Most days I don't refer to Zipf's law. Most days I don't need to. Some days I dig a little deeper in the dictionary because the stupidity I'm confronting is less ordinary.
Easy. 256MB configurations will quickly go the way of the Dodo bird. Retail competes on sticker price for the cheapest thing on the shelf. Some morons come along and buy that cheapest thing, the less moronic allow themselves to be "up sold" into something less incapacitated, while the super moronic hang around to get "up sold" to the highest margin piece of crap displayed on the shelves for exactly that purpose (anyone here like to part with $2k? I've got some *really* **awesome** 24 gauge zipcord looking for a good home).
Just imagine when you go across with the street with your 256MB price check and the oversexed 22 year old slick working there starts giving you the hairy eyeball about "Vista compatible".
Haven't you ever heard the retail lingo "oh, those guys, we get a lot of people in here after dealing with those guys"? That's the sound of retailers driving their own (who don't fall in line) into extinction.
Any store continuing to sell 256MB configurations in the Vista epoch is going to be portrayed by every slick-haired commissioned sales droid within a five mile radius as the fat kid with the black hairs growing out of his pimple.
Right after posting that screed, I looked on my desk at the book open there and read this quotation:
'Defendit numerus', [there is safety in numbers] is the maxim of the foolish; 'Deperdit numerus', [there is ruin in numbers] of the wise.
I'm not sure if he's referring to bones, boners, or boneheads. Bonus marks for anyone who can identify the book I'm reading. You have a long time to answer, it's printed on paper with a life expectancy of several hundred years.
Let's suppose some very wealthy snot picker decides to buy more iPods than the total number Apple has already sold and fill all of these with material of his own copyright which he puts into the public domain. Now the majority of music stored on iPods is user owned. Does the iPod collective sense their collective toppling into the legal side of majorism, and suddenly for everyone all file sharing becomes legal?
I find these majoristic arguments legally amusing for failing to isolate cause and effect. The iPod I bought behaves in certain ways solely because of presumptions about what other iPod purchasers are doing, at some select point in time.
Well, you could argue that because some RIAA controlled music exists, it becomes possible that a majority of iPod users have loaded mostly RIAA controlled tracks, and therefor sharing features should be disabled on the prospect that its use might be more illegal than legal, counting instance against instance.
But isn't that argument strange too? The main reason this argument holds sway is because RIAA is a functional monopoly. RIAA remains a functional monopoly in part because devices such as the iPod make it inconvenient, even in the legal case, to work around the RIAA monopoly.
Let's try another syllogism. Music is cool. RIAA has established themselves as the music monopoly. Apple is cool. Cool is a RIAA monopoly. Therefore Apple implements DRM.
It's no worse than your circular majorism, and about the level of the average person who drools over these devices. Let's face it. Cool has another property: it shuts off the brain. If you're cool, you're getting action, and if you're getting action, the brain is vestigial, and then what does it matter that RIAA is getting their cut? So is Pfizer, Carter-Wallace, Ortho McNeil, and most of the time, sooner or later, Proctor and Gamble.
I agree entirely. Only a truly Vogonesque mind would even think to pose such a question in those terms. "Ah, but if we could make it more stupid, why wouldn't we?" While we're at it, after we've cut off Britian and Australia and Taiwan, why don't we make it impossible to phone those places as well? And really, passports offices are for dodos, anyway. How many times have you been to Santiago this year? Oh come on, no one goes to Santiago twice in one year.
If eighty cents of every dollar I spend supporting OpenSSH gets flushed down the OpenBSD toilet, is that a good use of my contribution?
The cluelessness of this post defies belief.
I want to support this OpenFoil airplane wing because it supports me. However, if eighty cents of every dollar I spend supporting OpenFoil is vented through the OpenBlow high-test wind tunnel, is that a good use of my contributions?
NX protection, Pro-police, and priv-sep are all products of the two efforts coordinated together. Almost every dime OpenBSD spends is spent in the pursuit of enhancing security, and it's to imagine that those results are not immediately folded back into OpenSSH. Unlike FreeBSD, OpenBSD spends shockingly little on the OS itself. They aren't busy inventing disk geometry managers or porting to 150 different platforms.
90% of human stupidity originates in the capacity of the human mind to engage in intellectual shell games. Here is this dollar: let's split it up in to the 80 cents wasted on OpenBSD and the 20 cents invested in OpenSSH.
Or, my brother is dying of Leukemia. I want to donate blood because blood keeps him alive. Is that a good investment if 80% of the blood I donate is flushed down the toilet to replace blood lost during bone marrow transplants?
The guy who had the idea that all the consultants who install OpenBSD or OpenSSH for their clients should bill them for the official OpenBSD install set, so they can stack the shiny disks right beside the eight volumes of Vista they don't yet have was halfway to a good idea. [OK, I paraphrased.]
OpenBSD should release a special CD entitled "OpenBSD Recovery CD" for exactly this purpose. Fair game. You could use this CD to recover your system if you wanted to. Or, for a one time $50K payment, you can purchase the Enterprise-grade "OpenBSD Rescue CD" with exactly the same content.
I think that OpenBSD has not picked up the corporate support from big corporations who rely on OpenBSD such as HP et al. because the big corporations prefer to align themselves with feel good stories such as Barry Bonds, rather than difficult people like Theo.
It almost never fails when assessing claims for a new technology to look at the list of proposed applications. You get a pretty darn good "humbug" feeling long before you dig into the technical details, which are invariably thin on the ground. The other fabulous telltale is "commercial applications in five years". "Five years" is venture capital speak for "we have no clue". "Ten years" is primary science speak for "I've been cited three times already".
99 inventions out of a hundred that promise the moon deliver green cheese.
Even talking to my father's generation, the general level of ignorance that prevailed concerning marital relations likely did more harm than skin mags. Where was this idea originally hatched, that skin mags are harmful to minors? I've never had an adult person confess to me to having been harmed in his / her teenage years by coming across the wrong kind of printed material under their parent's bed. One could argue that children are harmed by exposure to emotionally charged materials in the *absense* of appropriate parental guidance, yet apparently many/most parents find it difficult to offer this guidance with destroying some cherished illusion I never had myself as a child. I fail to grasp the sentiment. I suspect the prevailing, unspoken theory runs along the lines that restrictions on certain kinds of information promotes God-fearing behaviours, and that skin mags falling into the hands of impressionable minors somehow upsets this delicate apple cart. I presume it must be religious at root, because it's so often invoked but never explained.
In saying this, you miss one of the essential points about having a few beefy supplies provide power for all the processors in a large rack: all those processing elements averaged out make it much easier to design the power supply to operate the majority of the time within the high efficiency zone. pair.com tunes their servers to an average 0.4 load factor. 200 servers in a rack all tuned to an expected 0.4 load factor would present a reasonably uniform load most of the time. Variance decreases on roughly the square root of population size.
Liquid propane stored in an approved wine cellar weighs about 5 pounds per gallon. To combust that propane you'll need about three times that mass of oxygen, producing 20 pounds per gallon of combustion products. A 350 gallon tank would produce 7000 pounds of combustion product. One tonne of TNT produces roughly one tonne of combustion product. IIRC the earth's atmosphere masses about 1kg per cubic meter, which is 16% oxygen by mass. The 5000 odd pounds of oxygen required to combust the propane represents about 12000 cubic meters of atmosphere, a rather large blimp to tether to the chimney out back. If you explosively decompress the propane into the kevlar blimp, you can burn off all that propane rather quickly. But the TNT still wins, not even counting the explosive multiple from pick-up truck with Goodyear tires to a puffy Goodyear blimp, at which point it is still just as hot and angry as the propane explosion before the kevlar rips.
I once knew a teacher who was nearly terminated from her job for physically interfering with a pre-school age boy who was standing on desks and peeing on the other preschoolers. The most important course a teacher takes is the one on anger management skills. If we all had better anger management skills, Windows would never have been such a problem in the first place.
There's hardly any subject that weeds out the people who can think from those who can't more effectively.
Darwin based his theory on morphological observations long before the genetic mechanisms were first postulated. Darwin couldn't even have defined "rate of change" within the modern view. I was reading about rare blood types just the other day.
Rare blood types can cause supply problems for blood banks and hospitals. For example, U-negative and Duffy-negative are two blood groups that occur only within people of African origin, and even then they are rare traits. The rarity of these factors can result in a shortage of U-negative and Duffy-negative blood for patients of African ethnicity.
The human species contains a lot of genetic diversity hidden away in small subpopulations. The origin of these unusual genes is most likely a slow process. However, changes in the configuration space of a population can take place extremely rapidly (tens of generations). For example, within the human species, red hair is more common among the Irish than it is among the Japanese. If it turned out that red hair confered immunity to next great plague, the configuration space of the human species could change extremely rapidly.
Imagine you have two human populations, one with the trait "narrow hips" and the other with the trait "wide hips". The mutation "big head" might confer some survival advantage if the infant/mother survive child-birth, but each time the mutation takes place within the "narrow hip" population it soon disappears again. Natural variation can supply the same mutation over and over again but it can't take hold until it occurs against the background of a viable configuration space.
What's most apparent in the fossil record is not the introduction of new traits, but the collapse in the diversity of alternatives. If a global pandemic wiped out the entire human population except for a small group of Duffy-negative individuals in Africa, it would seem like the "evolution" of human blood type had taken a surprising jump. Quite the contrary, it wouldn't even be anything new, but just a lot more of what you already had in the population (largely unnoticed) at the expense of a lot of diversity that used to exist.
We've had about fifty years of experience now whether evolution runs fast or slow with since the introduction of synthetic anti-biotics.
Resistance to individual antibiotics comes first, emergence of the superbug comes later. The superbug is just a configuration of genes that emerged elsewhere. Which portions of this process were the fast or slow parts, or does it just depend on what you are looking at (the emergence of novel genes, or the blood infection mortality spike thirty years later)?
On the longer time scale, we see mass extinction events every 50 to 100 million years or so.
It could be construed as periods of relative stability punctuated by great change. Or in the cartoon mentality that pervades these debates: fifty million years of not much happening, then blam an asteroid strikes and evolution kicks itself into high gear. Here's the question. Take a species that survives the extinction event (after 50 million years of relative stability) and ask yourself this: if the asteroid had struck twenty million years sooner, would that species have survived? Or did something not terribly obvious change within that species during that time period that permitted it to survive when it might not have otherwise? Or conversely, did "quiet" changes take place in other species who would have survived if the asteroid event had come sooner, but didn't in the e
Economical? You must uniformly clueless about the economics of the industrial world to suggest replacing titanium dioxide (a cheap food grade additive) with metalic titanium. Perhaps you know a good place to scavenge some from a retired YF22.
YF22 composition:
# 28% Composites (carbon-carbon, thermoplastics etc.) # 37% Titanium # 20% Metal (Aluminum and Steel), # 15% "other" material, probably RAM absorbent materials which are still highly classified.
Yesterday my brother told me that his attempt to purchase a circa 1900 wool carpet for a curatorial display (hand-made replica from England) was turned down by the provincial office on the grounds that it was "more expensive per square foot than the carpet in the premier's office". What a shock to discover that my keyboard has more bacteria than the premier's toilet seat. Yet not once in my life has anyone passed along to me any qualitative information on how much illness is passed along from toilet seats. The only factoid I've come across was a study that neat-nick cooks (mostly women) managed to spread salmonella almost everywhere when cooking raw chicken by swiping every surface with the same unbleached rag. But why worry about chicken, we can't catch anything from birds. Let's worry instead about the bacteria we are all catching from each other every day. It makes for a better sound bite.
I bet the average refrigerator door handle has even more bacteria than my keyboard and the toilet seat in the premier's office added together.
For every three seconds it takes to fix the spelling of a common word (its it's their there they're to too your you're etc.) it takes three minutes amortized over the slashdot faithful to moderate gripes off-topic and wastes valuable moderation points that could have been applied to outright stupidity instead. I don't get the logic here about leaving "low hanging fruit" to appease the born complainers. Not only does it set a low example for think before you post, it also sets a low standard for the use of moderation points.
I think the guy is an idiot. Survival is what got us into this mess in the first place. It will only get us into more trouble again if we keep doing it.
What I find amazing about the group of C++ detractors as a whole is how rarely I comprehend the claims put forward about the vaguely defined desirable language ~C++.
I think "you don't pay for what you don't use" is a fundamental design flaw of the language.
What is the precise claim here? That the entire language niche of pay-as-you-play languages should have remained empty? That C++ was the wrong language to occupy this niche? That there is a finite set of everyone-pays-all-the-time features that could have been added to C++ without compromising the language's scope or applicability? That any two people asked to write down such a list would produce a non-empty intersection?
Pay-as-you-play enables compositionality: the very idea that libraries like Boost can exist and be 90% as effective as if those same features had been designed into the language. It's the 10% that Boost doesn't achieve that gets folded back into the core language.
One guy was ranting that the true test of cohones is what the designer removes from the language, while another long post was devoted to a laundry list of "how could this language not have all these kitchen sinks so late in the day?" Which is it? You can't have minimalism in all places all of the time. Minimalism to the compiler vendor is a different beast than minimalism to the end user.
Ada had generics in 1983. Yada yada yada. What do you get when you start with a clean slate in 1995?
"Java is much more programmer-friendly than C or C++, or was for a few years there until they made just as complicated. It's become arguably even harder to learn than C++," Andreessen said. And the mantle of simplicity is being passed on: "PHP is such is an easier environment to develop in than Java."
The problem with J2EE really is that it became very, very academic and the complexity of all these perfectly designed systems in schools does not necessarily map when you have deadlines and all kinds of other things.
When Java designers decided to disallow operator overloading, they cited C++ as an example of the inherent woes of this feature. As usual, they got it wrong, which is why operator overloading is slowly but surely creeping into Java just as generics recently did.
Peter Yared, former CTO for Sun J2EE app server unit says Java/J2EE may lose out to Open Source technologies in the future, as IT managers are architects get tired of the time and cost of building in Java.
The sad fact is that few of the C++ detractors out there could do any better than Java, and Java didn't hit its own sweet spot any better than C++ mapped to its own misbegotten design criteria.
Because that's not how the free market economy works. Don't like rebates? Then don't buy products with rebates. If nobody goes along with it, they'll quit doing it on their own.
There are no end of regulations about how corporations can advertise their products on the principle that bad money drives out good. The sleezeballs out there taking advantage of the consumer have a chilling effect on ethical operations.
Recently I bought a bag of rice crisps on sale with "zero trans fats" and "now made with Sunflower oil" emblazened on the front of the package. When I arrived home I discovered this product contained mostly hydrogenated soybean oil. Sunflower oil was listed between onion powder and black pepper. In Canada, the rule is that ingredients are listed in order by relative contribution. This product was promoted on an end-of-aisle display where if you stopped to turn over the bag, you end up blocking three other carts behind you. Now there are many pie-wagons who don't seem to care about standing inertly in major choke points, but I prefer not to count myself among their numbers.
We don't need this crap. Hydrogenated oils are a burden on the public health-care system. People who are making a least a semblence of an effort to look after themselves shouldn't have to consult the fine print on every bag of cheesy crisps. If the bag says "now made with Sunflower oil" on the front, then that oil should be the most prominent oil on the listed ingredients. Anything else constitutes fraud and deception.
I believe in the free market. I'd like to sell the parent poster into indentured servitude.
Like there's no better use of time in this world than filling out rebate forms. Why, that's even worse than commenting on Slashdot. How low can you go? If the cheese-ball retailers wanted the rebate system to work, the rebate application would be filed automatically at the till. When the vendor guarantees me future in-store credits for every rebate without me having to lift one finger beyond paying for the product from a trackable account (to associate with my credit) then I'll believe that rebates are serious. In their present guise rebates are extremely undignified: sell your time and attention for a tiny spiff on a mostly irrelevent toy. Good thing India is taking over all the jobs, that leaves us with more time over here for prosecuting rebates on our 50 megabit wireless connection so we can download more celebrities.
Newspapers are filled with brilliance. I was laughing just last night about examples in the local rag. One was "overcame a stiff tailwind [to win the sprint]". Two days ago I read online that Ricky Ray, who had completed only one touchdown pass in his previous eight starts, now had "more detractors than touchdown passes". A valiant attempt at hyperbole, but I think the ball was dropped somewhere between the ears. Plus amazing leaps of mathematics: I remember one stat that boiled down to the claim that "such and such a goalie has remained undefeated in the last ten games where the other team fails to score a goal". The next step for search engine technology is the ability to filter on sentient thought.
I'm not sure the average local paper could overcome a stiff tailwind.
P4 has out-of-order execution, speculative execution, register renaming, branch prediction
All of those features were introduced with the Pentium Pro, which was savaged at the time relative to the Pentium (which is far more like the Cell) because the pre-NT Windows codebase ran like crap in that regime (one factor was partial register stalls, but there were many issues). A decade later the compilers and general codebase has become extremely tweaked in the other direction.
After the new code optimization framework in GCC 4.x has time to mature and fully target Cell, I'd be surprised if those loss factors of 2-10 don't settle down into the range of 1-3.
To see loss in the 2-10 range suggests to me that the Cell is blocking on memory loads far more often than it should be, which could be a compiler fault.
Here is a sequence that's hard to handle at the compiler level lacking OOO in hardware:
a = **p0; b = **p1; c = **p2;
If one of *p0, *p1, *p2 is an L1 cache miss, an OOO processor will still schedule two of **p0, **p1, **p2 while waiting for the cache miss to complete. This is impossible for a compiler to achieve on non-OOO hardware unless the compiler knows in advance which of those pointers will miss. I tend to refer to this class of optimizations as "stalling in parallel". I suspect C++ does a lot of double indirections to implement vtable mechanics. Maybe this is more of an issue than I thought.
The cell designers have comptley sacrificed instruction level parallelism in exchange for thread level parrallelism. It is certainly a valid and interesting way to achieve speed, but not for single threaded applications.
This analysis is incorrect, because it fails to recognize the fixed point. By sacrificing the out-of-order (OOO) mechanisms (which are brutal for heat production) they gained enough thermal headroom to effectively the double the clock rate. In the same thermal envelop, you either get an OOO processor running at 2GHz with three or four issues pathways (three has been the rule under x86) and a very deep pipeline, or you get a processor running at 4GHz with two issue pathways and a relatively short pipeline.
A deep pipeline grants (partial) immunity from stalls and bubbles. A short pipeline grants (partial) immunity from branch misprediction effects. To make the deep pipelines work well, huge investments are required in the branch-prediction unit, which is also infamous for throwing off a lot of heat.
The main Power Processing Element is crippled at best for simple single threaded applications...
Fortunately for Cell, this is also the wrong denominator for use in this discussion. Applications might be single threaded, but systems are hardly ever single threaded. While the SPU processors handle audio, video, encryption, block I/O and other compute/bandwidth intensive primitives that most systems engage, they also off-loading cache pollution from the main Cell processor threads, both in the data space and in the task scheduling space.
Nothing will ever best the Pentium IV for single thread peak performance with no calorie spared. News flash: Intel has already given up on this flawed approach. The Pentium IV could easily beat the Opteron by cranking itself up to 6GHz if there was any practical way to extract 200W from a small core with no hot spots.
OOO served its purpose in the era where cycle time was paramount and the processor to cache cycle time ratios were in closer balance. Now that heat has become the limiting factor, we'll be seeing a lot less of that from all parties.
The reality in silicon is that we need to start rethinking those portions of the code base which only perform well under an OOO execution regime.
This can be accomplished at so many different levels. The entire OpenSSL library can be recoded for SPU coprocessors with massive speed gains. Existing code can be recompiled with modern compilers which exploit large register sets to offset lack of hardware-level OOO. Key algorithms in system libraries can be recoded using better algorithms or memory access patterns.
Those of you who insist on putting all your eggs into one 100W single threaded basket, it's time to step off the Moore's law express train. Hope you enjoy the milk run.
Slashdot Mirror
http://en.wikipedia.org/wiki/Zipf's_law
Most days I don't refer to Zipf's law. Most days I don't need to. Some days I dig a little deeper in the dictionary because the stupidity I'm confronting is less ordinary.
Easy. 256MB configurations will quickly go the way of the Dodo bird. Retail competes on sticker price for the cheapest thing on the shelf. Some morons come along and buy that cheapest thing, the less moronic allow themselves to be "up sold" into something less incapacitated, while the super moronic hang around to get "up sold" to the highest margin piece of crap displayed on the shelves for exactly that purpose (anyone here like to part with $2k? I've got some *really* **awesome** 24 gauge zipcord looking for a good home).
Just imagine when you go across with the street with your 256MB price check and the oversexed 22 year old slick working there starts giving you the hairy eyeball about "Vista compatible".
Haven't you ever heard the retail lingo "oh, those guys, we get a lot of people in here after dealing with those guys"? That's the sound of retailers driving their own (who don't fall in line) into extinction.
Any store continuing to sell 256MB configurations in the Vista epoch is going to be portrayed by every slick-haired commissioned sales droid within a five mile radius as the fat kid with the black hairs growing out of his pimple.
Right after posting that screed, I looked on my desk at the book open there and read this quotation:
'Defendit numerus', [there is safety in numbers] is the maxim of the foolish;
'Deperdit numerus', [there is ruin in numbers] of the wise.
I'm not sure if he's referring to bones, boners, or boneheads. Bonus marks for anyone who can identify the book I'm reading. You have a long time to answer, it's printed on paper with a life expectancy of several hundred years.
Let's suppose some very wealthy snot picker decides to buy more iPods than the total number Apple has already sold and fill all of these with material of his own copyright which he puts into the public domain. Now the majority of music stored on iPods is user owned. Does the iPod collective sense their collective toppling into the legal side of majorism, and suddenly for everyone all file sharing becomes legal?
I find these majoristic arguments legally amusing for failing to isolate cause and effect. The iPod I bought behaves in certain ways solely because of presumptions about what other iPod purchasers are doing, at some select point in time.
Well, you could argue that because some RIAA controlled music exists, it becomes possible that a majority of iPod users have loaded mostly RIAA controlled tracks, and therefor sharing features should be disabled on the prospect that its use might be more illegal than legal, counting instance against instance.
But isn't that argument strange too? The main reason this argument holds sway is because RIAA is a functional monopoly. RIAA remains a functional monopoly in part because devices such as the iPod make it inconvenient, even in the legal case, to work around the RIAA monopoly.
Let's try another syllogism. Music is cool. RIAA has established themselves as the music monopoly. Apple is cool. Cool is a RIAA monopoly. Therefore Apple implements DRM.
It's no worse than your circular majorism, and about the level of the average person who drools over these devices. Let's face it. Cool has another property: it shuts off the brain. If you're cool, you're getting action, and if you're getting action, the brain is vestigial, and then what does it matter that RIAA is getting their cut? So is Pfizer, Carter-Wallace, Ortho McNeil, and most of the time, sooner or later, Proctor and Gamble.
I agree entirely. Only a truly Vogonesque mind would even think to pose such a question in those terms. "Ah, but if we could make it more stupid, why wouldn't we?" While we're at it, after we've cut off Britian and Australia and Taiwan, why don't we make it impossible to phone those places as well? And really, passports offices are for dodos, anyway. How many times have you been to Santiago this year? Oh come on, no one goes to Santiago twice in one year.
If eighty cents of every dollar I spend supporting OpenSSH gets flushed down the OpenBSD toilet, is that a good use of my contribution?
The cluelessness of this post defies belief.
I want to support this OpenFoil airplane wing because it supports me. However, if eighty cents of every dollar I spend supporting OpenFoil is vented through the OpenBlow high-test wind tunnel, is that a good use of my contributions?
NX protection, Pro-police, and priv-sep are all products of the two efforts coordinated together. Almost every dime OpenBSD spends is spent in the pursuit of enhancing security, and it's to imagine that those results are not immediately folded back into OpenSSH. Unlike FreeBSD, OpenBSD spends shockingly little on the OS itself. They aren't busy inventing disk geometry managers or porting to 150 different platforms.
90% of human stupidity originates in the capacity of the human mind to engage in intellectual shell games. Here is this dollar: let's split it up in to the 80 cents wasted on OpenBSD and the 20 cents invested in OpenSSH.
Or, my brother is dying of Leukemia. I want to donate blood because blood keeps him alive. Is that a good investment if 80% of the blood I donate is flushed down the toilet to replace blood lost during bone marrow transplants?
Almost too dumb to live, really.
The guy who had the idea that all the consultants who install OpenBSD or OpenSSH for their clients should bill them for the official OpenBSD install set, so they can stack the shiny disks right beside the eight volumes of Vista they don't yet have was halfway to a good idea. [OK, I paraphrased.]
OpenBSD should release a special CD entitled "OpenBSD Recovery CD" for exactly this purpose. Fair game. You could use this CD to recover your system if you wanted to. Or, for a one time $50K payment, you can purchase the Enterprise-grade "OpenBSD Rescue CD" with exactly the same content.
I think that OpenBSD has not picked up the corporate support from big corporations who rely on OpenBSD such as HP et al. because the big corporations prefer to align themselves with feel good stories such as Barry Bonds, rather than difficult people like Theo.
It almost never fails when assessing claims for a new technology to look at the list of proposed applications. You get a pretty darn good "humbug" feeling long before you dig into the technical details, which are invariably thin on the ground. The other fabulous telltale is "commercial applications in five years". "Five years" is venture capital speak for "we have no clue". "Ten years" is primary science speak for "I've been cited three times already".
99 inventions out of a hundred that promise the moon deliver green cheese.
Even talking to my father's generation, the general level of ignorance that prevailed concerning marital relations likely did more harm than skin mags. Where was this idea originally hatched, that skin mags are harmful to minors? I've never had an adult person confess to me to having been harmed in his / her teenage years by coming across the wrong kind of printed material under their parent's bed. One could argue that children are harmed by exposure to emotionally charged materials in the *absense* of appropriate parental guidance, yet apparently many/most parents find it difficult to offer this guidance with destroying some cherished illusion I never had myself as a child. I fail to grasp the sentiment. I suspect the prevailing, unspoken theory runs along the lines that restrictions on certain kinds of information promotes God-fearing behaviours, and that skin mags falling into the hands of impressionable minors somehow upsets this delicate apple cart. I presume it must be religious at root, because it's so often invoked but never explained.
In saying this, you miss one of the essential points about having a few beefy supplies provide power for all the processors in a large rack: all those processing elements averaged out make it much easier to design the power supply to operate the majority of the time within the high efficiency zone. pair.com tunes their servers to an average 0.4 load factor. 200 servers in a rack all tuned to an expected 0.4 load factor would present a reasonably uniform load most of the time. Variance decreases on roughly the square root of population size.
Liquid propane stored in an approved wine cellar weighs about 5 pounds per gallon. To combust that propane you'll need about three times that mass of oxygen, producing 20 pounds per gallon of combustion products. A 350 gallon tank would produce 7000 pounds of combustion product. One tonne of TNT produces roughly one tonne of combustion product. IIRC the earth's atmosphere masses about 1kg per cubic meter, which is 16% oxygen by mass. The 5000 odd pounds of oxygen required to combust the propane represents about 12000 cubic meters of atmosphere, a rather large blimp to tether to the chimney out back. If you explosively decompress the propane into the kevlar blimp, you can burn off all that propane rather quickly. But the TNT still wins, not even counting the explosive multiple from pick-up truck with Goodyear tires to a puffy Goodyear blimp, at which point it is still just as hot and angry as the propane explosion before the kevlar rips.
I once knew a teacher who was nearly terminated from her job for physically interfering with a pre-school age boy who was standing on desks and peeing on the other preschoolers. The most important course a teacher takes is the one on anger management skills. If we all had better anger management skills, Windows would never have been such a problem in the first place.
Darwin based his theory on morphological observations long before the genetic mechanisms were first postulated. Darwin couldn't even have defined "rate of change" within the modern view. I was reading about rare blood types just the other day.
http://en.wikipedia.org/wiki/Blood_type
The human species contains a lot of genetic diversity hidden away in small subpopulations. The origin of these unusual genes is most likely a slow process. However, changes in the configuration space of a population can take place extremely rapidly (tens of generations). For example, within the human species, red hair is more common among the Irish than it is among the Japanese. If it turned out that red hair confered immunity to next great plague, the configuration space of the human species could change extremely rapidly.
Imagine you have two human populations, one with the trait "narrow hips" and the other with the trait "wide hips". The mutation "big head" might confer some survival advantage if the infant/mother survive child-birth, but each time the mutation takes place within the "narrow hip" population it soon disappears again. Natural variation can supply the same mutation over and over again but it can't take hold until it occurs against the background of a viable configuration space.
What's most apparent in the fossil record is not the introduction of new traits, but the collapse in the diversity of alternatives. If a global pandemic wiped out the entire human population except for a small group of Duffy-negative individuals in Africa, it would seem like the "evolution" of human blood type had taken a surprising jump. Quite the contrary, it wouldn't even be anything new, but just a lot more of what you already had in the population (largely unnoticed) at the expense of a lot of diversity that used to exist.
We've had about fifty years of experience now whether evolution runs fast or slow with since the introduction of synthetic anti-biotics.
http://en.wikipedia.org/wiki/Antibiotic_resistance
Resistance to individual antibiotics comes first, emergence of the superbug comes later. The superbug is just a configuration of genes that emerged elsewhere. Which portions of this process were the fast or slow parts, or does it just depend on what you are looking at (the emergence of novel genes, or the blood infection mortality spike thirty years later)?
On the longer time scale, we see mass extinction events every 50 to 100 million years or so.
http://en.wikipedia.org/wiki/Extinction_events
It could be construed as periods of relative stability punctuated by great change. Or in the cartoon mentality that pervades these debates: fifty million years of not much happening, then blam an asteroid strikes and evolution kicks itself into high gear. Here's the question. Take a species that survives the extinction event (after 50 million years of relative stability) and ask yourself this: if the asteroid had struck twenty million years sooner, would that species have survived? Or did something not terribly obvious change within that species during that time period that permitted it to survive when it might not have otherwise? Or conversely, did "quiet" changes take place in other species who would have survived if the asteroid event had come sooner, but didn't in the e
Economical? You must uniformly clueless about the economics of the industrial world to suggest replacing titanium dioxide (a cheap food grade additive) with metalic titanium. Perhaps you know a good place to scavenge some from a retired YF22.
YF22 composition:
# 28% Composites (carbon-carbon, thermoplastics etc.)
# 37% Titanium
# 20% Metal (Aluminum and Steel),
# 15% "other" material, probably RAM absorbent materials which are still highly classified.
Yesterday my brother told me that his attempt to purchase a circa 1900 wool carpet for a curatorial display (hand-made replica from England) was turned down by the provincial office on the grounds that it was "more expensive per square foot than the carpet in the premier's office". What a shock to discover that my keyboard has more bacteria than the premier's toilet seat. Yet not once in my life has anyone passed along to me any qualitative information on how much illness is passed along from toilet seats. The only factoid I've come across was a study that neat-nick cooks (mostly women) managed to spread salmonella almost everywhere when cooking raw chicken by swiping every surface with the same unbleached rag. But why worry about chicken, we can't catch anything from birds. Let's worry instead about the bacteria we are all catching from each other every day. It makes for a better sound bite.
I bet the average refrigerator door handle has even more bacteria than my keyboard and the toilet seat in the premier's office added together.
For every three seconds it takes to fix the spelling of a common word (its it's their there they're to too your you're etc.) it takes three minutes amortized over the slashdot faithful to moderate gripes off-topic and wastes valuable moderation points that could have been applied to outright stupidity instead. I don't get the logic here about leaving "low hanging fruit" to appease the born complainers. Not only does it set a low example for think before you post, it also sets a low standard for the use of moderation points.
I think the guy is an idiot. Survival is what got us into this mess in the first place. It will only get us into more trouble again if we keep doing it.
What I find amazing about the group of C++ detractors as a whole is how rarely I comprehend the claims put forward about the vaguely defined desirable language ~C++.
r e+Java+isnt/2100-1012_3-5903187.html
5 0,39129961,00.htm
s plus&seqNum=200&rl=1
I think "you don't pay for what you don't use" is a fundamental design flaw of the language.
What is the precise claim here? That the entire language niche of pay-as-you-play languages should have remained empty? That C++ was the wrong language to occupy this niche? That there is a finite set of everyone-pays-all-the-time features that could have been added to C++ without compromising the language's scope or applicability? That any two people asked to write down such a list would produce a non-empty intersection?
Pay-as-you-play enables compositionality: the very idea that libraries like Boost can exist and be 90% as effective as if those same features had been designed into the language. It's the 10% that Boost doesn't achieve that gets folded back into the core language.
One guy was ranting that the true test of cohones is what the designer removes from the language, while another long post was devoted to a laundry list of "how could this language not have all these kitchen sinks so late in the day?" Which is it? You can't have minimalism in all places all of the time. Minimalism to the compiler vendor is a different beast than minimalism to the end user.
Ada had generics in 1983. Yada yada yada. What do you get when you start with a clean slate in 1995?
Does it thrill Marc Andreessen?
http://news.com.com/Andreessen+PHP+succeeding+whe
"Java is much more programmer-friendly than C or C++, or was for a few years there until they made just as complicated. It's become arguably even harder to learn than C++," Andreessen said. And the mantle of simplicity is being passed on: "PHP is such is an easier environment to develop in than Java."
Does it thrill Miguel de Icaza?
http://www.builderau.com.au/program/work/0,390246
The problem with J2EE really is that it became very, very academic and the complexity of all these perfectly designed systems in schools does not necessarily map when you have deadlines and all kinds of other things.
http://www.informit.com/guides/content.asp?g=cplu
When Java designers decided to disallow operator overloading, they cited C++ as an example of the inherent woes of this feature. As usual, they got it wrong, which is why operator overloading is slowly but surely creeping into Java just as generics recently did.
Does it thrill Sun insiders?
http://idevnews.com/CaseStudies.asp?ID=170
Peter Yared, former CTO for Sun J2EE app server unit says Java/J2EE may lose out to Open Source technologies in the future, as IT managers are architects get tired of the time and cost of building in Java.
The sad fact is that few of the C++ detractors out there could do any better than Java, and Java didn't hit its own sweet spot any better than C++ mapped to its own misbegotten design criteria.
Because that's not how the free market economy works. Don't like rebates? Then don't buy products with rebates. If nobody goes along with it, they'll quit doing it on their own.
There are no end of regulations about how corporations can advertise their products on the principle that bad money drives out good. The sleezeballs out there taking advantage of the consumer have a chilling effect on ethical operations.
Recently I bought a bag of rice crisps on sale with "zero trans fats" and "now made with Sunflower oil" emblazened on the front of the package. When I arrived home I discovered this product contained mostly hydrogenated soybean oil. Sunflower oil was listed between onion powder and black pepper. In Canada, the rule is that ingredients are listed in order by relative contribution. This product was promoted on an end-of-aisle display where if you stopped to turn over the bag, you end up blocking three other carts behind you. Now there are many pie-wagons who don't seem to care about standing inertly in major choke points, but I prefer not to count myself among their numbers.
We don't need this crap. Hydrogenated oils are a burden on the public health-care system. People who are making a least a semblence of an effort to look after themselves shouldn't have to consult the fine print on every bag of cheesy crisps. If the bag says "now made with Sunflower oil" on the front, then that oil should be the most prominent oil on the listed ingredients. Anything else constitutes fraud and deception.
I believe in the free market. I'd like to sell the parent poster into indentured servitude.
It's the ONLY way that the manufacturer can VERY QUICKLY stimulate buyers to buy their products.
I agree. It's well known that designing a good product in the first place ranks among the slowest methods of stimulating buyers.
Like there's no better use of time in this world than filling out rebate forms. Why, that's even worse than commenting on Slashdot. How low can you go? If the cheese-ball retailers wanted the rebate system to work, the rebate application would be filed automatically at the till. When the vendor guarantees me future in-store credits for every rebate without me having to lift one finger beyond paying for the product from a trackable account (to associate with my credit) then I'll believe that rebates are serious. In their present guise rebates are extremely undignified: sell your time and attention for a tiny spiff on a mostly irrelevent toy. Good thing India is taking over all the jobs, that leaves us with more time over here for prosecuting rebates on our 50 megabit wireless connection so we can download more celebrities.
Newspapers are filled with brilliance. I was laughing just last night about examples in the local rag. One was "overcame a stiff tailwind [to win the sprint]". Two days ago I read online that Ricky Ray, who had completed only one touchdown pass in his previous eight starts, now had "more detractors than touchdown passes". A valiant attempt at hyperbole, but I think the ball was dropped somewhere between the ears. Plus amazing leaps of mathematics: I remember one stat that boiled down to the claim that "such and such a goalie has remained undefeated in the last ten games where the other team fails to score a goal". The next step for search engine technology is the ability to filter on sentient thought.
I'm not sure the average local paper could overcome a stiff tailwind.
P4 has out-of-order execution, speculative execution, register renaming, branch prediction
All of those features were introduced with the Pentium Pro, which was savaged at the time relative to the Pentium (which is far more like the Cell) because the pre-NT Windows codebase ran like crap in that regime (one factor was partial register stalls, but there were many issues). A decade later the compilers and general codebase has become extremely tweaked in the other direction.
After the new code optimization framework in GCC 4.x has time to mature and fully target Cell, I'd be surprised if those loss factors of 2-10 don't settle down into the range of 1-3.
To see loss in the 2-10 range suggests to me that the Cell is blocking on memory loads far more often than it should be, which could be a compiler fault.
Here is a sequence that's hard to handle at the compiler level lacking OOO in hardware:
a = **p0;
b = **p1;
c = **p2;
If one of *p0, *p1, *p2 is an L1 cache miss, an OOO processor will still schedule two of **p0, **p1, **p2 while waiting for the cache miss to complete. This is impossible for a compiler to achieve on non-OOO hardware unless the compiler knows in advance which of those pointers will miss. I tend to refer to this class of optimizations as "stalling in parallel". I suspect C++ does a lot of double indirections to implement vtable mechanics. Maybe this is more of an issue than I thought.
The cell designers have comptley sacrificed instruction level parallelism in exchange for thread level parrallelism. It is certainly a valid and interesting way to achieve speed, but not for single threaded applications.
...
This analysis is incorrect, because it fails to recognize the fixed point. By sacrificing the out-of-order (OOO) mechanisms (which are brutal for heat production) they gained enough thermal headroom to effectively the double the clock rate. In the same thermal envelop, you either get an OOO processor running at 2GHz with three or four issues pathways (three has been the rule under x86) and a very deep pipeline, or you get a processor running at 4GHz with two issue pathways and a relatively short pipeline.
A deep pipeline grants (partial) immunity from stalls and bubbles. A short pipeline grants (partial) immunity from branch misprediction effects. To make the deep pipelines work well, huge investments are required in the branch-prediction unit, which is also infamous for throwing off a lot of heat.
The main Power Processing Element is crippled at best for simple single threaded applications
Fortunately for Cell, this is also the wrong denominator for use in this discussion. Applications might be single threaded, but systems are hardly ever single threaded. While the SPU processors handle audio, video, encryption, block I/O and other compute/bandwidth intensive primitives that most systems engage, they also off-loading cache pollution from the main Cell processor threads, both in the data space and in the task scheduling space.
Nothing will ever best the Pentium IV for single thread peak performance with no calorie spared. News flash: Intel has already given up on this flawed approach. The Pentium IV could easily beat the Opteron by cranking itself up to 6GHz if there was any practical way to extract 200W from a small core with no hot spots.
OOO served its purpose in the era where cycle time was paramount and the processor to cache cycle time ratios were in closer balance. Now that heat has become the limiting factor, we'll be seeing a lot less of that from all parties.
The reality in silicon is that we need to start rethinking those portions of the code base which only perform well under an OOO execution regime.
This can be accomplished at so many different levels. The entire OpenSSL library can be recoded for SPU coprocessors with massive speed gains. Existing code can be recompiled with modern compilers which exploit large register sets to offset lack of hardware-level OOO. Key algorithms in system libraries can be recoded using better algorithms or memory access patterns.
Those of you who insist on putting all your eggs into one 100W single threaded basket, it's time to step off the Moore's law express train. Hope you enjoy the milk run.