He's not talking about having the OS ask you if you really wanted to double-click on that icon. "Did you really want to do that" boxes would probably have made the list if he'd made the list a few items longer - they train users to default accept dialog boxes that come up, because they're just in the way and don't actually mean anything. Windows itself does this in numerous places. A dialog box asking if I want to send debug information to Microsoft about an app that just crashed is stupid. Do it, or don't do it, but don't bother users with a dialog box that has no impact on their experience with the OS - it just teaches them to ignore dialogs in the future.
He's talking about asking the user if you really want the trojan that the screensaver installed to start when you boot your computer next. If the computer properly protected the dozen or so apps you use and asked every time something new ran, people might actually start reading dialog boxes because most of them would contain useful information that will directly impact the user's experience.
The correct approach is not to try to algorithmically determine what is bad and warn the user about it, as you suggest. The correct approach is to algorithmically detect strangeness, deviations from the norm, that may indicate something wierd is happening. And then ask the user about it - even the dumbest user is better at sorting out new information than the computer.
The topic / search icons at the top of the page (not the ones associated with individual stories) are being printed in an unordered list with bullets at the top of the page. I see no reason why these should print. Using the display property, some print-version-specific intro could be printed to introduce the page, but those icons are just a waste of space (about half a page at present).
A quick glance through print.css indicates this is the intended behavior, though I may have missed something.
They've done this sort of thing before. Often. To big and small. And yet, they're still raking in the dough. A quick search for "paypal sucks" brings up a quarter million or so pissed-off ex-paypal users.
As always, don't keep money at paypal that you can't afford to lose.
Oh, I deliberately shied away from the whole mess that current DST is. Half of Indiana believing in it, half not - what's up with that?
My ideal solution would be to use UTC. nyrk is right that there's too much inertia about arbitrary numbers on a clock face, and that's just considering this one country. When you start talking about even larger distances, time zones and all the myriad variations thereon just get messier and messier.
Changing to UTC adds no additional complexity to any calculation involving time zones, and simplifies very many situations indeed.
I understand the theory that by changing people's work habits relative to the solar day, we may be saving some energy. I think the reality of it is rather less impressive than the theory, and certainly doesn't justify the expense and hassle, but it's undoubtedly there.
Too bad changing DST is the stupidest way to do it. As has been mentioned before, DST impacts way more than just work schedules. It's buried deep in various applications where it doesn't belong. It's hard-coded into embedded systems where it can't be changed. It's stuck on old software installations that will never see an upgrade. Changing DST is bad enough, but a half-assed mix of new-DST machines and old-DST machines is just a recipe for disaster.
If the government really wants to save energy by changing work habits, there are enormously better ways to do it. Tax credits for corporations that stagger their workers' start times by a significant margin would save way more energy than this DST nonsense, and it wouldn't have the unpleasant ancillary effects that changing the definition of time of day would have. Unimaginably large (you can look for the true numbers as well as I) amounts of gasoline are wasted in rush-hour traffic across the nation. Tax credits for starting 1/3 of employees 2hr earlier than normal and 1/3 2hr later would motivate employers to do it, and reducing the time people spend idling their cars on the freeways, or worse, driving in stop-and-go traffic, would save tons of fuel.
Tax credits (or some other incentive) makes people happy because the government's not forcing anyone to do anything they don't want to. It would have very few unintended consequences. And it would save many times more energy.
The reasoning that leads to a change in DST is just tortured. The government wants people going to work at a different time. So rather than ask (bribe, punish/bribe, whatever) businesses to employ people at different hours, they change the meaning of 8am, and screw up the entire country. Where's the logic in that?!
"In the wake of 9/11..." is all too often used to start a paragraph which boils down to "...government spent a ton of money doing absolutely nothing for security while simultaneously doing far more harm to a free and democratic way of life than the terrorists ever could."
I'm pleased to see that at least in one case, someone actually gets it. You can't stop the terrorists. Random bad stuff happens in life, and the best thing for it is to be as prepared as you can. Communications is the single best way to spend money in preparation for insert-bad-stuff-here. Be it terrorism, natural disaster, industrial accident, or what have you, better communications saves lives. As has been said many times by everyone whose job doesn't involve spending billions of dollars, more money should not be spent on trying to prevent disasters, it should be spent on ways to clean up after them. Billions in airport screening is a complete waste of money because it just forces someone to bomb a mall or movie theater instead - billions in police training, EMT and EMS training, hospitals and clinics, etc. is money that will reap rewards no matter what happens next.
The venice core is the new one. It's a few unimportant percent faster than the others. It's got a few not-terribly-important new features. It's rather more than a few percent less power-hungry. AMD doggedly sticks to its 89W max figure, but Venice maxes out at more like 50W, according to measurements / approximations.
So, to recap:
Faster Better Lower power Nicely overclockable
Buy from a respectable retailer and they'll tell you not only which core you're buying, but the clockspeed, cache size, etc. rather than just the model number.
The computers are not beating the humans with math. They are not relying on superior computational capability. The computers are winning with superior algorithms. Even a fairly shallow complete traversal of the search space is many orders of magnitude away from being possible, and a machine using this approach will be consistently beaten by even middling players. Computer chess has advanced primarily due to algorithmic optimizations. The evaluation functions that a modern chess engine uses are extremely well-tuned, and while a chess engine may be backed by an enormous pre-computed opening book, this too is dependent on algorithmic advances, because the book is calculated using algorithms as well, not a brute-force search. The two sets of algorithms are different, and the opening book can benefit from hugely greater computational resources, but ultimately the search algorithm is the limiting factor.
In short, the recent successes of machine chess are due to human enginuity, to the same sort of creative processes that humans themselves use to play chess. Technology, in the machine sense, is almost irrelevant (see Fritz's victories on a dinky 8P Xeon with a few gig of RAM) when compared to the advances in understanding of the game of chess.
Interestingly, even as the programmers are developing an ever-greater understanding of chess, chess players are developing an ever-greater understanding of both the game and the way in which computers play it, though people with much greater understanding of this than I tell me that the newest algorithms are playing a very human-like game, minimizing the effect of understanding 'computer chess' on the game.
I'm having trouble making an informed buying decision here. Perhaps the collective/. wisdom could assist?
On the one hand, this mouse's $70 price tag (quite reasonable for a quality mouse, I think) is 'whopping'. On the other hand, 90-odd teraflops worth of supercomputer is 'whopping'. I really only have room for one whopping piece of consumer electronics in my home, and I'm torn between the two. Perhaps we could petition the manufacturer to have the mouse's designation changed? Or could we instead express the supercomputer's processing power in terms of the number of LoCs per hour it could perform a simple regexp on? Would that still be a whopping number? Perhaps the only hope for it is to buy the mouse with euros?
"Only buy a hybrid if you want to feel good about yourself."...or reduce your environmental impact on air and water quality....or reduce your country's dependence on oil....or spur development of ever-more-efficient vehicles.
There are more reasons to buy a fuel-efficient vehicle than saving money.
No, it will not work. Each CPU supports up to 4 DIMM slots - to use all 16, you will need 4 CPUs.
The Tyan board you want is the S4882, as mentioned by at least one other poster. Available for about $1500 from such places as Alvio: http://www.alvio.com/pmoreinfo.asp?iid=3256
2GB DIMMs have come down in price - they're now only about 3x the price of 1GB DIMMs, or 50% more expensive per MB. Using 2GB DIMMs rather than twice as many 1GB DIMMs is probably cost-effective, assuming you don't mind limiting your upgradeability. Otherwise, a quad board with 2 CPUs and 2GB DIMMs would leave quite a bit of room. That Tyan board supports up to 32GB of DDR400.
If the charging circuit (including battery) is 85% efficient during charging, you'd need 1555/0.85 = 1830W, and your charger / battery would be pumping out 275W of heat.
If you assume 90% efficiency, you're only talking about 173W of heat. At 95% efficiency, a mere 82W.
Then comes the power supply for this. Wall warts are in the 50% efficiency range; computer PSUs top out at about 90%. Even supposing you could make a special-purpose device 95% efficient, and the battery / charger is also 95% efficient, you'll be burning over 80W in the power supply and another 80W in the battery and charger. The total power losses are great - you're at 95% efficiency after all - but the heat to be dissipated would very quickly become problematic. Think of running a high-end Opteron or mid-range P4 for 5 minutes. You'd need a pretty beefy heatsink, and a fan too.
This doesn't mean it can't be done, of course. Some applications can handle that heat load, and one presumes the batteries can also be charged in 20 minutes to generate a quarter the waste power. It seems unlikely, but if it really works, I'll take 5.
Re:Info on what exactly SHA-1 is ...
on
SHA-1 Broken
·
· Score: 2, Interesting
Sure, there's a chance, but it's fairly remote.
The original SHA was released in 1993. 2 years later, SHA-1 was released (by the NSA, through NIST) to address unspecified weakness in the original algorithm. In 1998, a weakness in the original SHA was discovered, and in 2004 various attacks on SHA, SHA-1, and other hashes were published. It is unknown which, if any, of the known SHA weaknesses were discovered by the NSA and resulted in the 1995 release of SHA-1, though at least one known attack affects SHA but not SHA-1. It seems likely that if the NSA were capable of releasing SHA-1 as the original SHA in 1993, they would have done so. Thus, it appears that the NSA discovered a weakness in SHA in 2 years that took the academic community 5-10 years to discover. The gap is not as wide as it was when DES was released, but it's still very impressive.
End brief synthesis. Check out the wikipedia article (and links) for more information. That's where I got mine.
From the article: "I have seen some feedback that we should not have declared the doctype as XHTML Strict. If anything, we are closer to HTML 4.01. I agree. But our target is to get to XHTML strict."
And what is a doctype? That's correct: the type of the document. So if the document is really more like HTML 4.01, it should be labeled HTML 4.01. It should not be labeled with some marketing 'we'd really like this to work' drivel. Sort of a micrososm of the Microsoft Approach, actually. Lying to people fits under marketing - we're used to it and can take it into account. Lying to computers, computers which are trying to make your not-really-HTML into something presentable, is just stupid.
If you pervert Moore's law into a statement of speed, you end up coming out ahead for any computation that
1) is CPU-bound rather than interconnect-bound or disk-bound or memory-bound 2) will take 3 years+ with current technology / budget, and 3) produces no useful intermediate results
At 3 years, you come out even buying current tech and running it for 3 years versus waiting 18 months and buying spending the same money on tech that can do the job in 18 months.
There are few such computations. Note that the universe simulation being discussed here does not qualify, even if it were run for the requisite 3 years. Clearly interconnect latency/bandwidth was a significant concern, necessitating special high-speed components and a torus topography.
On the other hand, I can buy quality 1GB DIMMs for $250. Divide by 4 (rough guess. SRAM at 6T should be 6x the price, but DIMMs have caps too. 4x the manufacturing costs seems reasonable, assuming the infrastructure were in place), and you've got 256M SRAM modules for $250. Obviously that's a bit on the spendy side for large capacity RAM, but clearly there's a market for faster DIMMs. Unfortunately, DRAM access time, at about 5ns, isn't the major component of memory latency, which even on the best systems runs 10x that. The market won't bear 4x the price for a 10% increase in speed.
This means that for SRAM to be useful, it has to be paired with a lower-latency interconnect. Some apps would benefit tremendously from 128M of what would amount to an L3 cache, even to the point that the $400 or so extra it would cost might be worth it. It's clear however that the market doesn't consider that a worthwhile expenditure.
Although newer system architectures such as AMD's Opteron platform are moving to more closely-attached RAM, the engineering and manufacturing challenges involved in attaching memory as tightly as it is to a GPU have so far proven more expensive than the payoffs warrant. With improvements in manufacturing and interconnect technology, I'm sure we'll see ever-tighter CPU-memory integration. I doubt however the technology will move to SRAM or an SRAM-equivalent simply because the performance/heat trade-off isn't favorable. Saving a few ns of latency on the memory chips is peanuts compared to the 10s of ns of latency in the connection to the CPU, which is probably a much more tractable problem.
The whitepaper suggests they're tearing out the run-length encoding that's the final step in jpeg and replacing it with something more space-efficient.
In a nutshell, JPEG works like:
Original image data -> frequency domain -> high frequencies truncated (via division matrix) -> RLE
RLE is fast, but not terribly compact. Replacing it with something better improves compression. However, RLE generates not-very-compressable output, which is why traditional compression software does poorly. I imagine if you took a jpeg, undid the RLE, and zip compressed the result, you'd get something close to what the stuffit folk are claiming. If someone wants to try that, I'd be interested in the results.
The memory modules you put in your computer are composed of DRAM chips. DRAM uses a capacitor and a transistor per cell (plus sense amps, decoders, etc.). DRAM requires refresh (the charge on the cap leaks off) but is relatively low-power and very dense. SRAM uses no capacitors, but more transistors (4 or 6) per bit; it's higher-power but faster and doesn't require refresh.
So, SRAM density has nothing to do with DIMMs you put in a computer. It's used for on-chip caches (and off-chip caches), but is too expensive for main memory. Denser SRAM means that Opteron you've got with 1M L2 cache could have 4M or 8M if IBM can mass-produce the stuff.
DIMMs usually have 16 chips (18 for ECC modules). So, if you have 512Mbit DRAMs, you put 16 of them on a module and you get 8Gbit = 1Gbyte. Gigabit DRAMs can make 2GB DIMMs. 2Gbit DRAMs are needed to make 4GB DIMMs; they cost hundreds of dollars each (and you need 16!), which is why 4GB DIMMs are so amazingly expensive.
Q1 this year, IBM sold 2 Itanium boxes. Yup, 2. Up to 200 by second quarter. IBM sells that many Power machines in about 6 hours. Dell shipped fewer than IBM; Bull shipped 80 but made more money than Dell ($5m versus $4m) Only HP, with about 80% of the ia64 market, shipped more Itaniums than Opterons. Even with Opterons selling for roughly 10% the cost of Itaniums, ia64 barely beat out Opteron in profits, and hasn't a prayer in Q3 and Q4.
The ENTIRE ia64 market for the first half of 2004 was a pissant $600m. 80% of which went to HP.
So, while you're correct that Siemens, Bull, IBM, Dell, etc. are major vendors, they're not major Itanium vendors, and wouldn't suffer a whit if Itanium died.
You can forgive them now . . . for the DIVX debacle. You shouldn't shop there because of the rude, slow staff, big store and small selection, and the way prices strangely go up about 10% between their website and the store.
That hasn't worked in the past. AMD's design talent is better spent on CPUs, and the only reason they used to design chipsets was because nobody else would. Back when AMD was making pin-compatible chips, they didn't need chipsets since they used whatever Intel was making. Then along came the Athlon, and AMD needed a chipset. So they made one, and then stopped as soon as Via and SiS started making their (better; cheaper) own. Then came the Athlon MP, and AMD again released chipsets, first the 760MP and then the 760MPX. Both fairly low-performance, low-feature chipsets designed to get some market adoption and convince the real chipset designers it was worth their time.
These days, everybody knows that 1) Athlons rock, and 2) they'll sell. If AMD now produces their own chipsets, they're just competing with their current chipset partners, and diverting talent from CPU design (and flash, and the other things AMD does). Between ATi, NV, SiS, and Via, there's plenty of competition for the Athlon segment and with Intel's recent return to high-quality chipset manufacture, there's competition to produce the best platform. AMD's held up their end of the bargain, and shortly the chipset makers will catch up to Intel as well.
It'd be great if Dell started selling AMD kit. However, that's not going to happen any time soon. Firstly, Dell's getting huge discounts on CPUs and chipsets from Intel. Secondly, even if AMD were to match Intel on price, AMD doesn't have the fab space to keep up. Fab 30 is going all-out, and AMD's in the process of building another one; also working closely with IBM, etc. in an effort to increase that capacity as cheaply and quickly as they can to capitalize on their superior product. AMD's mid-30% market share corresponds to their maximum output. Dell might be able to bump that up to 50%+, but only if AMD brings another fab online. However, AMD has to be careful about investing too heavily in fab capacity they can't use, so they'll only bring online capacity to handle Dell signing on, if Dell agrees to it beforehand which Dell probably won't do, not knowing if AMD really can live up to their end of the bargain later. Kind of a catch-22, but AMD's doing pretty well these days anyway, and the consumer can't really complain either, so it's all OK.
Slashdot Mirror
He's not talking about having the OS ask you if you really wanted to double-click on that icon. "Did you really want to do that" boxes would probably have made the list if he'd made the list a few items longer - they train users to default accept dialog boxes that come up, because they're just in the way and don't actually mean anything. Windows itself does this in numerous places. A dialog box asking if I want to send debug information to Microsoft about an app that just crashed is stupid. Do it, or don't do it, but don't bother users with a dialog box that has no impact on their experience with the OS - it just teaches them to ignore dialogs in the future.
He's talking about asking the user if you really want the trojan that the screensaver installed to start when you boot your computer next. If the computer properly protected the dozen or so apps you use and asked every time something new ran, people might actually start reading dialog boxes because most of them would contain useful information that will directly impact the user's experience.
The correct approach is not to try to algorithmically determine what is bad and warn the user about it, as you suggest. The correct approach is to algorithmically detect strangeness, deviations from the norm, that may indicate something wierd is happening. And then ask the user about it - even the dumbest user is better at sorting out new information than the computer.
Firefox 1.0.6 / Windows XP
Slashcode.com. Print preview.
The topic / search icons at the top of the page (not the ones associated with individual stories) are being printed in an unordered list with bullets at the top of the page. I see no reason why these should print. Using the display property, some print-version-specific intro could be printed to introduce the page, but those icons are just a waste of space (about half a page at present).
A quick glance through print.css indicates this is the intended behavior, though I may have missed something.
They've done this sort of thing before. Often. To big and small. And yet, they're still raking in the dough. A quick search for "paypal sucks" brings up a quarter million or so pissed-off ex-paypal users.
As always, don't keep money at paypal that you can't afford to lose.
The comic is at piratejesus.com. It's drawn by a friend of mine, so go check it out. Full of delightfully geeky puns.
Oh, I deliberately shied away from the whole mess that current DST is. Half of Indiana believing in it, half not - what's up with that?
My ideal solution would be to use UTC. nyrk is right that there's too much inertia about arbitrary numbers on a clock face, and that's just considering this one country. When you start talking about even larger distances, time zones and all the myriad variations thereon just get messier and messier.
Changing to UTC adds no additional complexity to any calculation involving time zones, and simplifies very many situations indeed.
What an idiotic idea!
I understand the theory that by changing people's work habits relative to the solar day, we may be saving some energy. I think the reality of it is rather less impressive than the theory, and certainly doesn't justify the expense and hassle, but it's undoubtedly there.
Too bad changing DST is the stupidest way to do it. As has been mentioned before, DST impacts way more than just work schedules. It's buried deep in various applications where it doesn't belong. It's hard-coded into embedded systems where it can't be changed. It's stuck on old software installations that will never see an upgrade. Changing DST is bad enough, but a half-assed mix of new-DST machines and old-DST machines is just a recipe for disaster.
If the government really wants to save energy by changing work habits, there are enormously better ways to do it. Tax credits for corporations that stagger their workers' start times by a significant margin would save way more energy than this DST nonsense, and it wouldn't have the unpleasant ancillary effects that changing the definition of time of day would have. Unimaginably large (you can look for the true numbers as well as I) amounts of gasoline are wasted in rush-hour traffic across the nation. Tax credits for starting 1/3 of employees 2hr earlier than normal and 1/3 2hr later would motivate employers to do it, and reducing the time people spend idling their cars on the freeways, or worse, driving in stop-and-go traffic, would save tons of fuel.
Tax credits (or some other incentive) makes people happy because the government's not forcing anyone to do anything they don't want to. It would have very few unintended consequences. And it would save many times more energy.
The reasoning that leads to a change in DST is just tortured. The government wants people going to work at a different time. So rather than ask (bribe, punish/bribe, whatever) businesses to employ people at different hours, they change the meaning of 8am, and screw up the entire country. Where's the logic in that?!
Hot grits and all. We can take it. Let's hear from the old-school trolls.
Come on, now's your chance. It's on-topic, even!
I want to wake up this morning with 50 hot grits replies to this comment. I know you can do it, guys.
Bring it on!
"In the wake of 9/11..." is all too often used to start a paragraph which boils down to "...government spent a ton of money doing absolutely nothing for security while simultaneously doing far more harm to a free and democratic way of life than the terrorists ever could."
I'm pleased to see that at least in one case, someone actually gets it. You can't stop the terrorists. Random bad stuff happens in life, and the best thing for it is to be as prepared as you can. Communications is the single best way to spend money in preparation for insert-bad-stuff-here. Be it terrorism, natural disaster, industrial accident, or what have you, better communications saves lives. As has been said many times by everyone whose job doesn't involve spending billions of dollars, more money should not be spent on trying to prevent disasters, it should be spent on ways to clean up after them. Billions in airport screening is a complete waste of money because it just forces someone to bomb a mall or movie theater instead - billions in police training, EMT and EMS training, hospitals and clinics, etc. is money that will reap rewards no matter what happens next.
Venice.
The venice core is the new one. It's a few unimportant percent faster than the others. It's got a few not-terribly-important new features. It's rather more than a few percent less power-hungry. AMD doggedly sticks to its 89W max figure, but Venice maxes out at more like 50W, according to measurements / approximations.
So, to recap:
Faster
Better
Lower power
Nicely overclockable
Buy from a respectable retailer and they'll tell you not only which core you're buying, but the clockspeed, cache size, etc. rather than just the model number.
The computers are not beating the humans with math. They are not relying on superior computational capability. The computers are winning with superior algorithms. Even a fairly shallow complete traversal of the search space is many orders of magnitude away from being possible, and a machine using this approach will be consistently beaten by even middling players. Computer chess has advanced primarily due to algorithmic optimizations. The evaluation functions that a modern chess engine uses are extremely well-tuned, and while a chess engine may be backed by an enormous pre-computed opening book, this too is dependent on algorithmic advances, because the book is calculated using algorithms as well, not a brute-force search. The two sets of algorithms are different, and the opening book can benefit from hugely greater computational resources, but ultimately the search algorithm is the limiting factor.
In short, the recent successes of machine chess are due to human enginuity, to the same sort of creative processes that humans themselves use to play chess. Technology, in the machine sense, is almost irrelevant (see Fritz's victories on a dinky 8P Xeon with a few gig of RAM) when compared to the advances in understanding of the game of chess.
Interestingly, even as the programmers are developing an ever-greater understanding of chess, chess players are developing an ever-greater understanding of both the game and the way in which computers play it, though people with much greater understanding of this than I tell me that the newest algorithms are playing a very human-like game, minimizing the effect of understanding 'computer chess' on the game.
I'm having trouble making an informed buying decision here. Perhaps the collective /. wisdom could assist?
On the one hand, this mouse's $70 price tag (quite reasonable for a quality mouse, I think) is 'whopping'. On the other hand, 90-odd teraflops worth of supercomputer is 'whopping'. I really only have room for one whopping piece of consumer electronics in my home, and I'm torn between the two. Perhaps we could petition the manufacturer to have the mouse's designation changed? Or could we instead express the supercomputer's processing power in terms of the number of LoCs per hour it could perform a simple regexp on? Would that still be a whopping number? Perhaps the only hope for it is to buy the mouse with euros?
"Only buy a hybrid if you want to feel good about yourself." ...or reduce your environmental impact on air and water quality. ...or reduce your country's dependence on oil. ...or spur development of ever-more-efficient vehicles.
There are more reasons to buy a fuel-efficient vehicle than saving money.
No, it will not work. Each CPU supports up to 4 DIMM slots - to use all 16, you will need 4 CPUs.
The Tyan board you want is the S4882, as mentioned by at least one other poster. Available for about $1500 from such places as Alvio: http://www.alvio.com/pmoreinfo.asp?iid=3256
2GB DIMMs have come down in price - they're now only about 3x the price of 1GB DIMMs, or 50% more expensive per MB. Using 2GB DIMMs rather than twice as many 1GB DIMMs is probably cost-effective, assuming you don't mind limiting your upgradeability. Otherwise, a quad board with 2 CPUs and 2GB DIMMs would leave quite a bit of room. That Tyan board supports up to 32GB of DDR400.
That's assuming 100% efficiency.
If the charging circuit (including battery) is 85% efficient during charging, you'd need 1555/0.85 = 1830W, and your charger / battery would be pumping out 275W of heat.
If you assume 90% efficiency, you're only talking about 173W of heat. At 95% efficiency, a mere 82W.
Then comes the power supply for this. Wall warts are in the 50% efficiency range; computer PSUs top out at about 90%. Even supposing you could make a special-purpose device 95% efficient, and the battery / charger is also 95% efficient, you'll be burning over 80W in the power supply and another 80W in the battery and charger. The total power losses are great - you're at 95% efficiency after all - but the heat to be dissipated would very quickly become problematic. Think of running a high-end Opteron or mid-range P4 for 5 minutes. You'd need a pretty beefy heatsink, and a fan too.
This doesn't mean it can't be done, of course. Some applications can handle that heat load, and one presumes the batteries can also be charged in 20 minutes to generate a quarter the waste power. It seems unlikely, but if it really works, I'll take 5.
Sure, there's a chance, but it's fairly remote.
The original SHA was released in 1993. 2 years later, SHA-1 was released (by the NSA, through NIST) to address unspecified weakness in the original algorithm. In 1998, a weakness in the original SHA was discovered, and in 2004 various attacks on SHA, SHA-1, and other hashes were published. It is unknown which, if any, of the known SHA weaknesses were discovered by the NSA and resulted in the 1995 release of SHA-1, though at least one known attack affects SHA but not SHA-1. It seems likely that if the NSA were capable of releasing SHA-1 as the original SHA in 1993, they would have done so. Thus, it appears that the NSA discovered a weakness in SHA in 2 years that took the academic community 5-10 years to discover. The gap is not as wide as it was when DES was released, but it's still very impressive.
End brief synthesis. Check out the wikipedia article (and links) for more information. That's where I got mine.
"Would you be happy buying a 12-pack at the corner store, but having to sacrifice one can/bottle to some guy at the exit door for no apparent reason?"
If the alternative is spending an extra $0.10 a can on beer that tastes funny, I'll toss the bouncer a bottle every trip.
From the article: "I have seen some feedback that we should not have declared the doctype as XHTML Strict. If anything, we are closer to HTML 4.01. I agree. But our target is to get to XHTML strict."
And what is a doctype? That's correct: the type of the document. So if the document is really more like HTML 4.01, it should be labeled HTML 4.01. It should not be labeled with some marketing 'we'd really like this to work' drivel. Sort of a micrososm of the Microsoft Approach, actually. Lying to people fits under marketing - we're used to it and can take it into account. Lying to computers, computers which are trying to make your not-really-HTML into something presentable, is just stupid.
If you pervert Moore's law into a statement of speed, you end up coming out ahead for any computation that
1) is CPU-bound rather than interconnect-bound or disk-bound or memory-bound
2) will take 3 years+ with current technology / budget, and
3) produces no useful intermediate results
At 3 years, you come out even buying current tech and running it for 3 years versus waiting 18 months and buying spending the same money on tech that can do the job in 18 months.
There are few such computations. Note that the universe simulation being discussed here does not qualify, even if it were run for the requisite 3 years. Clearly interconnect latency/bandwidth was a significant concern, necessitating special high-speed components and a torus topography.
On the other hand, I can buy quality 1GB DIMMs for $250. Divide by 4 (rough guess. SRAM at 6T should be 6x the price, but DIMMs have caps too. 4x the manufacturing costs seems reasonable, assuming the infrastructure were in place), and you've got 256M SRAM modules for $250. Obviously that's a bit on the spendy side for large capacity RAM, but clearly there's a market for faster DIMMs. Unfortunately, DRAM access time, at about 5ns, isn't the major component of memory latency, which even on the best systems runs 10x that. The market won't bear 4x the price for a 10% increase in speed.
This means that for SRAM to be useful, it has to be paired with a lower-latency interconnect. Some apps would benefit tremendously from 128M of what would amount to an L3 cache, even to the point that the $400 or so extra it would cost might be worth it. It's clear however that the market doesn't consider that a worthwhile expenditure.
Although newer system architectures such as AMD's Opteron platform are moving to more closely-attached RAM, the engineering and manufacturing challenges involved in attaching memory as tightly as it is to a GPU have so far proven more expensive than the payoffs warrant. With improvements in manufacturing and interconnect technology, I'm sure we'll see ever-tighter CPU-memory integration. I doubt however the technology will move to SRAM or an SRAM-equivalent simply because the performance/heat trade-off isn't favorable. Saving a few ns of latency on the memory chips is peanuts compared to the 10s of ns of latency in the connection to the CPU, which is probably a much more tractable problem.
The whitepaper suggests they're tearing out the run-length encoding that's the final step in jpeg and replacing it with something more space-efficient.
In a nutshell, JPEG works like:
Original image data -> frequency domain -> high frequencies truncated (via division matrix) -> RLE
RLE is fast, but not terribly compact. Replacing it with something better improves compression. However, RLE generates not-very-compressable output, which is why traditional compression software does poorly. I imagine if you took a jpeg, undid the RLE, and zip compressed the result, you'd get something close to what the stuffit folk are claiming. If someone wants to try that, I'd be interested in the results.
The memory modules you put in your computer are composed of DRAM chips. DRAM uses a capacitor and a transistor per cell (plus sense amps, decoders, etc.). DRAM requires refresh (the charge on the cap leaks off) but is relatively low-power and very dense. SRAM uses no capacitors, but more transistors (4 or 6) per bit; it's higher-power but faster and doesn't require refresh.
So, SRAM density has nothing to do with DIMMs you put in a computer. It's used for on-chip caches (and off-chip caches), but is too expensive for main memory. Denser SRAM means that Opteron you've got with 1M L2 cache could have 4M or 8M if IBM can mass-produce the stuff.
DIMMs usually have 16 chips (18 for ECC modules). So, if you have 512Mbit DRAMs, you put 16 of them on a module and you get 8Gbit = 1Gbyte. Gigabit DRAMs can make 2GB DIMMs. 2Gbit DRAMs are needed to make 4GB DIMMs; they cost hundreds of dollars each (and you need 16!), which is why 4GB DIMMs are so amazingly expensive.
That'd be curricula.
u m
http://dictionary.reference.com/search?q=curricul
Dell. IBM. Right.
Q1 this year, IBM sold 2 Itanium boxes. Yup, 2. Up to 200 by second quarter. IBM sells that many Power machines in about 6 hours. Dell shipped fewer than IBM; Bull shipped 80 but made more money than Dell ($5m versus $4m) Only HP, with about 80% of the ia64 market, shipped more Itaniums than Opterons. Even with Opterons selling for roughly 10% the cost of Itaniums, ia64 barely beat out Opteron in profits, and hasn't a prayer in Q3 and Q4.
The ENTIRE ia64 market for the first half of 2004 was a pissant $600m. 80% of which went to HP.
So, while you're correct that Siemens, Bull, IBM, Dell, etc. are major vendors, they're not major Itanium vendors, and wouldn't suffer a whit if Itanium died.
You can forgive them now . . . for the DIVX debacle. You shouldn't shop there because of the rude, slow staff, big store and small selection, and the way prices strangely go up about 10% between their website and the store.
That hasn't worked in the past. AMD's design talent is better spent on CPUs, and the only reason they used to design chipsets was because nobody else would. Back when AMD was making pin-compatible chips, they didn't need chipsets since they used whatever Intel was making. Then along came the Athlon, and AMD needed a chipset. So they made one, and then stopped as soon as Via and SiS started making their (better; cheaper) own. Then came the Athlon MP, and AMD again released chipsets, first the 760MP and then the 760MPX. Both fairly low-performance, low-feature chipsets designed to get some market adoption and convince the real chipset designers it was worth their time.
These days, everybody knows that 1) Athlons rock, and 2) they'll sell. If AMD now produces their own chipsets, they're just competing with their current chipset partners, and diverting talent from CPU design (and flash, and the other things AMD does). Between ATi, NV, SiS, and Via, there's plenty of competition for the Athlon segment and with Intel's recent return to high-quality chipset manufacture, there's competition to produce the best platform. AMD's held up their end of the bargain, and shortly the chipset makers will catch up to Intel as well.
It'd be great if Dell started selling AMD kit. However, that's not going to happen any time soon. Firstly, Dell's getting huge discounts on CPUs and chipsets from Intel. Secondly, even if AMD were to match Intel on price, AMD doesn't have the fab space to keep up. Fab 30 is going all-out, and AMD's in the process of building another one; also working closely with IBM, etc. in an effort to increase that capacity as cheaply and quickly as they can to capitalize on their superior product. AMD's mid-30% market share corresponds to their maximum output. Dell might be able to bump that up to 50%+, but only if AMD brings another fab online. However, AMD has to be careful about investing too heavily in fab capacity they can't use, so they'll only bring online capacity to handle Dell signing on, if Dell agrees to it beforehand which Dell probably won't do, not knowing if AMD really can live up to their end of the bargain later. Kind of a catch-22, but AMD's doing pretty well these days anyway, and the consumer can't really complain either, so it's all OK.