Flash drives offer over an order of magnitude decrease in random read access times... now that is a vast improvement (there is a distinct lack of SAS flash drives BTW). Now ATA's command queuing has been fixed the advantages of SCSI has been eroded. SAS is very nice, but SCSI's contribution to that is rather slim and it's continued use resulting for a large part from the artificial market separation of high RPM drives.
Sometimes the I in RAID is important even with SAS, if all you are going for is large storage and high throughput SATA can be the right solution even if you are using 50 drives.
Not as long as black lists are used to force change through collateral damage, not as long as they can start flagging every IP for some random reason... but most importantly, not as long as they fuck up, which they inevitably do.
If it was just a case of wanting to drop e-mail if you are almost certain it's spam you could do that with a Bayesian filter too. A blacklist is only one indicator of many... no matter how reliable you think it is, ultimately by using it as a single indicator at the IP level you will block e-mails which have a lower chance of being spam than e-mails you actually let through.
In these discussions I can't escape feeling a similarity with discussions about Wikipedia delitionism... a whole lot of rationalization to cover up a God complex.
Flagging everything from those IPs as spam is obviously just as reliable as throwing them away, so lets forget about the reliability non issue... Which leaves us with the expense. How much would it cost to do it the Right Way from a user's point of view? (Flagging and opt-in or opt-out filtering.)
On chip they are pumping the signal over a traces with mm range lengths and um range widths, off chip it's over traces with dm range lengths and mm range widths. Timing and power consumption are hard enough problems on chip, off chip they become much harder... not to mention that most of the power consumed either goes into EM or gets coupled into other signals.
Serial connections help with the timing, but do diddly for power and noise. That's where optical comes in.
Cultures don't change in the space of a generation. Militant Islam might confuse civilized behavior for cowardice, but this is the same world Hitler was born in. If militant Islam keeps creating martyrs in the west then shit is going to hit the fan in the next big economic downturn.
For large drives you have to put quite a few chips on anyway, put them on in parallel... and hey presto there you have the throughput you want. No amount of drives however will lower the access times of hard drives to 0.1 ms.
It wasn't broken until Microsoft specifically set out to break it, but ISO is screwed now doesn't have bylaws which allow it to fix itself. Microsoft has put it's finger on a button which tones the death bell for ISO, everyone is now in full damage control mode to disguise that fact... but it's all in vain. If you give very small countries with corrupt governments disproportional voting power in international bodies your process will get corrupted. Look at FIFA.
You mean simulate the forward emanation of quantum wave functions from light sources to the camera? (Closest to real we can get with present physics knowledge.)
"Rasterization" scales perfectly well inside the chip, to scale it in the same way in a multichip setup takes more bandwidth than they are willing to allocate pins for at the moment (so they use AFR... not SLI BTW). I don't care though, single chip solutions are in the right price range for me and multichip solutions are not.
In my price range raytracing scales from unusably slow to still unusably slow. Even if there were dedicated hardware for it that would still be true, raytracing just has so many nasty aspects which make it slow for the kind of rendering we can do in realtime at the moment in consumer space... the way it hits memory, the need for fine grained hierarchies etc... it's just nasty. Even after swallowing overdraw and the need for hugely oversampled shadow maps "rasterization" is still the better algorithm.
Everything else is just a coarse approximation which doesn't correspond to our best knowledge of light propogation. Forward raytracing? Pshaw, just a complete and utter hack. Backward raytracing can handle caustics and GI... but at the same time, still atrocious hacks really which can't handle a whole host of optical effects.
What's the point in playing a rules heavy game if the DM is going to fudge every second roll anyway? Might as well play a storytelling game pure and simple.
The only way to test obviousness is with a jury of people skilled in the relevant art (a true jury of peers for an inventor). The problem is these people cost too much (both for patent examination and even trials). That's why obviousness is never going to be a significant part of patent examination, the best we get is lawyer developed objective methods (which boil down to "it's the first time it appeared in print so it's non obvious) and the amateur opinion of patent examiners (who will get shouted down by lawyers if they turn down a patent for something as subjective as their opinion on obviousness).
You can not make obviousness an integral part of the patent process, even if the lawyers would let you (which they won't) it's simply impossible.
So pick your poison, patents on everything... or patents on nothing. I chose patents on nothing.
If a really capable hacker just decided the next time a windows worm is discovered to trojan all the transactions for a large number of banks the damage he will be able to cause is going to be huge, if he wants to be nasty he could use the online transaction history to make the transactions look legit too to maximize the amount of money he could pump around before you guys simply shutdown online transactions entirely.
He'd be able to make his money off put options rather than directly stolen money...
A key is a lot better than either of those, people understand what keys do, they understand what they should do if they get stolen or lost. Digital keys are almost impossible to copy, while passphrases are trivial to intercept and fingerprints are trivial to copy... two things a lot of people don't understand!
An extra factor is fine, but start with what works best. What you have.
My own bank uses such a device, but they have been hit by bank specific trojans which simply let you authenticate a different transaction while you thought you were authenticating your own.
The only solution is a separate device less easily owned than a PC which displays all the transaction details. Mobile phones would work (would be nice if they used better cryptography, but even without it's a lot more difficult to exploit on a large scale without physical presence).
Linking is a vague term, but not quite vague enough to use it in the way you are using it. After reading what you said a couple of times you are talking about programs which can be linked at runtime, using "linked" in the past tense form suggests they have already been linked (as in statically linked).
The problem with NVIDIA's drivers is they are sometimes distributed as precompiled binaries, the fact that their proprietary code does not copy any code or data structures from the Linux kernel then stops mattering. IF use of programming can violate copyrights then distributing such binary driver without offering source code under the GPL is a violation.
Regardless of opinions about ideas and implementation though it was also a paraphrased form of the first claim of the patent. So either the patent clerk didn't think it was an idea, ideas can be patented... or the third option, the idea/implementation dichotomy is pure sophistry.
For it to go from a cool academic project to a mainstream compiler I think GCC (or rather Tree-SSA) will have to die, since they compete for developers, and I just don't see it happening... especially not now that Stallman has retracted his objection to exposing the necessary data for intermodule optimization (which was one of the main advantages of LLVM and also the main obstacle for it being adopted instead of Tree-SSA).
Unfortunately the situation with GCC just isn't as dire as in the EGCS era.
Slashdot Mirror
SAS is a SCSI transport, it's also a SATA transport. As for Flash drives, I meant like MTron.
Flash drives offer over an order of magnitude decrease in random read access times ... now that is a vast improvement (there is a distinct lack of SAS flash drives BTW). Now ATA's command queuing has been fixed the advantages of SCSI has been eroded. SAS is very nice, but SCSI's contribution to that is rather slim and it's continued use resulting for a large part from the artificial market separation of high RPM drives.
Sometimes the I in RAID is important even with SAS, if all you are going for is large storage and high throughput SATA can be the right solution even if you are using 50 drives.
Not as long as black lists are used to force change through collateral damage, not as long as they can start flagging every IP for some random reason ... but most importantly, not as long as they fuck up, which they inevitably do.
... no matter how reliable you think it is, ultimately by using it as a single indicator at the IP level you will block e-mails which have a lower chance of being spam than e-mails you actually let through.
... a whole lot of rationalization to cover up a God complex.
If it was just a case of wanting to drop e-mail if you are almost certain it's spam you could do that with a Bayesian filter too. A blacklist is only one indicator of many
In these discussions I can't escape feeling a similarity with discussions about Wikipedia delitionism
Flagging everything from those IPs as spam is obviously just as reliable as throwing them away, so lets forget about the reliability non issue ... Which leaves us with the expense. How much would it cost to do it the Right Way from a user's point of view? (Flagging and opt-in or opt-out filtering.)
On chip they are pumping the signal over a traces with mm range lengths and um range widths, off chip it's over traces with dm range lengths and mm range widths. Timing and power consumption are hard enough problems on chip, off chip they become much harder ... not to mention that most of the power consumed either goes into EM or gets coupled into other signals.
Serial connections help with the timing, but do diddly for power and noise. That's where optical comes in.
Cultures don't change in the space of a generation. Militant Islam might confuse civilized behavior for cowardice, but this is the same world Hitler was born in. If militant Islam keeps creating martyrs in the west then shit is going to hit the fan in the next big economic downturn.
The only historic context most of Islam seems to care about is which part of the Surahs came later.
I imagine heart muscle temperature is higher than blood temperature.
For large drives you have to put quite a few chips on anyway, put them on in parallel ... and hey presto there you have the throughput you want. No amount of drives however will lower the access times of hard drives to 0.1 ms.
It wasn't broken until Microsoft specifically set out to break it, but ISO is screwed now doesn't have bylaws which allow it to fix itself. Microsoft has put it's finger on a button which tones the death bell for ISO, everyone is now in full damage control mode to disguise that fact ... but it's all in vain. If you give very small countries with corrupt governments disproportional voting power in international bodies your process will get corrupted. Look at FIFA.
You mean simulate the forward emanation of quantum wave functions from light sources to the camera? (Closest to real we can get with present physics knowledge.)
"Rasterization" scales perfectly well inside the chip, to scale it in the same way in a multichip setup takes more bandwidth than they are willing to allocate pins for at the moment (so they use AFR ... not SLI BTW). I don't care though, single chip solutions are in the right price range for me and multichip solutions are not.
... the way it hits memory, the need for fine grained hierarchies etc ... it's just nasty. Even after swallowing overdraw and the need for hugely oversampled shadow maps "rasterization" is still the better algorithm.
In my price range raytracing scales from unusably slow to still unusably slow. Even if there were dedicated hardware for it that would still be true, raytracing just has so many nasty aspects which make it slow for the kind of rendering we can do in realtime at the moment in consumer space
WotC has squads of enforcers who go to nerd's houses and force them to buy and use new rule books!
Everything else is just a coarse approximation which doesn't correspond to our best knowledge of light propogation. Forward raytracing? Pshaw, just a complete and utter hack. Backward raytracing can handle caustics and GI ... but at the same time, still atrocious hacks really which can't handle a whole host of optical effects.
Quantum wave tracing baby, that's where it's at.
What's the point in playing a rules heavy game if the DM is going to fudge every second roll anyway? Might as well play a storytelling game pure and simple.
The only way to test obviousness is with a jury of people skilled in the relevant art (a true jury of peers for an inventor). The problem is these people cost too much (both for patent examination and even trials). That's why obviousness is never going to be a significant part of patent examination, the best we get is lawyer developed objective methods (which boil down to "it's the first time it appeared in print so it's non obvious) and the amateur opinion of patent examiners (who will get shouted down by lawyers if they turn down a patent for something as subjective as their opinion on obviousness).
... or patents on nothing. I chose patents on nothing.
You can not make obviousness an integral part of the patent process, even if the lawyers would let you (which they won't) it's simply impossible.
So pick your poison, patents on everything
If a really capable hacker just decided the next time a windows worm is discovered to trojan all the transactions for a large number of banks the damage he will be able to cause is going to be huge, if he wants to be nasty he could use the online transaction history to make the transactions look legit too to maximize the amount of money he could pump around before you guys simply shutdown online transactions entirely.
...
He'd be able to make his money off put options rather than directly stolen money
A key is a lot better than either of those, people understand what keys do, they understand what they should do if they get stolen or lost. Digital keys are almost impossible to copy, while passphrases are trivial to intercept and fingerprints are trivial to copy ... two things a lot of people don't understand!
An extra factor is fine, but start with what works best. What you have.
My own bank uses such a device, but they have been hit by bank specific trojans which simply let you authenticate a different transaction while you thought you were authenticating your own.
The only solution is a separate device less easily owned than a PC which displays all the transaction details. Mobile phones would work (would be nice if they used better cryptography, but even without it's a lot more difficult to exploit on a large scale without physical presence).
Nuff Said.
Linking is a vague term, but not quite vague enough to use it in the way you are using it. After reading what you said a couple of times you are talking about programs which can be linked at runtime, using "linked" in the past tense form suggests they have already been linked (as in statically linked).
The problem with NVIDIA's drivers is they are sometimes distributed as precompiled binaries, the fact that their proprietary code does not copy any code or data structures from the Linux kernel then stops mattering. IF use of programming can violate copyrights then distributing such binary driver without offering source code under the GPL is a violation.
Regardless of opinions about ideas and implementation though it was also a paraphrased form of the first claim of the patent. So either the patent clerk didn't think it was an idea, ideas can be patented ... or the third option, the idea/implementation dichotomy is pure sophistry.
If we are going to play that silly little semantic game how about raising the stakes ... is the following an idea or an implementation :
Putting a matrix of LEDs on each key of a keyboard?
Of course changing 2006 to 2005 in a research notebook isn't that hard ...
For it to go from a cool academic project to a mainstream compiler I think GCC (or rather Tree-SSA) will have to die, since they compete for developers, and I just don't see it happening ... especially not now that Stallman has retracted his objection to exposing the necessary data for intermodule optimization (which was one of the main advantages of LLVM and also the main obstacle for it being adopted instead of Tree-SSA).
Unfortunately the situation with GCC just isn't as dire as in the EGCS era.