Fortunately for the UK, we have an active resistance group that goes by the initials EDS. Their typical modus operandi for thwarting this kind of attack on our freedoms is to bribe or mislead civil servants to be awarded the contract for delivering the system, and then to delay and delay, while pushing the budget up, until another government takes power. The new government then blames the cost overruns on the previous one and cancels the project.
Intel has always had the fab advantage. Previously, they used this to make bigger caches and to ramp up clock speeds. The Pentium 4 was over 2GHz when AMD wasn't shipping anything above 1.5GHz - AMD had to make up the performance gap with more efficient designs. With the Core microarchitecture, they switched their focus to power efficiency. AMD kept focussing on raw performance. This left AMD with nothing competitive in the two biggest-growing markets: low-power servers and laptops.
You're talking a complete cycle earlier. AMD bet on servers at the height of the.com boom. The result was the Opteron. Cut down versions (Athlon64) did well in the desktop. In the next cycle, they bet that they could just keep on delivering more of the same, but the market changed. Server buyers started to care about performance per Watt, because they started being limited by the cooling capacity in their data centres more than their hardware budget. More people bought laptops than desktops, and this let Intel put things that are basically laptop chips (with more cache) and a much bigger heatsink into desktops.
Their branch predictors are the best in the business.
And this isn't even new. I found that some of my code was suffering quite badly from a single unpredicted branch. Rewriting it so that the processor got it right gave about a 20% speedup across the whole program on a 1GHz Athlon. On a 1.2GHz the Pentium M (which uses the P4's branch predictor), it might have seen a 2% speedup, but that was in the error margin for the benchmark.
Otherwise, how else does one explain that their processors are still popular in some supercomputers,
They were popular with supercomputer vendors for one very simple reason: HyperTransport. This made it much easier to plug in a proprietary interconnect for doing NUMA with a thousand or so processors than the northbridge arrangement that Intel used. With QPI, this advantage is gone. Now they are popular because the big supercomputers typically take 2-3 years to build, and the current set were designed before QPI was introduced.
Not even remotely true. GPUs are stream processors. They are designed to take a big blob of memory, run the same algorithm on every subunit, and create the output as another big blob of memory. They are also designed for code that is very light on branching. They are terrible for things like finite element simulations (which are used in a lot of engineering and physics simulation).
AMD made good products too, they just made them for the wrong market. This is why commercial semiconductor manufacturing is so difficult. You give your engineers a set of constraints and then about 5 years later you have a product. Intel, back around 2003, bet heavily on laptops and power-conscious servers. AMD bet on desktops. Intel's market predictions were better and so the products that they brought to market turned out to be the ones customers wanted by the time the related products made it to market. AMD's were not. Of course, Intel only made this bet after seeing how badly they'd underestimated the importance of power consumption with the Pentium 4 so, if anything, their later products were an overreaction to the poor performance of the P4.
So, in summary, AMDs problem was that they didn't screw up the previous generation, so assumed that the next one could be more of the same and missed the industry shift to mobile devices.
66 non-whitespace characters is about optimal. Unfortunately, the people who insist on 80-colums as a hard limit tend to be the same people who insist on 8-character tabs for indent, so you end up with a lot fewer than 66 characters per 80-column line.
It's a fine line to pick the exact ratio that's believable. If you're offering 10% above market rate, then people will likely just believe that they got a good deal. If you're offering 100% above the market rate, then they'll think that there's something a bit fishy going on. You need to offer a good enough deal that they thing they're getting the better of you, but not such a good one that they become suspicious. This is where most 419 scammers fail: people are far more likely to believe that they can get $100 due to a bank error than $1,000,000.
Does the script run in a sandboxed environment or an already-installed program (e.g. mathematica, MS Office, cmd.exe)? If so, fine. If not, justify it to your boss. If it's really that useful, then maybe it should be part of the default image...
Unfortunately, melting them down is illegal in the USA, as is exporting them in anything other than very small quantities, so melting them down in another country is difficult too...
Your 'standard Windows corporate lockdown setup' allows end users to run untrusted code that they downloaded from the Internet. I can think of many reasons for calling Google evil, but in this case they are simply doing something that, since Vista, has been a requirement for the 'Designed for MS Windows' logo and part of the recommended practices: allowing non-admin user to install for their own user. It's only 'a nice little unpublicized exploit' if you don't count the articles on MSDN telling you 'this is what you must do in a UAC world'.
It's not Google's fault that you think removing write access to C:\Program Files is the same as preventing users from running their code. Windows has fine-grained ACLs. Learn how to use them. Remove the user's ability to run programs that are installed in any location that they have write access to.
And now I've defended Windows, I need to go and have a shower...
Well, unless they choose to share that data, of course. Google can easily give you some server-side code that just forwards requests made by gStalker.js to their servers, rather than processing it locally.
There's also nothing stopping Google from making Google Ads and Analytics users set a subdomain for these to come from.
I've seen it called 'cheese food slices', presumably in the same way that 'juice drink' can mean anything that contains 10% juice and tastes vaguely of fruit.
Moore's law doesn't say what you seem to think it says. It says that the number of transistors that you can fit on a chip for a fixed cost doubles every 18 months (for some value of 18). This may mean that the price of the same chip halves every 18 months (which is roughly true, although if you're buying something like Z80s the cost of the package became the limiting factor some time around the mid '90s). Or it may mean that for the same cost you get twice as many transistors to play with. For a while, it was easy to use these for things like extra integer units, then bringing floating point on-die, then adding SIMD units, and adding extra cache and seeing immediate returns. A few years ago this stopped being the case and the easiest way of getting more speed from more transistors became to add extra cores. Clock frequency is totally independent of Moore's Law.
It's actually a misquote. He said that there 64KB ought to be enough for anyone, and it was a comment on one of the limitations of Microsoft BASIC for 8-bit systems. It's not clear whether it was a serious comment or was intended to be tongue in cheek...
Actually, if you check, the quote is pretty well sourced. However, since the person who said it worked for a company that had sold over a dozen computers that year alone, it is more evidence of total executive ignorance than anything else.
No, actually. I wrote an article about some of the alternatives last year. MRAM, FeRAM, and PCRAM have all been under development by various companies for well over a decade. The only real new discovery is memrister memory. The reason the other three are starting to appear on sites like Slashdot is that they're now getting to the point where they can make the transition to shipping product.
The problem is, having a 50Hz CRT does not require a license, operating a device capable of receiving broadcast TV signals does. If you have a TV that is connected to a VCR, DVD player, or games console, but not an aerial or a cable or satellite TV feed, then you don't need a license.
Slashdot Mirror
Fortunately for the UK, we have an active resistance group that goes by the initials EDS. Their typical modus operandi for thwarting this kind of attack on our freedoms is to bribe or mislead civil servants to be awarded the contract for delivering the system, and then to delay and delay, while pushing the budget up, until another government takes power. The new government then blames the cost overruns on the previous one and cancels the project.
Your phraseology suggests that you have not, in fact read 1984. Don't worry - this is something that you have in common with most people who quote it.
Intel has always had the fab advantage. Previously, they used this to make bigger caches and to ramp up clock speeds. The Pentium 4 was over 2GHz when AMD wasn't shipping anything above 1.5GHz - AMD had to make up the performance gap with more efficient designs. With the Core microarchitecture, they switched their focus to power efficiency. AMD kept focussing on raw performance. This left AMD with nothing competitive in the two biggest-growing markets: low-power servers and laptops.
You're talking a complete cycle earlier. AMD bet on servers at the height of the .com boom. The result was the Opteron. Cut down versions (Athlon64) did well in the desktop. In the next cycle, they bet that they could just keep on delivering more of the same, but the market changed. Server buyers started to care about performance per Watt, because they started being limited by the cooling capacity in their data centres more than their hardware budget. More people bought laptops than desktops, and this let Intel put things that are basically laptop chips (with more cache) and a much bigger heatsink into desktops.
Their branch predictors are the best in the business.
And this isn't even new. I found that some of my code was suffering quite badly from a single unpredicted branch. Rewriting it so that the processor got it right gave about a 20% speedup across the whole program on a 1GHz Athlon. On a 1.2GHz the Pentium M (which uses the P4's branch predictor), it might have seen a 2% speedup, but that was in the error margin for the benchmark.
Otherwise, how else does one explain that their processors are still popular in some supercomputers,
They were popular with supercomputer vendors for one very simple reason: HyperTransport. This made it much easier to plug in a proprietary interconnect for doing NUMA with a thousand or so processors than the northbridge arrangement that Intel used. With QPI, this advantage is gone. Now they are popular because the big supercomputers typically take 2-3 years to build, and the current set were designed before QPI was introduced.
as well as vendors such as Sun?
You know Sun doesn't exist anymore, right?
Not even remotely true. GPUs are stream processors. They are designed to take a big blob of memory, run the same algorithm on every subunit, and create the output as another big blob of memory. They are also designed for code that is very light on branching. They are terrible for things like finite element simulations (which are used in a lot of engineering and physics simulation).
AMD made good products too, they just made them for the wrong market. This is why commercial semiconductor manufacturing is so difficult. You give your engineers a set of constraints and then about 5 years later you have a product. Intel, back around 2003, bet heavily on laptops and power-conscious servers. AMD bet on desktops. Intel's market predictions were better and so the products that they brought to market turned out to be the ones customers wanted by the time the related products made it to market. AMD's were not. Of course, Intel only made this bet after seeing how badly they'd underestimated the importance of power consumption with the Pentium 4 so, if anything, their later products were an overreaction to the poor performance of the P4.
So, in summary, AMDs problem was that they didn't screw up the previous generation, so assumed that the next one could be more of the same and missed the industry shift to mobile devices.
I don't think the problem is that there is too much politics, it's that there is too little technical content.
66 non-whitespace characters is about optimal. Unfortunately, the people who insist on 80-colums as a hard limit tend to be the same people who insist on 8-character tabs for indent, so you end up with a lot fewer than 66 characters per 80-column line.
It's a fine line to pick the exact ratio that's believable. If you're offering 10% above market rate, then people will likely just believe that they got a good deal. If you're offering 100% above the market rate, then they'll think that there's something a bit fishy going on. You need to offer a good enough deal that they thing they're getting the better of you, but not such a good one that they become suspicious. This is where most 419 scammers fail: people are far more likely to believe that they can get $100 due to a bank error than $1,000,000.
Does the script run in a sandboxed environment or an already-installed program (e.g. mathematica, MS Office, cmd.exe)? If so, fine. If not, justify it to your boss. If it's really that useful, then maybe it should be part of the default image...
Unfortunately, melting them down is illegal in the USA, as is exporting them in anything other than very small quantities, so melting them down in another country is difficult too...
In the UK, a lot of shops have charity boxes by the till for putting this kind of change into. Apparently they are quite lucrative.
On the down side, no one will be able to afford to shop in a dollar store anymore...
Buying penny sweets must be awesome there...
Your 'standard Windows corporate lockdown setup' allows end users to run untrusted code that they downloaded from the Internet. I can think of many reasons for calling Google evil, but in this case they are simply doing something that, since Vista, has been a requirement for the 'Designed for MS Windows' logo and part of the recommended practices: allowing non-admin user to install for their own user. It's only 'a nice little unpublicized exploit' if you don't count the articles on MSDN telling you 'this is what you must do in a UAC world'.
It's not Google's fault that you think removing write access to C:\Program Files is the same as preventing users from running their code. Windows has fine-grained ACLs. Learn how to use them. Remove the user's ability to run programs that are installed in any location that they have write access to.
And now I've defended Windows, I need to go and have a shower...
Well, unless they choose to share that data, of course. Google can easily give you some server-side code that just forwards requests made by gStalker.js to their servers, rather than processing it locally. There's also nothing stopping Google from making Google Ads and Analytics users set a subdomain for these to come from.
CM9 is not out yet. The HTC Sense UI is significantly better than the CM7.1 UI, yet still manages to find myriad ways in which to suck.
I've seen it called 'cheese food slices', presumably in the same way that 'juice drink' can mean anything that contains 10% juice and tastes vaguely of fruit.
Moore's law doesn't say what you seem to think it says. It says that the number of transistors that you can fit on a chip for a fixed cost doubles every 18 months (for some value of 18). This may mean that the price of the same chip halves every 18 months (which is roughly true, although if you're buying something like Z80s the cost of the package became the limiting factor some time around the mid '90s). Or it may mean that for the same cost you get twice as many transistors to play with. For a while, it was easy to use these for things like extra integer units, then bringing floating point on-die, then adding SIMD units, and adding extra cache and seeing immediate returns. A few years ago this stopped being the case and the easiest way of getting more speed from more transistors became to add extra cores. Clock frequency is totally independent of Moore's Law.
It's actually a misquote. He said that there 64KB ought to be enough for anyone, and it was a comment on one of the limitations of Microsoft BASIC for 8-bit systems. It's not clear whether it was a serious comment or was intended to be tongue in cheek...
Actually, if you check, the quote is pretty well sourced. However, since the person who said it worked for a company that had sold over a dozen computers that year alone, it is more evidence of total executive ignorance than anything else.
No, actually. I wrote an article about some of the alternatives last year. MRAM, FeRAM, and PCRAM have all been under development by various companies for well over a decade. The only real new discovery is memrister memory. The reason the other three are starting to appear on sites like Slashdot is that they're now getting to the point where they can make the transition to shipping product.
The problem is, having a 50Hz CRT does not require a license, operating a device capable of receiving broadcast TV signals does. If you have a TV that is connected to a VCR, DVD player, or games console, but not an aerial or a cable or satellite TV feed, then you don't need a license.