I think that the Register did them all when they noticed that we've bought a bunch of "Hunter Killer" drones that will be run by this "Skynet". New Labour, New Britain - doing the best we can to help Judgement Day along one death at a time...
Wiki is meant to be authoritive - that means all the way from a beginner's entry to the subject to the accurate detailed facts about the topic. This thread is a false dichotomy. Wiki should not have to lean towards one extreme or the other - the only reason to do so is because of lack of space. Remember "wiki is not paper".
You'd be suprised. Down here in the Navy we talk about sodomy all the time. And those other mainstays of civilization; rum and the lash.
More seriously, language patterns have changed. This is quite a bad example as "Sod" and "sodomy" are not commonly used any more, but "bugger" is, or "anal". It just reflects language patterns of a couple of centuries ago. Which would be the rough timeframe of the "Rum, sodomy and the lash" quip.
Maybe I'm bucking the trend so far, but I found the reformatted versions harder to read than normal text. You're right about their bad comparison - but comparing their "poetic" formatting against normal text on a webpage (not their example) makes me think that ther technique makes it harder to read.
Their "revelation" about how the eyes scan a page is well known and understood in page design and layout. Also, the idea that the brain has to remove "clutter" from the surrounding words is false. The brain uses the pattern of the text above and below to help the eye scan back to the beginning of the line quickly. Also the brain interprets the surrounding text to get an earlier chance to parse what is coming. The line underneath is processed before it is consciously read, kind of a warm-up run.
Sadly I can't remember where I read this, or find a reference to it...
In Sony's case it's even worst. Since the beginning of the PS3 marketing they kept throwing fake pre-rendered videos at us (Killzone 2 anyone?). then once it's out their console can't even match their fake videos.
Actually you can go right back to when they were throwing fake pre-rendered videos at us before the launch of the PS2. Or the dev demos from back then - anyone remember the disembodied head that we were told would be an accurate indication of characters in PS2 games...
I still play on the ps2 now and again... but I really don't see the point in a ps3. It's not the graphics that are the problem. I will be buying a Wii just as soon as the uk price comes down from its ridiculous level. They just look like more fun.
Actually it's the other way around. It's a latency issue - so in a movie you can start the loading 1/4 second before you need that data because you know it's coming. In a game you are responding to the players movements so you don't know that you need the data until they cross some line, then it's 1/4 second too late. You can't hide that glitch in the background so it affects the game performance.
When loading a move you know exactly what order you're going to need the data off of the drive as it is linear. So the instant layer switch doesn't mean that the laser changes any quicker - it's more likely that it prebuffers.
Guess how that maps onto a seamless game with constant loading - yup, you don't know when you're going to cross from the top to the bottom layer and hence the game may glitch (very noticably) for a 1/4 second.
OK, now that we've established that you're a dumbass, there is a way around it for rockstar - but it is more work, and they might be pushing the drive to its limits to stream their data already. They could replicate all non location data (everything but the maps, geometry and local textures) on both layers, and then use some kind of spatial subdivision to split the world. They probably do this already using a quadtree or something similar. Have a line down the middle of the gameworld for the top/bottom layer and then replicate the stuff around that line on both layers. Use the hystersis this creates do a little preloading when approaching the layer so that you can take the 1/4 second hit...
<< Who? I know of no outspoken proponents of environmentalism that come off like "rambling kooks".>>
Your examples are a little hit and miss (well, ok mostly hits wit a couple of misses).
The right to have children should be a marketable commodity, bought and traded by individuals but absolutely limited by the state. Bloody good point. Not for environmental reasons though, just because anyone who is an idiot shouldn't be allowed to breed. Let's actually get some environmental selection going for the species. I would recommend Takeshi's Castle as a method of selecting who is eligible.
Free Enterprise really means rich people get richer. They have the freedom to exploit and psychologically rape their fellow human beings in the process.... Capitalism is destroying the earth. Almost right. Most of us in the world don't have "Free Enterprise" because unlimited capitalism would be as she describes. That's why we have capitalist societies within the limits regulated by law.
The rest sounded like freaks and kooks, perfect material for the B-ark. Now all we need is some kind of impending disaster story to get them off-world....
Yesterday there was an article in the Independent about a large wave powered station off the coast of Cornwall. The thing that struck me as odd is that in the UK the 20MW station will supply about 7500 "homes" - always a strange piece of statistics. In Canada the 40MW solar station will supply about 10000. Is this purely down to different levels of power consumption on either side of the Atlantic, or is the exchange rate for Canadian Watts pretty bad?
Interesting. It must be possible, because it happened. As far as I know, the engineer didn't lose his job. Our number (and DSL account) was moved from one line to another. And then it worked afterwards... go figure.
You sound American (by your use of phone companies - over here BT still has a defacto monopoly in all but name). When you pay a DSL fee over here part of the money goes from the ISP to BT as rental for the equipment installed at the line exchange. So you are still paying some money to them indirectly.
Other people have replied and asked about how to go about this. We did it by accident once, I'm not sure how to do it officially. BT's phone system is a maze, but somewhere on the web there are direct dial numbers for a bunch of companies / shortcut keys to drop out of voicemail menus. When we had a problem I got transferred to some random engineer at an exchange who adjusted our account for us. YMMV.
To be fair, this guy also transferred our DSL from one phone line to another, something that the ISP denied was possible, and tried to charge us a £60 re-activation fee for. They argued the case until we cancelled the account and signed up with another ISP - without an activation fee as DSL was setup on our new line. One more reason to avoid plusnet like the plague...
Finally, the phone companies might not like the loss of revenue, but if there is no technical reason to enforce the bundling they may be forced to accept it. Take a look at the recent bank charges scandal here in the UK for a similar example.
Yes and no. It doesn't use the circuit as the higher-frequencies piggyback the line, but you do still need a line for those higher frequencies to piggyback. Here in the UK as long as the line is in place it can be deactivated for voice (so no line rental) but still used for DSL.
Okay, there are two lines of argument here that I'll summarise (correct me if you disagree):
x Is TRIPS an advance on academic research in the area? x Does TRIPS describe a shift in commercially ready parallel platforms?
From what you've said you seem to think that the second point is down to me shifting the argument. I would disagree, the article summary and press releases from the team seem to support the idea that they think they have something that radically shifts the current state of the art. From what I've read in scanning their publications I would disagree with this point. I'm not sure if you support it or not, but you do seem to say that this is not an issue, when the slashdot article would suggest otherwise.
From a commercial point of view I don't think that what they've done will shift the market. Over the years there have been many exotic parallel architectures that have not shifted the mainstream because their gains do not apply to real-world code. In the compiler papers section there is some evidence that they have made an advance here. In particular it's interesting that they can accelerate gzip, given the workload in that benchmark.
Of course there have also been exotic architectures that have become mainstream - GPUs are one example. So I don't think their research has market potential to replace the current mainstream x86 variants. This does not mean that I don't think that they've done anything interesting, as I said, and you decided not to quote.
From a research point of view, they have two advances; a new architectural layout that exploits parallelism with a certain tradeoff, and compiler technology that restructures conventional code to take advantage of this layout. They do argue quite heavily in each paper that it is a very different architecture to a dataflow machine, or a vector layout. I'm not convinced about this point, mainly because their arguments are quite buzzword heavy, and I don't see any substance.
The compiler side is more interesting, but it doesn't appear to be validated. They are comparing their performance on benchmarks against a standard compiler. They should be comparing the speedup they achieve against rival approaches. In particular this would give some idea of how efficiently they are exploiting the transistor budget with their design. This is standard in the literature, but it doesn't make it correct to phrase the validation that way.
I hadn't heard of Yale Patt but it seems quite a nice comparison (for me). My background is more in language design and compilation than in architecture. I've read quite a lot of papers on random exotic designs for parallel architectures that haven't gone anywhere, so I have a healthy (but possible misplaced) cynicism on the topic.
My comments about media applications were a guess that their architecture would require coarse-grained data parallelism to show off any real gains in performance. Their description of loop mining and tiling techniques seem to back that up, although they show interesting results on fairly sequential benchmarks.
Apart from your selective misquoting you haven't really said anything. From your other posts I would guess that you are (loosely) associated with the project / people on it. Please back up your claim about breaking code up into packets (preferably with a paper citation), because if they have done that I would like to read it.
That's not the point of research
From the way that you misquoted me and then attacked a strawman, either you don't understand what research is about, or you do know but scoring points is more important to you. At no point did I claim that research must be directed at commercial results. As someone who did his PhD in an area that has no (foreseeable) commercial application I don't really need an explanation of this from you.
Multiway branching is ancient, and it's not used much because it's very inefficient. At least half of the instruction stream after a branch will be canceled, two branches deep it is 75% and so on. No matter how much parallelism ou throw at this there are only marginal gains to made (exponential increase in number of execution units for a linear increase in depth). It still doesn't get around data dependencies which will be the major bottleneck if looking that far ahead in the instruction stream.
Having read the articles that were easy to get to, and the abstract of the PhD student: this is buzzword bollocks. There is no innovation in what they have done. As other people have pointed out this is a vector / datastream architecture. It's not a very good one at that. Although it has the "potential" to scale to terraflops, so does my toothbrush. On a 130 process they can fit 2 cores with 32-wide dispatch clocked at 500Mhz. My 7800 is fab'ed on a 130 process with 24*4*4 = 384 operation wide vector dispatch. This prototype would hit about 16 billion ops/sec, versus 180 Gflop on the 7800. This is a long way from terraflops, and doesn't convince me that it can scale.
As the 7800 is close to a systolic model there is a limited class of programs that can be executed; but those that are in that class exhibit (near)perfect parallelism and so have zero hit from memory access costs. Actually the internal bandwidth on the 7800 is a bottleneck for some computations but I'm just going for coarse detail here.
Edge appears mix and match ideas from several parallel designs; every one of which suffers from hard code generation problems. I suspect that the only sample applications that hit 32 ops / cycle are media apps (or dataflow problems as they used to be called) which normal architectures run at high speed anyway.
Interesting research, as it's always good to see people explore different designs, but it sounds overhyped and I believe that it has zero commercial appeal. Finally, as a sidenote, you are right about cache latencies being a memory defect rather than processor but there are ways around it. If you are willing to limit yourself to a certain class of applications (roughly the same one that executes well on most parallel architectures such as this, or GPUs) then you can completely avoid the latency. This provides a much bigger performance hike than any other technique as memory latency is a dominating factor in most runtimes. The only snag is that it is very hard to do, requires different fabrication technology (largely solved now), and lots of compiler advances... If you're interested then google for intelligent ram. It's about a decade of research now...
It sounds more like a patent on a make system. The description in the technical content section describes regenerating web-pages after assets have been changed without a programmer being involved. I'd say that make, or even visual studio would count as prior art. The patent was filed in 1999 but build systems have been around for decades.
Thankyou. That was a very rare post on slashdot, with some genuine insight. You've nailed exactly what disappoints me in Oblivion. It is certainly a step-up from Morrowind (pretty graphics, fixing the obvious game mechanics holes) but... it's not a very large step. When I walk into the Imperial city of Tameriel, that is so imposing I can see it from everywhere in the surrounding valley, it looks like a sparse little village with one or two people dotted around. The game sets up expections but then it dashes them.
The new outdoors look with vegetation and trees looks awesome... at about 5fps. Turning the vegetation off gets an acceptable framerate, but the limitation here is not the grunt on the graphics card. It's the CPU not doing enough LOB calculation and billboard generation. This is exactly where adding some power to the brains (CPU) of the system would improve the result of applying the brawn (GPU).
The game that I really want to see somebody write is quite like you describe, it's Nethack with the Oblivion game engine. It's starting each game in a random city in a generated world that is consistent, and has a generated back story. Oh well, it's going to be a wet dream for a few more years at least... back to my thieves guild initiation test...
Because despite what you claim you didn't read the article (properly) ? Fair enough, it was almost a whole page without the adverts.
The doubling in speed is for a different technology - "turbo memory" under a particular (memory-bound) application.
The speed-up for "overclocking" the core is unlikely to be as much as 2x. When you have a multi-threaded app (or several apps) then you want both cores because you'll get more performance that way. When one core is not being utilised the other core can increase its clockspeed by a small amount because the heat generated by the other core is not a problem. As somebody else pointed out above, the heat produced is roughly proportional to the square of the clockspeed, so you can get away with boosting one core a little bit.
In short, the reason that you have two cores in the first place is that doubling the performance of a single core is not feasible within a reasonable thermal envelope.
It is not really suprising that statistics can be made to say anything that you want. It would be suprising if the web presence of the US / Europe / Asia was in decline. What could cause this momentous change in direction? Is this a new trend that could change the face of the world as we know it???
Err, no. It's just some twat pumping traffic to his site. So lets look at what he's done shall we:
* Traffic is declining to first-world web domains!!! No, not true. The relative share of domains in the top-500 has decreased. Overall traffic and numbers of domains are still increasing. Ahh, so what is being measured as a "top-500" site? Obviously we can bias this any way we want. Does it explain anywhere on the site how this measurement is performed.... no.
* All URL's are geographically based!!! No,.com does not imply an American company. So the shift in traffic from generic TLD's to country code TLD's for sites in this mythical "top-500" could be explained by a change in presence. That companies no longer want a generic TLD for a "global presence" and instead want a national image. That would be an interesting explanation of the "data", but no, lets go for a screaming headline to pump up traffic in our slashvertisment.
* Believe what I've told you!!! We have bold claims about traffic to a wide range of internet domains. There is no description on the site of how the data is gathered. Is this opt-in traffic reporting? Does this guy happen to own a large amount of internet infrastructure? Is it one of the largest benign bot-nets in existence? Or is it the answer behind door D...
Slashdot Mirror
I think that the Register did them all when they noticed that we've bought a bunch of "Hunter Killer" drones that will be run by this "Skynet". New Labour, New Britain - doing the best we can to help Judgement Day along one death at a time...
No, I'm afraid to say you're completely wrong.
Wiki is meant to be authoritive - that means all the way from a beginner's entry to the subject to the accurate detailed facts about the topic. This thread is a false dichotomy. Wiki should not have to lean towards one extreme or the other - the only reason to do so is because of lack of space. Remember "wiki is not paper".
GP uses a humorous post. Technical criticism is not applicable and is thus useless.
You'd be suprised. Down here in the Navy we talk about sodomy all the time. And those other mainstays of civilization; rum and the lash.
More seriously, language patterns have changed. This is quite a bad example as "Sod" and "sodomy" are not commonly used any more, but "bugger" is, or "anal". It just reflects language patterns of a couple of centuries ago. Which would be the rough timeframe of the "Rum, sodomy and the lash" quip.
Maybe I'm bucking the trend so far, but I found the reformatted versions harder to read than normal text. You're right about their bad comparison - but comparing their "poetic" formatting against normal text on a webpage (not their example) makes me think that ther technique makes it harder to read.
Their "revelation" about how the eyes scan a page is well known and understood in page design and layout. Also, the idea that the brain has to remove "clutter" from the surrounding words is false. The brain uses the pattern of the text above and below to help the eye scan back to the beginning of the line quickly. Also the brain interprets the surrounding text to get an earlier chance to parse what is coming. The line underneath is processed before it is consciously read, kind of a warm-up run.
Sadly I can't remember where I read this, or find a reference to it...
Actually you can go right back to when they were throwing fake pre-rendered videos at us before the launch of the PS2. Or the dev demos from back then - anyone remember the disembodied head that we were told would be an accurate indication of characters in PS2 games...
I still play on the ps2 now and again
...or you could encode the 3-states on two wires and reinvent null convention logic as used in Asychronous Circuits
Actually it's the other way around. It's a latency issue - so in a movie you can start the loading 1/4 second before you need that data because you know it's coming. In a game you are responding to the players movements so you don't know that you need the data until they cross some line, then it's 1/4 second too late. You can't hide that glitch in the background so it affects the game performance.
Nicely put.
... he was an idiot and I'm impatient
But
Ok, you're stupid.
When loading a move you know exactly what order you're going to need the data off of the drive as it is linear. So the instant layer switch doesn't mean that the laser changes any quicker - it's more likely that it prebuffers.
Guess how that maps onto a seamless game with constant loading - yup, you don't know when you're going to cross from the top to the bottom layer and hence the game may glitch (very noticably) for a 1/4 second.
OK, now that we've established that you're a dumbass, there is a way around it for rockstar - but it is more work, and they might be pushing the drive to its limits to stream their data already. They could replicate all non location data (everything but the maps, geometry and local textures) on both layers, and then use some kind of spatial subdivision to split the world. They probably do this already using a quadtree or something similar. Have a line down the middle of the gameworld for the top/bottom layer and then replicate the stuff around that line on both layers. Use the hystersis this creates do a little preloading when approaching the layer so that you can take the 1/4 second hit...
Your examples are a little hit and miss (well, ok mostly hits wit a couple of misses).
The right to have children should be a marketable commodity, bought and traded by individuals but absolutely limited by the state. Bloody good point. Not for environmental reasons though, just because anyone who is an idiot shouldn't be allowed to breed. Let's actually get some environmental selection going for the species. I would recommend Takeshi's Castle as a method of selecting who is eligible.
Free Enterprise really means rich people get richer. They have the freedom to exploit and psychologically rape their fellow human beings in the process.... Capitalism is destroying the earth. Almost right. Most of us in the world don't have "Free Enterprise" because unlimited capitalism would be as she describes. That's why we have capitalist societies within the limits regulated by law.
The rest sounded like freaks and kooks, perfect material for the B-ark. Now all we need is some kind of impending disaster story to get them off-world....
Yesterday there was an article in the Independent about a large wave powered station off the coast of Cornwall. The thing that struck me as odd is that in the UK the 20MW station will supply about 7500 "homes" - always a strange piece of statistics. In Canada the 40MW solar station will supply about 10000. Is this purely down to different levels of power consumption on either side of the Atlantic, or is the exchange rate for Canadian Watts pretty bad?
Interesting. It must be possible, because it happened. As far as I know, the engineer didn't lose his job. Our number (and DSL account) was moved from one line to another. And then it worked afterwards... go figure.
You sound American (by your use of phone companies - over here BT still has a defacto monopoly in all but name). When you pay a DSL fee over here part of the money goes from the ISP to BT as rental for the equipment installed at the line exchange. So you are still paying some money to them indirectly.
Other people have replied and asked about how to go about this. We did it by accident once, I'm not sure how to do it officially. BT's phone system is a maze, but somewhere on the web there are direct dial numbers for a bunch of companies / shortcut keys to drop out of voicemail menus. When we had a problem I got transferred to some random engineer at an exchange who adjusted our account for us. YMMV.
To be fair, this guy also transferred our DSL from one phone line to another, something that the ISP denied was possible, and tried to charge us a £60 re-activation fee for. They argued the case until we cancelled the account and signed up with another ISP - without an activation fee as DSL was setup on our new line. One more reason to avoid plusnet like the plague...
Finally, the phone companies might not like the loss of revenue, but if there is no technical reason to enforce the bundling they may be forced to accept it. Take a look at the recent bank charges scandal here in the UK for a similar example.
Yes and no. It doesn't use the circuit as the higher-frequencies piggyback the line, but you do still need a line for those higher frequencies to piggyback. Here in the UK as long as the line is in place it can be deactivated for voice (so no line rental) but still used for DSL.
Okay, there are two lines of argument here that I'll summarise (correct me if you disagree):
x Is TRIPS an advance on academic research in the area?
x Does TRIPS describe a shift in commercially ready parallel platforms?
From what you've said you seem to think that the second point is down to me shifting the argument. I would disagree, the article summary and press releases from the team seem to support the idea that they think they have something that radically shifts the current state of the art. From what I've read in scanning their publications I would disagree with this point. I'm not sure if you support it or not, but you do seem to say that this is not an issue, when the slashdot article would suggest otherwise.
From a commercial point of view I don't think that what they've done will shift the market. Over the years there have been many exotic parallel architectures that have not shifted the mainstream because their gains do not apply to real-world code. In the compiler papers section there is some evidence that they have made an advance here. In particular it's interesting that they can accelerate gzip, given the workload in that benchmark.
Of course there have also been exotic architectures that have become mainstream - GPUs are one example. So I don't think their research has market potential to replace the current mainstream x86 variants. This does not mean that I don't think that they've done anything interesting, as I said, and you decided not to quote.
From a research point of view, they have two advances; a new architectural layout that exploits parallelism with a certain tradeoff, and compiler technology that restructures conventional code to take advantage of this layout. They do argue quite heavily in each paper that it is a very different architecture to a dataflow machine, or a vector layout. I'm not convinced about this point, mainly because their arguments are quite buzzword heavy, and I don't see any substance.
The compiler side is more interesting, but it doesn't appear to be validated. They are comparing their performance on benchmarks against a standard compiler. They should be comparing the speedup they achieve against rival approaches. In particular this would give some idea of how efficiently they are exploiting the transistor budget with their design. This is standard in the literature, but it doesn't make it correct to phrase the validation that way.
I hadn't heard of Yale Patt but it seems quite a nice comparison (for me). My background is more in language design and compilation than in architecture. I've read quite a lot of papers on random exotic designs for parallel architectures that haven't gone anywhere, so I have a healthy (but possible misplaced) cynicism on the topic.
My comments about media applications were a guess that their architecture would require coarse-grained data parallelism to show off any real gains in performance. Their description of loop mining and tiling techniques seem to back that up, although they show interesting results on fairly sequential benchmarks.
From the way that you misquoted me and then attacked a strawman, either you don't understand what research is about, or you do know but scoring points is more important to you. At no point did I claim that research must be directed at commercial results. As someone who did his PhD in an area that has no (foreseeable) commercial application I don't really need an explanation of this from you.
Yeah, it's been under appreciated :)
Multiway branching is ancient, and it's not used much because it's very inefficient. At least half of the instruction stream after a branch will be canceled, two branches deep it is 75% and so on. No matter how much parallelism ou throw at this there are only marginal gains to made (exponential increase in number of execution units for a linear increase in depth). It still doesn't get around data dependencies which will be the major bottleneck if looking that far ahead in the instruction stream.
Having read the articles that were easy to get to, and the abstract of the PhD student: this is buzzword bollocks. There is no innovation in what they have done. As other people have pointed out this is a vector / datastream architecture. It's not a very good one at that. Although it has the "potential" to scale to terraflops, so does my toothbrush. On a 130 process they can fit 2 cores with 32-wide dispatch clocked at 500Mhz. My 7800 is fab'ed on a 130 process with 24*4*4 = 384 operation wide vector dispatch. This prototype would hit about 16 billion ops/sec, versus 180 Gflop on the 7800. This is a long way from terraflops, and doesn't convince me that it can scale.
As the 7800 is close to a systolic model there is a limited class of programs that can be executed; but those that are in that class exhibit (near)perfect parallelism and so have zero hit from memory access costs. Actually the internal bandwidth on the 7800 is a bottleneck for some computations but I'm just going for coarse detail here.
Edge appears mix and match ideas from several parallel designs; every one of which suffers from hard code generation problems. I suspect that the only sample applications that hit 32 ops / cycle are media apps (or dataflow problems as they used to be called) which normal architectures run at high speed anyway.
Interesting research, as it's always good to see people explore different designs, but it sounds overhyped and I believe that it has zero commercial appeal. Finally, as a sidenote, you are right about cache latencies being a memory defect rather than processor but there are ways around it. If you are willing to limit yourself to a certain class of applications (roughly the same one that executes well on most parallel architectures such as this, or GPUs) then you can completely avoid the latency. This provides a much bigger performance hike than any other technique as memory latency is a dominating factor in most runtimes. The only snag is that it is very hard to do, requires different fabrication technology (largely solved now), and lots of compiler advances... If you're interested then google for intelligent ram. It's about a decade of research now...
It sounds more like a patent on a make system. The description in the technical content section describes regenerating web-pages after assets have been changed without a programmer being involved. I'd say that make, or even visual studio would count as prior art. The patent was filed in 1999 but build systems have been around for decades.
Thankyou. That was a very rare post on slashdot, with some genuine insight. You've nailed exactly what disappoints me in Oblivion. It is certainly a step-up from Morrowind (pretty graphics, fixing the obvious game mechanics holes) but ... it's not a very large step. When I walk into the Imperial city of Tameriel, that is so imposing I can see it from everywhere in the surrounding valley, it looks like a sparse little village with one or two people dotted around. The game sets up expections but then it dashes them.
The new outdoors look with vegetation and trees looks awesome... at about 5fps. Turning the vegetation off gets an acceptable framerate, but the limitation here is not the grunt on the graphics card. It's the CPU not doing enough LOB calculation and billboard generation. This is exactly where adding some power to the brains (CPU) of the system would improve the result of applying the brawn (GPU).
The game that I really want to see somebody write is quite like you describe, it's Nethack with the Oblivion game engine. It's starting each game in a random city in a generated world that is consistent, and has a generated back story. Oh well, it's going to be a wet dream for a few more years at least... back to my thieves guild initiation test...
Come on, you're not getting with the numerology vibe here. It's obvious:
The prime factors of 6 are 2 and 3.
2/3 = 0.666666666
If we start at digit 2 and take 3 digits then we get 666.
Carol Vorderman eat your heart out.
Because despite what you claim you didn't read the article (properly) ? Fair enough, it was almost a whole page without the adverts.
The doubling in speed is for a different technology - "turbo memory" under a particular (memory-bound) application.
The speed-up for "overclocking" the core is unlikely to be as much as 2x. When you have a multi-threaded app (or several apps) then you want both cores because you'll get more performance that way. When one core is not being utilised the other core can increase its clockspeed by a small amount because the heat generated by the other core is not a problem. As somebody else pointed out above, the heat produced is roughly proportional to the square of the clockspeed, so you can get away with boosting one core a little bit.
In short, the reason that you have two cores in the first place is that doubling the performance of a single core is not feasible within a reasonable thermal envelope.
I didn't think there was anything wrong with the set of laws in the reg article.But then I use a very simple "rule of thumb" to navigate these laws:
If you don't like the AI's ability to fire at your weapon system then get out of the tank, and run like hell.
It is not really suprising that statistics can be made to say anything that you want. It would be suprising if the web presence of the US / Europe / Asia was in decline. What could cause this momentous change in direction? Is this a new trend that could change the face of the world as we know it???
.com does not imply an American company. So the shift in traffic from generic TLD's to country code TLD's for sites in this mythical "top-500" could be explained by a change in presence. That companies no longer want a generic TLD for a "global presence" and instead want a national image. That would be an interesting explanation of the "data", but no, lets go for a screaming headline to pump up traffic in our slashvertisment.
Err, no. It's just some twat pumping traffic to his site. So lets look at what he's done shall we:
* Traffic is declining to first-world web domains!!!
No, not true. The relative share of domains in the top-500 has decreased. Overall traffic and numbers of domains are still increasing. Ahh, so what is being measured as a "top-500" site? Obviously we can bias this any way we want. Does it explain anywhere on the site how this measurement is performed.... no.
* All URL's are geographically based!!!
No,
* Believe what I've told you!!!
We have bold claims about traffic to a wide range of internet domains. There is no description on the site of how the data is gathered. Is this opt-in traffic reporting? Does this guy happen to own a large amount of internet infrastructure? Is it one of the largest benign bot-nets in existence? Or is it the answer behind door D...
Complete, and utter, bullshit.