90 posts • joined 5 Mar 2008
We owe a lot to AMD .. but there's a limit
AMD had a great run. The K6 and K6-2 were great (though I never owned one .. I kept a Pentium Pro 200 Linux machine for quite a while).
I built an Athlon 700 in early 2000. And then a few years later in early 2004 an Athlon 3200+. Both matched or beat Intel's chips at the time, and for less money.
And then AMD saved us all from Intel's vision of 32 bit x86 forever (and the awful Pentium4 at that) and Itanic if you wanted 64 bits.
I never did own a true AMD64. The Core 2 Duo was just as fast in a laptop and on the desktop, and by the time that felt slow my best option seemed to be (in Jan 2010) building a quad core i7 and overclocking it.
And I've just this week built an i7-4790K box (reliably doing 4.6 GHz on 4 cores, 4.9 GHz single core) that should see me for the next few years.
AMD doesn't have anything to touch the i7, and certainly not for $400. (they're ok if you want a low end box, but I don't)
It's very very sad. I'd buy AMD by preference if they were at all close.
dead simple programs are equally fast to write in any language
If you look at their code, it's just some statically allocated multidimensional arrays, some simple nested loops (corresponding to array dimensions), and a small amount of arithmetic and occasional if statement.
The C code doesn't give you the safety net of array bound checking, and the for loop syntax is annoying to the uninitiated, but there's really not a lot of difficulty in using C or C++ in this way. Certainly the development time difference between it and Python would be absolutely minimal.
Interesting that their test Mac is a 2.3 GHz quad core i7 (model unspecified), while the Windows machine is a quad core i7-3770 at 3.4 GHz, and yet the execution times are pretty much identical or even faster.
The conclusion from that is the Mac is a laptop with massive SpeedStep taking it up to to around 3.3 or even 3.5 GHz (current 15" Retina MBP).
Python sounds reasonable. But I'd be likely to do it as a cron job, written largely in bash. It might well call a bit of python for some reason or other.
p.s. I am well acquainted with optimising the sh*t out of things in C (or asm if necessary) having, for example, been involved in writing a commercially successful (right up until iPhone and Android came out) compiler to turn Java games into native code on cruddly little feature phones with 1 MHz CPUs and as little as 400 KB of RAM.
I've got an iPad with an A7 in. It benchmarks and feels nearly identical to my 2.5 GHz core2duo laptop. That's plenty for a MacBook Air style laptop even today.
Where, by "over a year", you mean "the last TWO generations of iPhone and iPad"?
Mr Chirgwin misses the point
It's not about whether this particular model is or is not a good or mature one. It's about long running numerical calculations being pretty much inherently unstable unless heroic measures are taken to organize every step in them for minimum error. Just getting a simple gaussian elimination right is a complex and specialized task.
We KNOW that typical climate scientists are not numerical analysis whizzes. Recall that in FOIA2011 email 1885.txt, Phil Jones admits to not knowing how to calculate a linear trend in Excel (or anything else):
It's very likely that other, more mature, climate models used in publications:
1) have serious numerical analysis instabilities, to the point of being GIGO
2) have never been run on more than one kind of computer
3) would show exactly the same kinds of problems if they were
trivially easy to do as a 1 off
And I have a few times.
A bit tough if you're only going to have one quid on the first day. Easy if you get five pounds at the start. If you've only got to achieve a £1/day overall spend rate then you can have your choice of rice, pasta, potato, flour each day. Ingredients for a home made loaf of bread are about 30p. Spices to make it interesting are dirt cheap once you acquire them as you use very little.
Avoiding feelings of hunger will be easy.
Getting enough energy intake will be easy.
Even getting enough protein will be easy with eggs and cheese.
Getting a long term healthy balanced diet won't be at all easy, but doesn't matter if it's only five days.
seems like the wrong niche
Flywheels are really great at absorbing or providing huge amounts of power over short timescales: a few seconds to maybe a few minutes. They're also great for a very large number of charge/discharge cycles. I don't think there's any successful example of using them to store energy for days.
A Trojan IND17-6V battery weights 188 kg, stores 7.21 kWh of energy (if you take 4 days to extract it, about 5.5 kWh if you extract it in one day). It is designed for 1500 cycles at 80% discharge depth or 5000 cycles at 20% discharge depth. It costs about $1250.
So two of these batteries will hold as much energy, weigh 380 kg (840 lb), cost $2500, and last 4 - 12 years depending on how far you discharge them each time on average.
NYT and WSJ had simply been reading blogs...
Clearly, both newspapers had simply read early Apple UI guru Bruce Tognazzini's purely "what if?" blog post from a week ago:
It doesn't take any more than that to explain the close timing in the media, and *certainly* it doesn't mean Apple has a product on the verge of release.
It would be out of character for them not to have been playing with the idea in the labs for a few years already though, waiting for the gating technology to hit.
It's entirely possible to make a video unavailable in only one country while others elsewhere continue to watch and comment on it.
the underlying trend
"Remove the cycles and look at the underlying trend."
The problem is we don't have enough hard data to know whether the underlying trend is accelerating, or just part of a longer cycle.
We do know that there have been some pretty amazingly big temperature extremes in the last 20 or 30 million years. And that the Earth has not raced off to become another Venus or another Mars in the process … it's corrected itself somehow and come back to what we have now.
20 or 30 million years seems like a very long time compared the couple of thousand years of recorded history, until you realize that mammals have been around for 200 million years and great apes (i.e. very like us) for 40 million years. They didn't have our technology to help them adapt to extremes, but they survived them.
The point that natural cycles, if they exist (as the Met Office is now agreeing with skeptics on), can both accelerate and stall temperature rises is an excellent one.
The alarmists are now saying "sure temperatures are stable now, but that just means they'll go up in a hurry later on".
That's likely to be correct, but misses the point.
The point is that what we care about is the possibility that warming is being caused by humans AND will be harmful AND that we could do something about it by decreasing our activities. If it turns out that the rapid warming in the 80s and 90s wasn't caused by us then there will be very little that we can do about any future rapid warming anyway, and we'll just have to learn to live with it.
Climate scientists work out all the natural things that they can think of that affect the climate, put them into mathematical models, and then just ASSUME that any temperature rise in excess of that was caused by us, and specifically by our CO2 emissions. This has in the past caused them to think that doubling CO2 from pre-industrial times could cause a temperature increased of 3 degrees or 4 degrees or even in some scary reports 6 degrees.
If you instead assume that there is an approximately 60 year natural cycle that is flattening temperatures now and caused them to rise faster in the 80s and 90s (and decrease in the 50s and 60s), and continue to assume that all the rest of the unaccounted-for change was caused by CO2, what you come up with is that a doubling of CO2 levels from pre-industrial times might cause a 1.5 degree temperature increase.
That's a lot less scary than 4 or 6 degrees. And we've already seen 0.8 C of that.
There is also the possibility that there are more yet longer cycles not yet accounted for in the models. Good temperature records only go back 150 years, but there is considerable anecdotal evidence of an approximately 1000 year cycle that we're in the rising part of.
There is no such single thing as "considered statistically significant". It depends what your purpose is.
Among real scientists 2 sigma or 95% is considered to be the point at which you say "oh, that looks interesting, lets research it more to see if it's real" (i.e. try to add a few more sigma on to the probability.
The following article shows how physicists deal with such statistics, and don't consider something to be proven unless it reaches 5 sigma or 99.99994% probability that the observations didn't happen by chance.
That discussion is largely in the context of deciding whether or not Higgs' boson is real, a subject of absolutely no practical importance to anyone not in that academic field.
That something as serious as hurting the world economy to the tune of trillions of dollars and condemning millions of the most vulnerable people to starvation and death via raising the prices of food and energy should be decided by mere 2 sigma evidence is utterly ludicrous.
Re: No 'coffee' shortage as long as we have Soy Beans
I roast my own in a popcorn machine too. Green beans cost about 40% of what roasted beans cost in 200g quantity (I can't buy more otherwise it's stale before I use it), but before roasting they keep no problem for many months. It takes about ten minutes to roast a batch and it's super fresh.
Thanks for the twitter link to the research: http://t.co/HhvDJuAk
The paper's three scenarios assume an old IPCC prediction of 1.8°C to 4°C temperature increase by 2100, which was based on the sudden increase from 1980-1998 continuing. No one believes that any more.
The actual temperature increase by 2100, if any, is much more likely to be similar to the last century overall, with its rises and falls (e.g. 1940-1980), and therefore considerably better than his "best" scenario.
Aron also raises an excellent point, that wild coffee has clearly survived for many thousands of years, and quite probably across several ice ages and interglacial temperatures warmer than today.
The Ethiopian "plateau" with altitudes from 1300m - 3000m (an 11 C temperature range at 6.5 C/km atmospheric lapse rate) would seem to provide an ideal environment for the wild plants to be able to naturally migrate upwards and downwards tracking temperature changes.
I quickly found one coffee producer in Ethiopia whose web site says they are selling wild coffee, not cultivated, growing at altitudes of 1750m - 1850m, which is below the middle of the plateau altitude range (i.e. in the colder half of the survivable climatic temperature range).
The Arduino documentation is actually really opaque on just what Wiring is. They try to aim things at "artists" who are scared of programming.
What Wiring turns out to be is a source code compatible subset of both Java and C++, along with a library containing classes and functions for memory management, timing, serial I/O, digital and analogue I/O, and so forth.
You can write exactly the same code and run it in either a Java or a C++ environment. This is done with the aid of a preprocessor: in the case of Java the preprocessor wraps your code (variable and function definitions) with scaffolding for a class containing everything; in the case of C++ the preprocessor adds declarations for your functions at the top of the file. In both cases the Wiring library is automatically imported.
When you are using the Arduino IDE there is in fact nothing preventing you from using full C++ rather than merely the Wiring subset. The IDE (which is written in Java) uses a standard copy of gcc (targeting AVR or ARM as appropriate) for the code generation and what runs on the Arduino is machine code.
You can easily link in your own assembly language routines if you want.
this is probably because of maps
I'll bet this is happening now because SO MANY people have installed the beta of iOS6 (possibly as many as have ICS on Android!), and many are then complaining about how bad the data is in the new (non Google) Maps application.
The beta is NOT there for random people to use in real life — it's there for developers to test their own apps to make sure they will be compatable with the new OS, and to start to use the new APIs. It's perfectly reasonable for the maps data to be strictly sample data at this stage.
If it's still bad when iOS6 is released to the general public then that's the time to complain, not when you jump through hoops to install an early access version meant only for developers.
I was a bit slow
I was asleep and ordered my Pi about two hours after orders went live on February 29.
On April 27 I got an email from Farnell (nz.element14.com, actually), saying "Having successfully passed its CE compliance testing, we can now confirm that your Raspberry Pi will be dispatched week commencing 28-May-12."
I haven't heard anything more since, so hopefully that's still true.
not really the same thing
While the price is similar to Arduinos intended for general prototyping (e.g. the Uno), they're not really the same thing at all.
On the one side, Arduinos go down to much lower prices, with a bare chip with Arduino boot loader available for about 5 EUR. And Arduinos are suitable for hard real-time programming as they don't run an operating system at all — you're programming the bare metal.
On the other side, the Pi has at least 100 times more CPU power and 100 thousand times as much RAM. And while the Pi has a few pins of digital I/O, it has no analogue I/O at all.
The Pi is much more comparable to something like an old 700 MHz - 1 GHz Pentium III box that you can pick up used for about the same price (I have half a dozen of them in use as routers and so forth so I know). The Pi is much physically smaller and uses less power. But it's lacking any ability to add multiple ethernet cards or the like.
Here for a while yet
I'm an instructor at a gliding club in New Zealand. We do about 50 credit card transactions a month for trial flights worth on average around £100 each. In the last five years we've gotten internet (and electricity!) in our clubhouse, but at this transaction level online authorization machine rental plus transaction fees would cost us far more than the fees for our old zip-zap machine. We get a considerable reduction in fees by phoning in each transaction to get an authorisation number, so it is pretty much as safe for the card issuer as an electronic transaction would be.
The situation may well be different for organisations with a lower average transaction value.
Can't argue with the result, but that seems like a crazy flight path to use to get maximum distance! That thing very nearly didn't pull out of the dive.
Here in NZ...
Dick Smith NZ just in the second half of 2011 stopped stocking components. I'm actually surprised they lasted that long. I remember that even in the 90's Radio Shack in the USA didn't have components any more. That Dick Smith did surely meant that it was a personal non-economic decision by someone in upper management.
I'm going to miss it, but It's a bit rich for Mr Dick Smith to complain about management decisions 30 years after he sold the business.
"attributed to"? The style is unmistakable.
Is the write of the article really not familiar with Steve's history of previous rants? I used to read them religiously, back in his Amazon days.
Stevey's Drunken Blog Rants™
This should be an absolutely painless transition for almost everyone.
Virtually all Arduino programming is done in C++.
They call it "Wiring" but that's just:
- a standard library that provides interface functions and symbolic constants for the I/O pins, timers, USB etc
- a preprocessor that adds a #include for the library header and prototypes for all your functions so you don't have to care about the order
- a driver main() that calls setup() and then repeatedly calls loop()
- the avr-gcc c++ compiler
You already have to select which board you are compiling for from a menu. Different boards are using different AVR chips with varying MHz, varying amounts of flash and RAM and EEPROM, and different constants for the I/O devices. The ARM chip just has more and faster of everything.
There will be an "Arduino Due" item in the "Board" menu which as well as the normal selection of the appropriate library header file will select the arm-gcc compiler instead of the avr-gcc compiler.
No ordinary end-users are likely to have written anything in AVR assembler. 99.9% of everything will Just Work.
Making a big profit at £479
@Ojustaboo, you're 90% right. The only thing I take issue with is that while Apple is making big profits at that price (and could lower it if they needed to), the other manufacturers are barely breaking even, if that.
The reason is that Apple is buying – hell paying in advance, or even giving suppliers the cash to build the factory with – components and assembly services in such volume that they get much better prices than anyone else.
I've had one for years
Here in New Zealand I've had such a smart meter for 2 1/2 years. At present it's only reporting one reading a day at 4 PP, via Vodafone. I don't know whether it's 2G or 3G but I do know it's got a microsim like an iPhone 4.
It's really quite useful to be able to go to the web site and see what usage has been:
The main reason, for me, for getting the smart meter was so that I could get access to cheaper night rates for electricity. If you have separate metering in the day and at night (11 pm - 7 am in this case) then the day rate is the same as for people with a single meter, and the night rate is considerably cheaper (about 25% - 35%, depending on time of year).
I've been able to make some changes that have meant this winter I've used around 45% night and 55% day, while in the autumn (and it should be now again in spring) it was 60% night and 40% day.
That's a considerable saving.
Man who sells middleware for Android predicts Android will win.
While 50% of Mac buyers have never bought a Mac before, it must surely be more like 80% - 90% of iPhone and iPad buyers who have never bought a Mac.
Quite a lot of them will have owned an iPod but as cumulative iOS sales surpass cumulative (non touch) iPod sales that is changing too.
One on't cross beams gone owt askew on treadle?
what took the x86 so long?
Something that hasn't been mentioned is that both the 68000 family (from the 68010 onward) and the PowerPC family supported full hardware-assisted virtualizaton without the kind of partial-emulation hacks needed in VMWare and without the active cooperation of the guest operating system as required by the likes of Xen.
Licensed and trained are not the same thing.
The post is required, and must contain letters.
A machine with no purpose.
I loved the idea of the Cube, but hated the price. $1799? Ridiculous! The Mac Mini might not have been quite so pretty, but at $499 it actually felt like a bargain.
It's a funny old world when being the biggest buying rather than the biggest seller gets you called a "monopoly".
The difference between Apple and the rest is that they're willing to go to suppliers and say "we'll take umpty-million of your product next year and, oh, if you don't have the factory capacity for that then here's some cash now to build the factory with".
It's not only the cash that's the difference, it's also the willingness to commit to buying a huge quantity. For most companies that'd be asking to be left with warehouses of unsold product.
And even then, there are usually shortages of new Apple kit.
"Now how much is $NZ8m in *real* money."
Doesn't matter a lot .. it's a pretty trivial amount of money as tech development goes. But anyway, present conversion as at today is:
They've only got to sell a few hundred to military or search & rescue or similar worldwide to get into profit.
yes that's right
It is "just" a small personal helicopter, able to operate from a couple of square meters and extremely close to obstacles with no risk of rotor strike. It's the smallest practical man-carrying helicopter the world has seen (let's exclude toy quadrocopters, which this has a lot in common with, but which can't carry people).
Some are complaining about the 30 minute endurance. This turns out to be a legal limit for the class of ultralight aircraft it is being operated as. If you are a customer able to operate it under other rules then they could fit a bigger tank for you.
Of course you'd never get even close to that using a rocket or even a jet turbine. The old peroxide-powered Bell Rocket Belt (it's not a belt, it's a pack!!) has an endurance of 30 seconds.
A fundamental property of a rocket engine is the "Isp" (specific impulse). This has units of "seconds" and can be interpreted as the amount of time that the rocket could hover in 1 gravity, ignoring the weight of the engine itself, the tanks, and anything else on board.
Hydrogen peroxide monopropellant rockets have a theoretical maximum Isp of 160 seconds. The improved RB2000 Rocket Belt produces 145 kgf (enough to lift it's own 60 kg weight plus a human weighing less than 85 kg) for 30 seconds from 32 kg (23 liters) of fuel, implying an Isp of about 135 seconds, so there's not much room for improvement. The 1960's version had an Isp of about 100.
The very best chemical rockets give an Isp of about 450, so you're unlikely to ever get a rocket belt with more than about three times the endurance (90 seconds), and it's certainly impossible to get more than 7.5 minutes even if it's carrying its own fuel and nothing else.
A turbojet can achieve an effective Isp of 2000 - 3000, which is much better, but still would give an endurance of no more than about 10 minutes.
It seems to me Martin is using the right technology for a practical machine, regardless of whether it fit's someone's Buck Rogers definition of what a "Jetpack" should be.
 542 seconds was achieved in a rocket engine using liquid lithium, fluorine, and hydrogen but this is .. er .. expensive, dangerous, highly toxic, and just generally impractical.
So, how does a small independent software vendor (aka someone coding in their parents' basement) get into the Amazon store? And what is the revenue split, or other costs?
Or is it just for the big boys such as Microsoft and Adobe?
I haven't seen that one, but I was given this one at about the same time:
C and garbage collection are not mutually exclusive
It's a very long time since I wrote a significant C/C++ program without using the Boehm garbage collector.
It makes life far more pleasant. You have to write a lot less code, think about housekeeping a lot less leaving more time to think about the actual problem you're solving. And it usually makes for faster programs, to boot.
Yes, faster. The overall execution time is usually lower than programs using malloc&free, and certainly much lower than anything where you'd otherwise resort to reference counting.
The downside is slightly higher memory use, and the odd pause while the GC runs. Unless your program uses gigabytes of RAM (some do of course, but most don't) the extra memory is inconsequential these days, and the pauses down under 100 mS which is undetectable unless you're writing a video game.
they lost a feature?
I picked up last year's TH-P42U20Z cheap (about 400 pounds if I convert the NZ$) a month or so ago and it shows a live picture while browsing the FreeView program schedule.
Surely they wouldn't take that away in what appears to be a slightly higher spec new model?
A quarter century of HP28S
I still use my HP 28S, bought in 1986, almost every day. I recently had to put in its 4th set of batteries.
It's a far far better machine than the 12C! Rather than 20 memories and 99 program steps it's got 32 KB of memory, and the stack is as deep as you want it to be, not just a puny 4 levels. Plus it supports complex numbers, binary operations with a selectable word size, strings, vectors, matrices, lists, and algebraic expressions in addition to the native RPN.
It's my 3rd and last calculator. Previous ones were some Casio scientific in 1977, and a Texas Instruments TI57 programmable (8 memories an 50 program steps) in 1979.
what happened to NOTAR?
I don't know why they'd be mucking about with extra tail rotor blades to make it quieter when the NOTAR system was developed 20 years ago!
It's hard to tell the absolute volume level from the video, but I saw a NOTAR flying overhead at 100 ft or so in Phoenix in the early 90's and it was vastly quieter than a standard Hughes 500.
cops can't casually get this data!
No program on the iPhone can access this database unless it has root privileges.
Programs you write yourself, or programs you download, can't do it unless your phone is jailbroken.
The usual jailbreaking techniques result in a wiped phone, and you have to reload your data from the latest backup on your computer.
This is not something that a cop can do in a few seconds at a traffic stop.
The data that people are analyzing is coming form the backup copy on their computer, not directly from their phone.
I've looked at my own data (http://a.yfrog.com/img640/2665/qso.png) and it shows scattered sites that I haven't ben within 50 or 100 km of. The three across on the South Island, for example, along with the ones near New Plymouth and also the vast number on the east coast near Napier.
I can explain these by having been flying at altitude in a glider 50 - 100 km away, and/or being high on a skifield on Mt Ruapehu in the central North Island, and my phone being able to pick up cell sites unusually far away.
A glider flight I made in February, along the far side of mountains 60 - 70 km from Napier:
For those not in NZ, most of the actors are AirNZ staff, but there's also this guy:
No problem in practice
The funny thing is, when I was camping in some remote spots this (southern) summer, very often the people with iPhone 4's were happily using SMS or on the internet while other types of smartphones had nothing at all.
When you're not bridging it, the external antenna works brilliantly.
There is nothing to stop someone who gets a short tweet from passing that message on. Why should they be stopped from passing the URL of the full version on?
the problem is clear
The privacy problem is obvious. When a message is too long for twitter it is further truncated and a URL is inserted to the full version stored elsewhere. Something like http://tweet.dk/x17dg for example. The problem is the URL is too easily guessed and thus accessed by people who can't see the tweet itself. It might even be listed under the user's name.
They have chosen to fix this by simply not allowing long messages in private messages or if you have a locked profile.
A better fix would be to use the MD5 or SHA1 of the long message as its URL. e.g. http://tweet.dk/8ed3f9c68c4313a70b3dce05391e805c.
That's 48 characters but could be shortened to 38 if it used base64 instead of hex. Worse than the 20 or so now, but not too horrid and far harder to guess.
The latest JailBreak has already worked around this.
The Drobo S is a crazy price! Almost double the price of the original Drobo for what? eSata, 1 more bay, and the option to enable double-failure protection (which then uses up that extra bay...).
It makes the US$349 pricing on the original look very reasonable.
As for rolling your own using a PC chassis with lots of drive bays ... that's fine except when you run out of capacity. With standard RAID software you'll either have to add a new RAID array (yes, even if you use ZFS you can't just add to an existing RAID set), or you'll have to replace all the existing drives at once which is also going to require a 2nd RAID array at least temporarily while you copy the data.
It's the ability to expand the storage literally by pulling out one drive and pushing a new one in that makes the Drobo attractive. I've currently got 3 x 1.5 TB drives in mine and it's about 70% full. By the time I need more space I'm hoping that 3 TB drives will be the cheapest per byte (the 2's are at the moment).
I have one .. works for me
I've had a 2nd generation 4 bay Drobo for nearly two years. That's with Firewire800 instead of USB-only, and a much improved controller which can actually give a decent data transfer speed (I just checked and got 23 MB/sec copying a movie from the Drobo to this PC over my home network, and 10 MB/sec to an old MacBook Air over WIFI).
The prices charged by the Australian and NZ agent have been quite obscene -- over NZ$1000 two years ago and still $750 now. I bought mine from Amazon for US$349 (NZ$500 at the time) on a trip to the USA.
As far as I can see it does the job and does it well. As an earlier commenter suggested, I have all my CDs and DVDs on it (in both lossless and lower bitrate versions), and also backups for my various PCs.
Everything that is on the Drobo that is not just a backup of another machine, I also have on a stand-alone (non-redundant) external drive. Maybe the Drobo can get corrupted somehow, but for sure someone could steal it.
If they could drop their prices a bit (and I don't see why they can't), I'd recommend them to a LOT of home users I know. But at the moment it's far cheaper to just buy a couple of those 2 TB WD external drives that go for about $100 each.
Gosh .. at those trivial data rates you could very easily have even a simple Perl script with a huge regexp that tail'd the feed and fired off an SMS or Tweet to the relevant person whenever a car of interest was spotted.
- Just TWO climate committee MPs contradict IPCC: The two with SCIENCE degrees
- 14 antivirus apps found to have security problems
- Apple winks at parents: C'mon, get your kid a tweaked Macbook Pro
- Feature Scotland's BIG question: Will independence cost me my broadband?
- Driverless car SQUADRONS to hit Britain in 2015