Perhaps it's time for Linux to split into two seperate camps. A version for Linux for servers, and a version for the desktop.
No. Adding high-end server features to the kernel won't negatively impact the desktop market; they just won't use ReiserFS or hot-swap their CPUs. And making X not suck won't change the servers at all, because they won't be using it. Forking the code again and again just makes for more incompatibily, and people will port many major improvements anyway. Work on the two issues separately, but leave the code in one piese (or in as many pieces as it already is).
It used to be that when people were going to make a dumb joke that had been made a thousand times before, they would just make it. And they got modded down, but they deserved it.
Now, it seems that people are making half a dumb joke which his been done to death, or slightly more than half, thinking that this is somehow ironical or something.
Hm. That's pretty sad. I actually don't believe you, unless you had about 18 MB HD space free at the time, but I'll see if Macs today have that problem. I'm sitting at a Mac right now, albeit a bit newer one (eMac 700). Let's see if I can reproduce your "Macs are inferior" problem.
I'll keep this browser in the foreground, and copy around a 154 MB file that I happen to have sitting around, from one folder to another.
Hmmm.... 17 seconds...
Try again.
Re:is this really a privacy concern?
on
NYT on RFID Tags
·
· Score: 1
I'm sure if this were possible, it would also be possible to get nonstandard equipment to locate and destroy them.
Who shall reward them? If you want to reward them, that's fine. But why should I be forced to do so? To compel me to reward them amounts to a gross imposition on my own liberties.
While I agree with most of your post, I think you are making a serious mistake here. While it is certainly a "gross imposition on your own liberties" if the government were to, say, force everyone to pay Microsoft, this is not the issue. The problem here isn't with rights, as it is ridiculous to claim that your rights are being violated; rather, it is an efficiency issue.
The issue is that you are profiting by the inventor's labor, while you are giving him nothing in return. Why shouldn't he be able to require that you must pay him something in order to profit by his labor? Let $foo be a consumer good (eg books). Suppose it were legal to take a $foo right out of the store without paying for it. Of course, you still could pay for it, but most people wouldn't. There wouldn't be any rights violated, but nobody would make $foos anymore, because they wouldn't get anything for it.
Suppose now that you have to pay for the $foo before leaving the store. Your rights haven't been violated; if you don't think the $foo is worth the price, don't buy it. It is, after all, a consumer good. You don't need it to survive, and are not forced in any way to buy it; if you do buy it, you do so because you think it is worth price($foo). Therefore, you gain on the whole.
The question then becomes, is this system the most efficient one? It is certainly quite efficient for physical ("rivalrous") goods, but it is not very efficient for intellectual or "non-rivalrous" goods, because everyone's utility could be increased at a cost of \epsilon simply by copying the $foo around to them. The current economic models deal very poorly with intellectual property, and therefore need to be modified or replaced! But what can they be replaced with? I don't know. In fact, I would say that nobody knows. Nonetheless, it is clear that not everyone can be allowed to copy the goods for free, as the lack of rewards for the inventor would stifle innovation even more than patents do today.
All you open source zealots out there, remember that the cost of programming is relatively small compared to other R&D. It may be one thing to copyleft some program that you wrote in your spare time (and I have done this myself), but it is quite another to copyleft pharmaceutical research which took years to develop in a multi-billion-dollar facility, plus several more years of expensive tests to get FDA approval. And these costs don't even take into account the failure rate of such research, ie the number of promising leads which turn out to be dead ends.
Unfortunately not. I asked my credit union when I opened a new account if I could specify the PIN and the answer was no. They claimed they were not equipped to handle it.
My bank, when you first open an account, chooses the password for you. But it's not just a hash of your account number or completely random, but it's the digits based on the telephone keypad letters of a four letter word...
Hm. New possibilities for dictionary attacks on cards held by 15-year-olds come to mind...
No. It hasn't even been proved that cracking RSA requires factoring, even with simplified assumpions about how you're going to be using it. And nobody knows whether tomorrow someone will solve factoring. You can't prove anything secure without P!=NP or some small-n approximation to that. A proof like that would be revolutionary.
They didn't say it was a dictionary attack on the password that took only an hour. Furthermore, if the description they gave was correct, their attack should be linear in the length of the password, not exponential. This means that it can be cracked easily in a few hours with brute force anyway.
The cause of all these problems is simple, and I'm amazed that Apple overlooked it. The new version number, 10.2.4, overruns the 11-bit signed integer field that Apple devotes to the version numbers for their operating systems. This causes the system to think it's version -10.2.4, and extrapolating on the rate of Apple software releases allows one to easily calculate the date that this release should have taken place: Version 1.0.0 in 1984 and Version 10.2.3 in 2003 gives version -10.2.4 to be precisely January 1, 1970. The 1969 dates are due to a roundoff error, and only occur in G3s without Altivec support.
As no version earlier than OS 8 has network time support, the new "old" system does not attempt to update from a network time server. The reverted files and dock icons are simply the computer's attempt to show the files as they exist at the "current" time.
A patch may take awhile, due to the difficulty of updating all the old system routines that use the 11-bit signed int format.
A) when you buy an EmperorLinux laptop, you are clearly still paying the Microsoft tax - they are just wiping Windows XP off of it (or you can still get it dual boot if you want it as such). And B) you are paying a shit price.
While you may be right about Windows, there's something you forgot: the care package. You get a physical manual. You get boot disks and system boot preconfigured. You get a system installation customized for that particular laptop model. And most importantly, you get really good tech support. One of my friends has a laptop by them (not that specific model though), and she says that the tech support is more than worth the added price. She isn't a total newbie, either, she's gotten a bunch of help with network problems (Harvard network's fault, not Emperor's), adding an ancient hacked 802.11b card on, etc etc.
Remember that time is money, and how much time would you spend custom-compiling your kernel, adding on whatever obscure hardware, and other tasks that these guys can help you with.
Note for the skeptical: I'm a Harvard student and a Mac zealot, not an Emperor Linux sales rep. I'm just passing on a strongly positive review.
No magic; they are reasonably fast (but not as fast as custom designed chips), and way more expensive..... although, it would be really cool to run Magic (ASIC chip design software) on these things. Probably bitching fast, too, as you have a prototype board just sitting there. And you could use it to design those "custom chips" that are so much more efficient:-)
BTW, you're right, context-switch would be a bitch, probably take 10 milliseconds.
I looked at this site several years ago, and thought, "whoa, cool idea, FPGAs would make a really fast computer." Then for two years, nothing to show for this idea. And after I programmed some FPGAs, I realized (at least partly) why: they're too slow to program. It takes on the order of milliseconds to reprogram even a moderate-sized FPGA.
And even a very large FPGA would be pretty lousy at doing SIMD, vector ops, etc. Basically, they would suck at emulating a computer's instruction set, which is (fairly well) optimized for what software actually needs to do. I can't think of many algorithms used by software today that would work much better in an FPGA, except for symmetric crypto. And if you need to do that, get an ASIC crypto chip, 10s of dollars for umpity gigs/second throughput. SPICE might also run a bit faster on these (understatement), but those types already have decent FPGA interfaces.
Furthermore, the processor programming these FPGAs must have some serious power... if you have to do many things on an FPGA at once (which you do if there are only 11 of them), you basically have to place & route on the fly, which is pretty slow.
So, I don't think that these "hypercomuters" will ever be any better than a traditional supercomputer in terms of price/performance, except for specialized applications. And even then, it won't be any better than an application specific setup. And not many people need to go back and forth between specialized tasks. (Who am I to complain of price/performance, I'm a Mac user?)
That said, if they *can* put a hypercomputer on everyone's desk for $1,000.00, more power to them!
No. I said that since 64-bit RC5 was cracked in the spare time of a few thousand guys on Earth, using a tiny fraction of the earth's energy (which is order of a billionth of the energy emitted by the sun) over less than a billionth of the sun's lifetime, and not terribly efficiently, that there is plenty of energy in the Universe's >2^64 stars to break a 128-bit key. In fact, given the above figures, there is probably enough left in our sun alone to crack millions of 128-bit keys, depending on that "tiny fraction" and on how much efficiency can be improved. I still wouldn't call that easy.
The comment about 2^256 meant that if you could somehow extract all the energy given off by all the stars in this galaxy for their lifetimes, you might have enough to count to 2^256. I'm not calling that easy either.
I put these figures in because the person I was responding to had said there was not enough energy in the universe to do 2^128.
Mike
Sounds like a reflection attack...
on
Unreal Security Hole
·
· Score: 2, Insightful
- Local and remote denial of service. - Distributed denial of service (flooding remote computers with data packets to freeze it). - Bounce attacks with spoofed UDP packets
This bit sounds an awful lot like the GameSpy reflection attack: you send them a forged UDP packet asking for some resource, they send out 400 times as much data to the poor bloke whose IP you put on it. Rinse, lather, repeat and you have yourself a pretty big DRDOS (not the guys MS killed, rather a Distributed Reflection Denial Of Service).
Therefore a 128 bit key can assume 2^128 different values and, as some other poster pointed out, there is not enough energy in the universe to overcome the background radiation as many times as it would take to count to 2^128, let alone try and brute force the cypher.
hmmm.... simple calculation: 2^64 ~ 2x10^19 #stars in the universe ~ order of 10^20 "Some things are better left unread" = 64-bit key brute-forced already using a tiny fraction of the energy available to one planet orbiting a typical star.
So... yes there is easily enough energy in the universe to brute-force a 128-bit key. In fact, there is probably (barely) enough in our galaxy to brute-force a 256-bit key, if you could get all of it. (This according to a Schneier calculation: see Applied Cryptography). But it's not going to happen anytime soon.
Furthermore, these assume that the cipher is perfect, ie that brute force is the fastest possible attack. That's the hardest part of designing ciphers, and proving that any particular cipher is perfect would require other important things like P!=NP.
Asymmetric cryptography on the other hand derives its features from mathematical properties of some of the numbers used. For example, some systems require the a product of large prime numbers, or discrete logarithms etc. This means that, for example in RSA, you cannot use all of the 2^128 values of a 128 bit key.
The security per bit of an asymmetric cipher is less mostly because you give the attacker so much information (the public key). Furthermore, the attacker can trivially generate any number of (chosen plainext)-ciphertext pairs.
If you just had to choose a pair of primes for RSA, 134 bits or so would suffice to replace that 128-bit key (yay, prime number theorem!). The catch is that you have to publish their product.
Slashdot Mirror
Perhaps it's time for Linux to split into two seperate camps. A version for Linux for servers, and a version for the desktop.
No. Adding high-end server features to the kernel won't negatively impact the desktop market; they just won't use ReiserFS or hot-swap their CPUs. And making X not suck won't change the servers at all, because they won't be using it. Forking the code again and again just makes for more incompatibily, and people will port many major improvements anyway. Work on the two issues separately, but leave the code in one piese (or in as many pieces as it already is).
They [freebsd, openbsd, netbsd] are real unix. They won't easily die for a long time.
I think the *BSD trolls would disagree with you.
This story gives a whole new meaning to that phrase.
For those of you who came in late, Unix and its workalikes (Linux etc) have grown in use exponentially since 1980.
Exactly. The point of the article is that Unix has fallen behind its workalikes, specifically Linux in this case.
And no, *BSD trolls, he did not say that *BSD is dying.
It used to be that when people were going to make a dumb joke that had been made a thousand times before, they would just make it. And they got modded down, but they deserved it.
Now, it seems that people are making half a dumb joke which his been done to death, or slightly more than half, thinking that this is somehow ironical or something.
It's not. The horse is dead. Quit beating it.
Hm. That's pretty sad. I actually don't believe you, unless you had about 18 MB HD space free at the time, but I'll see if Macs today have that problem. I'm sitting at a Mac right now, albeit a bit newer one (eMac 700). Let's see if I can reproduce your "Macs are inferior" problem.
I'll keep this browser in the foreground, and copy around a 154 MB file that I happen to have sitting around, from one folder to another.
Hmmm.... 17 seconds...
Try again.
I'm sure if this were possible, it would also be possible to get nonstandard equipment to locate and destroy them.
Barring that, there's always the microwave.
Who shall reward them? If you want to reward them, that's fine. But why should I be forced to do so? To compel me to reward them amounts to a gross imposition on my own liberties.
While I agree with most of your post, I think you are making a serious mistake here. While it is certainly a "gross imposition on your own liberties" if the government were to, say, force everyone to pay Microsoft, this is not the issue. The problem here isn't with rights, as it is ridiculous to claim that your rights are being violated; rather, it is an efficiency issue.
The issue is that you are profiting by the inventor's labor, while you are giving him nothing in return. Why shouldn't he be able to require that you must pay him something in order to profit by his labor? Let $foo be a consumer good (eg books). Suppose it were legal to take a $foo right out of the store without paying for it. Of course, you still could pay for it, but most people wouldn't. There wouldn't be any rights violated, but nobody would make $foos anymore, because they wouldn't get anything for it.
Suppose now that you have to pay for the $foo before leaving the store. Your rights haven't been violated; if you don't think the $foo is worth the price, don't buy it. It is, after all, a consumer good. You don't need it to survive, and are not forced in any way to buy it; if you do buy it, you do so because you think it is worth price($foo). Therefore, you gain on the whole.
The question then becomes, is this system the most efficient one? It is certainly quite efficient for physical ("rivalrous") goods, but it is not very efficient for intellectual or "non-rivalrous" goods, because everyone's utility could be increased at a cost of \epsilon simply by copying the $foo around to them. The current economic models deal very poorly with intellectual property, and therefore need to be modified or replaced! But what can they be replaced with? I don't know. In fact, I would say that nobody knows. Nonetheless, it is clear that not everyone can be allowed to copy the goods for free, as the lack of rewards for the inventor would stifle innovation even more than patents do today.
All you open source zealots out there, remember that the cost of programming is relatively small compared to other R&D. It may be one thing to copyleft some program that you wrote in your spare time (and I have done this myself), but it is quite another to copyleft pharmaceutical research which took years to develop in a multi-billion-dollar facility, plus several more years of expensive tests to get FDA approval. And these costs don't even take into account the failure rate of such research, ie the number of promising leads which turn out to be dead ends.
Unfortunately not. I asked my credit union when I opened a new account if I could specify the PIN and the answer was no. They claimed they were not equipped to handle it.
What's your account number?
My bank, when you first open an account, chooses the password for you. But it's not just a hash of your account number or completely random, but it's the digits based on the telephone keypad letters of a four letter word...
Hm. New possibilities for dictionary attacks on cards held by 15-year-olds come to mind...
Brilliant! Because if everyone is reading the article, nobody will be able to get through to... oh wait...
No. It hasn't even been proved that cracking RSA requires factoring, even with simplified assumpions about how you're going to be using it. And nobody knows whether tomorrow someone will solve factoring. You can't prove anything secure without P!=NP or some small-n approximation to that. A proof like that would be revolutionary.
They didn't say it was a dictionary attack on the password that took only an hour. Furthermore, if the description they gave was correct, their attack should be linear in the length of the password, not exponential. This means that it can be cracked easily in a few hours with brute force anyway.
I think she got one of the less inflated models, only a hundred or two hundred above the "Windows" price.
The cause of all these problems is simple, and I'm amazed that Apple overlooked it. The new version number, 10.2.4, overruns the 11-bit signed integer field that Apple devotes to the version numbers for their operating systems. This causes the system to think it's version -10.2.4, and extrapolating on the rate of Apple software releases allows one to easily calculate the date that this release should have taken place: Version 1.0.0 in 1984 and Version 10.2.3 in 2003 gives version -10.2.4 to be precisely January 1, 1970. The 1969 dates are due to a roundoff error, and only occur in G3s without Altivec support.
As no version earlier than OS 8 has network time support, the new "old" system does not attempt to update from a network time server. The reverted files and dock icons are simply the computer's attempt to show the files as they exist at the "current" time.
A patch may take awhile, due to the difficulty of updating all the old system routines that use the 11-bit signed int format.
A) when you buy an EmperorLinux laptop, you are clearly still paying the Microsoft tax - they are just wiping Windows XP off of it (or you can still get it dual boot if you want it as such). And B) you are paying a shit price.
While you may be right about Windows, there's something you forgot: the care package. You get a physical manual. You get boot disks and system boot preconfigured. You get a system installation customized for that particular laptop model. And most importantly, you get really good tech support. One of my friends has a laptop by them (not that specific model though), and she says that the tech support is more than worth the added price. She isn't a total newbie, either, she's gotten a bunch of help with network problems (Harvard network's fault, not Emperor's), adding an ancient hacked 802.11b card on, etc etc.
Remember that time is money, and how much time would you spend custom-compiling your kernel, adding on whatever obscure hardware, and other tasks that these guys can help you with.
Note for the skeptical: I'm a Harvard student and a Mac zealot, not an Emperor Linux sales rep. I'm just passing on a strongly positive review.
No magic; they are reasonably fast (but not as fast as custom designed chips), and way more expensive.. ... although, it would be really cool to run Magic (ASIC chip design software) on these things. Probably bitching fast, too, as you have a prototype board just sitting there. And you could use it to design those "custom chips" that are so much more efficient :-)
BTW, you're right, context-switch would be a bitch, probably take 10 milliseconds.
I looked at this site several years ago, and thought, "whoa, cool idea, FPGAs would make a really fast computer." Then for two years, nothing to show for this idea. And after I programmed some FPGAs, I realized (at least partly) why: they're too slow to program. It takes on the order of milliseconds to reprogram even a moderate-sized FPGA.
And even a very large FPGA would be pretty lousy at doing SIMD, vector ops, etc. Basically, they would suck at emulating a computer's instruction set, which is (fairly well) optimized for what software actually needs to do. I can't think of many algorithms used by software today that would work much better in an FPGA, except for symmetric crypto. And if you need to do that, get an ASIC crypto chip, 10s of dollars for umpity gigs/second throughput. SPICE might also run a bit faster on these (understatement), but those types already have decent FPGA interfaces.
Furthermore, the processor programming these FPGAs must have some serious power... if you have to do many things on an FPGA at once (which you do if there are only 11 of them), you basically have to place & route on the fly, which is pretty slow.
So, I don't think that these "hypercomuters" will ever be any better than a traditional supercomputer in terms of price/performance, except for specialized applications. And even then, it won't be any better than an application specific setup. And not many people need to go back and forth between specialized tasks. (Who am I to complain of price/performance, I'm a Mac user?)
That said, if they *can* put a hypercomputer on everyone's desk for $1,000.00, more power to them!
There was a map in MacSki where you got points off for not killing all the penguins :-)
This saying baffles me. "Looking like ass" could be good (J Lo) or bad (Goatse guy).
Sorry, but "look like ass" conveys no information.
I dunno. Most of the snowboarding games I've played have looked like snowboarding.
No. I said that since 64-bit RC5 was cracked in the spare time of a few thousand guys on Earth, using a tiny fraction of the earth's energy (which is order of a billionth of the energy emitted by the sun) over less than a billionth of the sun's lifetime, and not terribly efficiently, that there is plenty of energy in the Universe's >2^64 stars to break a 128-bit key. In fact, given the above figures, there is probably enough left in our sun alone to crack millions of 128-bit keys, depending on that "tiny fraction" and on how much efficiency can be improved. I still wouldn't call that easy.
The comment about 2^256 meant that if you could somehow extract all the energy given off by all the stars in this galaxy for their lifetimes, you might have enough to count to 2^256. I'm not calling that easy either.
I put these figures in because the person I was responding to had said there was not enough energy in the universe to do 2^128.
Mike
- Local and remote denial of service.
- Distributed denial of service (flooding remote computers with data packets to freeze it).
- Bounce attacks with spoofed UDP packets
This bit sounds an awful lot like the GameSpy reflection attack: you send them a forged UDP packet asking for some resource, they send out 400 times as much data to the poor bloke whose IP you put on it. Rinse, lather, repeat and you have yourself a pretty big DRDOS (not the guys MS killed, rather a Distributed Reflection Denial Of Service).
Therefore a 128 bit key can assume 2^128 different values and, as some other poster pointed out, there is not enough energy in the universe to overcome the background radiation as many times as it would take to count to 2^128, let alone try and brute force the cypher.
hmmm.... simple calculation:
2^64 ~ 2x10^19
#stars in the universe ~ order of 10^20
"Some things are better left unread" = 64-bit key brute-forced already using a tiny fraction of the energy available to one planet orbiting a typical star.
So... yes there is easily enough energy in the universe to brute-force a 128-bit key. In fact, there is probably (barely) enough in our galaxy to brute-force a 256-bit key, if you could get all of it. (This according to a Schneier calculation: see Applied Cryptography). But it's not going to happen anytime soon.
Furthermore, these assume that the cipher is perfect, ie that brute force is the fastest possible attack. That's the hardest part of designing ciphers, and proving that any particular cipher is perfect would require other important things like P!=NP.
Asymmetric cryptography on the other hand derives its features from mathematical properties of some of the numbers used. For example, some systems require the a product of large prime numbers, or discrete logarithms etc. This means that, for example in RSA, you cannot use all of the 2^128 values of a 128 bit key.
The security per bit of an asymmetric cipher is less mostly because you give the attacker so much information (the public key). Furthermore, the attacker can trivially generate any number of (chosen plainext)-ciphertext pairs.
If you just had to choose a pair of primes for RSA, 134 bits or so would suffice to replace that 128-bit key (yay, prime number theorem!). The catch is that you have to publish their product.
Funny. All the emails I have gotten on that subject have suggested that I "Give her something BIG for Valentine's day!!!