible emulation layer. Apple leveraged this well. But it doesn't make OSX a tweaked BSD. What OSX is, is NeXT. If you want to see where the OS really came from look there, and the linkage back to Mach.
* A purely mechanical system, with no electronics
Given the current state of electronics I'm not convinced this is a good requirement. It presupposes that electronics are in some way inherently less reliable than mechanical ones. I am far form convinced this is a reasonable assumption, and it may be unduly swaying the rest of the design choices. That the backup system be independent of the main system is good, but there is no reason it should be purely mechanical.
* Triggered at, or after, balloon burst
This seems to be a revision of the requirements, curiously you are on the way to making balloon burst the preferred primary launch time.
* Guaranteed to work, no matter what
Good requirement :-)
* Easy to test
The proposed design fails totally here. Unless you can test parachute deployment at the equivalent altitude you are not going to be able get a useful test.
My suggestion:
Get a three axis accelerometer, and a micro-controller. All you have to do is run a suitably damped sum of the accelerations. If they all sum to zero for more than a few seconds the thing is in freefall. You can test this easily. You can wave it about, drive around in a car, hang it from a flagpole in the wind, simulate just about every sort of buffeting and pathological scenario you can, and then chuck it off a roof to see if detects that it is falling. Power the device with its own battery. It will be vastly more reliable than any Heath Robinson contraption of springs and such that will most likely freeze solid or simply plummet.
I used to joke about a Facebook phone. Trouble is, we all knew it wasn't entirely a joke. The business model was simple. It only runs Facebook. And they give it away for free. Other than that it is business as usual. They sell all your details, habits, interactions, and interesting liaisons, to whoever will pay, mine everything they can out of your life, and push targeted adverts at you.
As a condition of use you agree that the geotracking stays on. They add a payment portal, from which they mine all the transactions details.
Oh, and as a condition of ownership, it never gets turned off, and you keep in on you all the time.
Simple really.
Curiously the Helium we get from the wells is has not been produced by stellar processes. It is the result of radioactive decay in the rocks. In most places this will slowly make it way to the surface and eventually vanish into space. But where you have a natural impervious barrier that traps any gas, you the Helium is retained. These natural barriers are the same barriers that trap natural gas (aka Methane) and is why all natural gas wells produce Helium along with the methane. Some wells have a higher proportion of Helium than others, and those wells near the US Helium store were quite high. Any natural gas facility that creates liquefied natural gas can also produce Helium with only small additional effort. Qatar and Australian LNG plants will. But if you don't extract the Helium, it simply goes up the flue when the natural gas is burnt. Whenever you put the kettle on you probably waste as much if not more Helium as a child's balloon holds. The vast majority of Helium is lost this way. Simply because it was never recovered in the first place. The limits to the accessible Helium are simple. When we run out of natural gas, we run out of cheap Helium. Forever.
Lewis' analysis is spot on. Space ports are just plain silly, and of themselves provide little to no value. Sealaunch simply use a converted oil platform. Building a sodding huge runway with no technology on the horizon to use it is "rain following the plough" in the extreme.
A nick pick however. Not all launches are to geosynchronous orbit. A great many are to polar orbits (the majority of Earth observation craft) and these require the exact opposite of an equatorial launch location. They benefit from a launch from as high a latitude as possible.
The temperature varies with altitude quite nicely, and it actually starts to increase again above about 80,000 feet. Around here you get about 1 C every thousand feet, which isn't difficult to track with a simple semiconductor thermometer.
Clearly there are variations, but this altitude is above pretty much all of the weather, and you should see a stable relationship with temperature to altitude as good as you see with pressure.
What you might want to look at is the NRL empirical model that can give you a prediction of the temperature versus altitude taking into account current conditions and location. Wikepedia here: http://en.wikipedia.org/wiki/NRLMSISE-00 and in particular check out the links at the bottom of the page.
If you thought that marketing triumphing over the engineers was bad enough, this exercise would appear to be gearing up to be a triumph of lawyers over marketing. The amount of publicity this has garnered in Oz (all the major news conduits covered it prominently) is enough that no-one is going to be fooled by a lawyer driven victory over a technical loophole. The damage is done, and anyone in marketing with any sense understands the first rule of holes. Stop digging, and work out a way to spin a backdown into some form of positive light. But if you continue digging, it never gets better. A victory will not be perceived as Apple acting in a manner they imagine they are viewed in the marketplace, but rather something much more squalid. It takes a long time to build a reputation, but a very short time to damage it.
With a pedigree like that, I'm sure William Blake was probably wondering just how he ended up at MS, and has been looking for a suitable time to jump ship. Given he never actually asked to be at MS I don't think he can be tarred with the same brush as some other exports from Redmond. There is a certain irony in getting someone from MS, given that they stole Burton Smith from Cray. Burton was arguably one of the last real technical visionaries of supercomputing. Sadly he was probably just looking for a sinecure and has been happy to quietly ossify.
Hmm, trying to guess there. Meter is calibrated in % and db. % is almost certainly modulation depth. There is a switch for audio, tv & audio, tv. There are a heap of devices that look like parametric equalisers. Given the clues this looks like part of a very early satellite ground station. Given it is in Germany it is possible it was even used for the first Telstar, which would make it a very significant bit of history. Then again the above comment about the Olympic games might mean it was used for Syncom 3, and receiving the 1964 summer Olympics. It could of course have been used for both.
It was touched on above, but in real HPC systems, the Linpack performance is mostly ignored. (Unless your only workload is linear algebra.) The key concepts are how well the architecture scales, and what the sustained performance is. The cheap laptop gets the best bang for buck simply because it is incapable of scaling. You cannot join a boxload of them together and get much useful sort of speedup. Not in a way which can produce good sustained numbers. Supercomputers have never been about raw flops. They have always been about a critically balanced design, with equal attention paid to cpu speed, memory performance, and I/O performance. There is no value in a high speed CPU if you can't feed it with data fast enough. Caches are often not as useful as you might hope.
The cost of interconnect interfaces and fabrics are always a significant fraction of the cost of the machine. From simple Infiniband, through SGI's NumaLink, IBM's Blue Gene and beyond, you get what you pay for, and depending upon the workload, you have no choice about which fabric to use. Tuning the interconnect topology to the problem is useful too.
Linpack has the disadvantage that it can be tuned so that it is insensitive to the interconnect. Because it does little more than time solving large matrices, the amount of time spent communicating data is related to the length of the edges that divide the matrix into subdivision distributed across the nodes. But the amount of data is related to the area of the subdivisions. The bigger the data size the less communication is needed relative to the work done. The more memory you add to a machine, the larger the test dataset you can configure, and the faster the Linpack result. Simply because the less important the interconnect becomes for the contrived test. This usually does not relate to the machine's actual performance on real world workloads. Hence the need to spend real money on interconnects, even if it doesn't actually appear to improve the simple performance. An old college coined the phrase "Gigaflop harlotry" to describe the focus on simple Linpack number and rank in the Top 500.
It really is nice to see the MTA going somewhere. I saw Burton Smith talking up MTA nearly 20 years ago, and even then everyone considered it to be about the only new idea in computer architecture. In that time not much new has arisen, and for a long time MTA looked pretty moribund with little more than some neat parallel sort results. So this really is rather heartening.
"Doctor Who is a vitally important BBC brand"
There is a deep truth here that will talk louder than any other issue. The BBC makes more money off the Dr Who brand than any other income stream. This a big difference from a decade odd ago, when it was unloved by management, and prone to the idiotic whim of whoever was in charge. Damaging the brand would be what is known as a "career limiting move". The BBC already sells millions of DVDs, toys, and books, has had a number of spinoff shows, and has a continuing core TV series. One that sells for serious money all around the planet. This is the sort of success that makes a Hollywood exec cream themselves just imagining the prospect. Hollywood movies don't make the real money at the box office. It is this second stream where the gravy is. The BBC does not need Hollywood. There is nothing that Hollywood could teach it or provide to it that it does not already have. Not that that would stop various Hollywood executives pitching to the BBC to try to get the rights to do a film. But this should be recognised as the weaker party (Hollywood), bereft of any originality or actual creative talent, attempting to make money off the back of the BBC's brand.
There is no upside for the BBC is getting into bed with Hollywood. The best that can happen is a continuation of the current astounding strong brand, and the money that already rains from the sky from it. Everything else is downhill, with a damaged brand, reduced income, and no doubt any money that is made from the film being syphoned by the Hollywood moguls before the BBC sees a share.
The difference being that the fuel pump on even a 50 year old RL-10 is driven by a turbine that generates 500kw. So just the fuel system has more power than any car you can ever dream of owning. And the entire rocket motor weights 140kg, so less than your car's engine. When things are are crazy as this you don't want complexity. You want every last possible thing that might go wrong removed. The sophistication is in the simplicity.
Whilst 37 x 49mm isn't full frame 645, it isn't as far short as this. 6cm is the full width of the film not the frame width, the frame size of 645 is 41.5 x 56mm. Given the chip yield goes down with about the cube of the size, and given these sensors are already about $10,000 for just the chip, I think they can be forgiven for for the size.
I still shoot 645 film. £35,000 buys one hell of a lot of film, even at the ruinous prices charged now. On the other hand, there is no film around that can achieve the resolution described here. When I win the lottery I will be buying one.
A nitpick, but of a very common misconception.
Dithering does not mask the quantisation artefacts with random noise. That would imply that the amplitude of the noise added needs to be of an amplitude high enough to drown out the quantisation. In fact the amplitude of the dither is one half of the least significant bit. It does not mask the quantisation artefacts, it actually removes them totally. In its simplest form the use of AIWN (additive independent white noise) will totally decorrelate quantisation artefacts. (It is the fact that the errors in quantisation are correlated with one another that leads them to contain objectionable audible sound.) However my noting that the ear has vastly greater sensitivity in some frequencies, and much less in others - especially in the top octave - it is possible to craft the dither in such a way that increased noise in those parts of the audible spectrum we are less sensitive to can be traded for higher resolution in those areas we have greater sensitivity to. Thus it is possible to get a genuine 18 bits of resolution in the mid bands of a 16 bit digital recording. Most recording systems provide the engineer with a range of dither options, with different profiles being suited to different recordings.
It is rather hard to see how this prize is really going to work. It is clearly modelled directly on the Nobel. The rules allow for one to three recipients, and the criteria include: "a groundbreaking advance in engineering which has created significant benefit to humanity".
But the nature of engineering isn't like science. Whittling down an engineering contribution to three prime people is going to be a huge problem. The nature of engineering is building on what has gone before, and delivering results in a timely, safe, and risk free manner. And it involves teams.
Take an example from computing. How about we nominate Ken Thomson, Denis Ritchie, and Brian Kernighan? Whilst the creation of Unix was important, the face of computing as we know it would probably be almost the same if it had not been written. The most common desktop OS isn't a Unix variant. Perhaps any prize should include Dave Cutler. But then, the true operating system innovations were long before any of these guys. The PDP-7 Unix was written on already had an operating system. There are so so so many people that contributed that makes no sense at all to go looking for "the three". Try to nominate the three giants of engineering the Apollo programme.
The danger is that it might simply become a surrogate science prize for areas of science that don't fit into the Nobel categories.
(On the other hand we all probably have favourites. Personally, if he were still alive, I would be nominating Tommy Flowers for the prize.)
It would be interesting to have some example nominations.
Sort of an interesting idea. However what could be a really good idea is to talk to a real actuary, and see what they think. Actuaries certainly work in risk, but they are to a large extent statisticians. Without a statistical basis for risk assessment it becomes a different problem.
This is the issue with quantifying internet security. There isn't the equivalent of the underwriters laboratory that certifies materials and components, and there aren't standards bodies that build standards built upon centuries of experience. Worse, there is no way of quantifying the effects of a security breach ahead of time. There are no easy risk/cost curves. There are huge discontinuities in the problem. This isn't likely to be a place where actuaries play. But a professional actuary might well differ with me. Hearing from one would be interesting.
I really don't think the specifics of this case are the real game.
Pundits arguing the minor details of the individual patents and the usual fanboi versus ant-fanboi are just not what the game is. We have a couple of the largest companies on the planet facing off, with the lingering threat of unleashing the oft mentioned global patent wars. In the end I imagine the outcome won't be something that is all that satisfactory. I will assume that eventually we will see an even more messy and complex global cross licensing system that is owned by the very very big players, to the even greater exclusion of the smaller players. Te only alternative we can hope for is that global patent Armageddon really is unleashed. Out of cahos comes opportunity. Maybe, we an hope, will come some sanity. Maybe.
Given the much wider and serious threat of a global patent Armageddon, these spats seems puzzling. Neither Samsung or Apple are stupid. They both know that allowing patent litigation to escalate will hurt both of them, as it has the potential to draw many other players in, in an uncontrolled and ultimately destructive manner. Success in any patent litigation also encourages the patent trolls, something no major player wants to see.
The seems to be something about this particular round as being born of spite, after a breakdown in the previously cordial relationship between Apple and Samsung, but no large player can allow such things to direct their operations. Or certainly should never do so.
However, Apple have been on the outside of the global patent cross licensing club for phones, and these rounds of litigation may turn out to be part of a longer game, aimed at securing a permanent place at the table. What is reasonably certain, the public face of these little arguments are little to do with the real game being played.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
