That's decent reasoning, but starting from a mistaken premise.
You see, the decision you cite has the _defendant_ being protected from the _government_.
Yes, that's always the case in the application of the ex post facto provision.
In the case of what we're talking about, the defendant _is_ the government.
No, wrong. The defendant in a criminal case is never the government. The defendant may be a government officer (or, in the case of the President, a former officer since the President, while in office, has absolute immunity to criminal prosecution, since that is an executive function and he can't prosecute himself, even by proxy.)
But the government is always the prosecution, and some individual person against whom the government is acting is always the defense, in a criminal case. At least in the US: some foreign systems allow private criminal prosecution.
Allowing the government to act in the way the White House and the Congress are trying to allow hurts the non-government defendants in any investigation and trial for which the surveillance is used.
Actually, no. In general, material gathered under FISA (or even the EMSA modifications) procedures for foreign intelligence material, but not under the procedures required for surveillance for law enforcement purposes which are different, may not be used in any criminal prosecution.
By legalizing what the _government_ is doing after the fact, it _narrows_ the rights of the _defendant_ after the fact.
Changing the rules of information gathering after the fact is not "ex post facto", only changing the substantive law (the definition of crimes or their punishment) is "ex post facto".
Changing the rules of information gathering for criminal cases may violate the fourth amendment whether or not the change is retroactive, and probably the provisions of FISA and even moreso those of EMSA would violate the fourth amendment if some of the information gathered that is currently prohibited from use in criminal cases were allowed to be used in criminal cases. It may violate the Fourth Amendment to gather it in any case, especially with the expanded authorization in EMSA, but in either case that concern is separate from the ex post facto issue.
The illegality of the surveillance makes it inadmissable as evidence. That's a legal defense.
Properly speaking, its a legal argument for the inadmissibility of the evidence, not a defense. And that it is illegal under statute may make it inadmissible, but that it is illegal under statute is not required for it to be inadmissible, and much of the the information gathered under FISA might be inadmissible even if it weren't prohibited by statute to use it. Further, even if it were an ex post facto issue to change the information gathering rules after the fact because of the impact on other defendants, all that would mean is that such evidence would be inadmissible against affected defendants, not that Congress' removal of the criminal penalties for the illegal surveillance would not be applicable retrospectively.
No doubt it'll be more than 8MB of cache for a behemoth like this, but do you really think they're going to be able to squeeze, say, 2MB per core without burning down the house?
I think there is a reason they have a RAM stack that the processor attaches directly to: I'm not sure how big it is supposed to be, but I would imagine its pretty big.
Also, other stories on the announcement, (like this one) highlight that the main initial target is the "mega data center", saying that Intel pointed specifically to Google and YouTube as examples.
Prior to the 486, much of what you said would, I believe, have been true of CPUs and FPUs, too; CPUs handled instructions FPUs didn't, and vice versa, applications were written to use FPUs where appropriate, etc. There were different FPUs with different instruction sets available to complement the same CPUs.
Then Intel rolled the FPU up into the CPU chip with the 486DX, keeping the functionality pretty much the same as their standalone FPU, and that was that. Sure, for a while, the low end 486SX without the FPU that could still take an external FPU was around, too, but from the standalone FPUs days were numbered.
The real problem isn't, I think, that CPUs and GPUs do different things, its that the people who pay money for premium GPUs are likely to want to upgrade in the life of the CPU, and aren't going to want to pay more to lose that freedom unless they are going to get a big performance boost out of the gate; this wasn't an issue with FPUs.
OTOH, if putting the GPU functionality on the CPU chip gives a big performance boost, that would be an incentive to adopt that kind of architecture.
I think the question was about compilers designed to produce code optimized for multiple CPU or multiple core systems, not about compilers that themselves were made to get a huge performance boost running on multiple cores.
No, microsoft is not infallible. But they do have experience with tilt controllers, and they probably also have a good bit of user feedback from the sidewinder freestyle, and they must have considered making a follow up, so their opinion on whether this is a good idea should be a bit better informed.
"should be" =/= "is". Yes, its interesting that Microsoft isn't pursuing this as a core feature, OTOH, the Xbox 360 was the first of this round of consoles out, and therefore the least able to be influenced by expected features in the others: the fact is that the response to the Wii's inclusion of this feature can reasonably be expected to influence expectations in the market, and even if the some of the people who pay intense attention to the business end of things see Sony as playing "copycat" (which they clearly are), in the long run, it may be having the feature is valuable because of something Microsoft didn't anticipate when it set the core features for the Xbox 360.
Raytracing alone isn't the be-all end-all, but if you can do what, say, POV-Ray does (which can augment raytracing with an radiosity-based rendering which can really improve lighting realism), but in realtime at a reasonable framerate with fullscreen coverage, that would be great.
I'm not sure 1 TFLOP is enough to do that and do everything else you need to do for a game, though.
Writing it in a asynchronous CSP-based language like Erlang is much easier. There's a language I saw a presentation from some guys at IBM on that looks potentially even more promising, although I can't recall its name at the moment.
I've seen quite a number of languages, mostly kinda fringey academic languages now, that focus on simple, efficient, and safe distribution of tasks across processors. Presumably, if hardware demanding that becomes more common, ideas will be ripped from them and blended with features from currently popular languages to create new languages appropriate to the new needs, just as increasing demand for features which fit well with the OO paradigm saw features from OO languages grabbed and tacked onto, among other existing non-OO languages, C, in several different forms...
No, as to all of those scientific laws that propose a mechanism which explains and predicts empirical observations; Moore's law is a fairly consistent empirical observation for which no explanation is offered or, more importantly, testably supported.
Without any understanding of the particular processes which might be producing this observed behavior, we don't have a lot of basis for guessing whether, as the context in which technology develops changes, Moore's law will continue to hold true or not.
Seriously, just wait. Don't upgrade until this sucker comes out, otherwise what ever you buy will be obsolete fast.
I generally upgrade every 2-3 years, anyhow, and I'm due for one on my "desktop" (actually, a floortop, but anyhow) about now, anyway, so I figure I can do this one and the next one without any worry that I might want to buy one of these in 5 years, since that'd be the natural time to upgrade again, anyway.
You know, that would be a more useful syscall if its return was more like is_computer_on(), immediately preceding. I do wonder, not having much exposure to BeOS, if those are jokes in the documentation, or in the actual sytem libraries.
Comparing where we are today to twelve years ago, and expecting the same or greater multiplier is absurd.
Er, I wasn't pointing to any particular multiplier. I was pointing out that, even if you are right that, when released, these would be prohibitively expensive for most purchasers, that history suggests that processors go from "prohibitively expensive for most user", to "common", to "you really need to upgrade that old piece of crap" pretty quickly.
Nobody is going to design 80 core systems unless someone is prepared to buy them and nobody is going to design 80-core chips if nobody can show how to design effective systems with them.
As intel already has apparently designed and produced prototype 80 core chips, are you telling me that somebody has shown how to design effective systems with them? Certainly, Intel has already talked (as recounted in TFA) about some of the approaches to the problems you point to. Its like you are ignoring that and sticking your fingers in your ears and screaming "Can't be done! Can't be done! Can't be done!"
If all they did was increase clock speeds, we wouldn't need as many major advancements in compiler theory.
Yeah, well its possible that the returns from investment in technology to increase clock speed is less than the return on a similar investment in compiler technology and multicore technology. The clock speed games been worked for decades, and modern processors are on the order of 1,000 times the clock speed of the first PCs. The easy returns in that avenue may already have been acheived.
But the more pressing issue is bus bandwidth. It won't take long for multiple cores to exceed the bandwidth limit to/from the system RAM.
Which probably has something to do with why they want to directly attach this 80-core monstrosity to a block of RAM, and why they are also pointing to not only its processing speed, but its enormous data transfer throughput. While the article isn't that detailed, it certainly seems to suggest that Intel is looking at a very different way of laying out the architecture to make this beast work.
If AMD manages to develop higher Ghz speed CPUs rather than continuing down Intel's "more cores = better" path, they stand a good chance of having the superior price vs. performance CPU again in the not too distant future.
Wasn't constantly working the "more cycles = better" angle while AMD did more finding way to do more with the same clock speed exactly how Intel initially lost its price/performance and overall performance dominance to AMD?
The major limitation to the effectiveness multi-cores is somewhat described in Amdahl's Law.
Things like memory bandwidth are already constraining 2-core chips. The only way to effectively mitigate this is to make wider bus paths. That's relatively easy for 2 core chips, but to get any benefit from 80-core chips you're going to need 40x the memory bandwidth you have now. That means huge pin-outs, huge amounts of RAM, huge everything.
Enormous memory bandwidth doesn't require enormous amounts of RAM, the two are completely independent. TFA does refer to the approaches being taken on the bandwidth front. Clearly, its a challenge, but new technologies usually are. They also, in the world of modern computing, often go from high-priced products with narrow markets to everyone has one to "you're still using that old stuff" fairly quickly. So, yeah, if they are introduced 5 years from now, I doubt many people will have them at first. But I wouldn't be surprised if 20 years from now they are looked at the way we'd look at, say, a 60MHz Pentium I today.
Isn't it kind of immature for a company to think "hmm, we have had relative sucess with dual core processors over our competitors. . so lets fit as many cores into a mobo as possible . ..that will get amd"
It might be, but its pretty immature of you to assume that that is what they are thinking.
While I believe that multi-core technology needs to be developed further, there are also other things for intel to be researching.
And...so? Does this announcement imply that Intel isn't researching other things?
The first thing I thought when I saw this was that they really ought to dial in Quad Core before boasting twenty times that.
Apparently, AMD will be peddling them withtin the next year.
From TFA:
As expected, Intel announced plans to have quad-core processors ready for its customers in November. An extremely fast Core 2 Extreme processor with four cores will be released then, and the newly named Core 2 Quad processor for mainstream desktops will follow in the first quarter of next year, Otellini said.
The quad-core server processors are on a similar trajectory, with a faster Xeon 5300 processor scheduled for November and a low-power Xeon slated for the first quarter. Intel's first quad-core processors are actually two of its dual-core Core architecture chips combined into a multichip package.
The costs to make use of 80 cores (you're going to need hugely complex chips and hugely complex memory buses) mean that these chips will be severe overkill for PCs and will be outside any typical user's price range.
Unless the complexity makes the manufacturing vastly expensive, rather than just the development, this won't be true: the more widely its sold, the less of a development premium there will be, because the development costs will be spread more widely.
I smell marketing horseshit. I think they're just saying this to get people to start thinking of multi-core options. Most people don't see the need for multi-core (even 2 core) systems.
Most people probably won't even know (or care) how many cores there computer has, any more than they know how many transitors it has. People who make computers understand the value of multicore systems and are adopting dual core systems for consumer PCs without problem, and will no doubt do the same thing with the quad core systems Intel plans to release soon, and even more cores as they become available.
What massively parallel tasks would possibly need 80 cores?
I'm not sure this is 80 general-purpose processing cores: the article claims that there are "80 floating point cores". Clearly, the big selling points of the chip are, in Intel's view, its data transfer at 1 TB/sec, and its floating point speed at 1 TFLOP.
I can see uses for two, maybe 4 cores but what are advantages of 80 core chip as opposed to system with 40 2-core processors we can have now?
An 80-core chip with RAM attached directly to the processor chip, as TFA discusses, is going to have an advantage in transferring data between cores, and plus it'll probably be a lot smaller. Than 40 dual core (or 20 quad core) chips.
I know of no movie that offers free watching the first 20 mins to let me know if I want to pay to watch it.
And I know of know game that gives you 1/6 of the total gameplay free, either. Both movies and games often have selected promotional material available as a means of drumming up interest, yes.
If you don't do the research for the product you want to spend your money on, I have no sympathy for you.
Who cares? No one was suggesting not doing what research can be done, and no one was asking for your sympathy for anything. I was answering your question of why people might be happy with spending money on movies and not happy with games even though the $/hour for the former was generally greater than the latter, and the latter offered more interactivity.
So that means the temperatures were that high before manufacturing and automobiles?
Yes, temperatures were this high before manufacturing and automobiles. What is concerning is not so much the current temperature, but the trend of rapid increase that has been strong since about the Industrial Revolution and appears unprecedented.
And will, if continued, in not too long reach temperatures that are, themselves, sources of grave concern. At which point, it may well be impossible to prevent catastrophic results.
Slashdot Mirror
Yes, that's always the case in the application of the ex post facto provision.
No, wrong. The defendant in a criminal case is never the government. The defendant may be a government officer (or, in the case of the President, a former officer since the President, while in office, has absolute immunity to criminal prosecution, since that is an executive function and he can't prosecute himself, even by proxy.)
But the government is always the prosecution, and some individual person against whom the government is acting is always the defense, in a criminal case. At least in the US: some foreign systems allow private criminal prosecution.
Actually, no. In general, material gathered under FISA (or even the EMSA modifications) procedures for foreign intelligence material, but not under the procedures required for surveillance for law enforcement purposes which are different, may not be used in any criminal prosecution.
Changing the rules of information gathering after the fact is not "ex post facto", only changing the substantive law (the definition of crimes or their punishment) is "ex post facto".
Changing the rules of information gathering for criminal cases may violate the fourth amendment whether or not the change is retroactive, and probably the provisions of FISA and even moreso those of EMSA would violate the fourth amendment if some of the information gathered that is currently prohibited from use in criminal cases were allowed to be used in criminal cases. It may violate the Fourth Amendment to gather it in any case, especially with the expanded authorization in EMSA, but in either case that concern is separate from the ex post facto issue.
Properly speaking, its a legal argument for the inadmissibility of the evidence, not a defense. And that it is illegal under statute may make it inadmissible, but that it is illegal under statute is not required for it to be inadmissible, and much of the the information gathered under FISA might be inadmissible even if it weren't prohibited by statute to use it. Further, even if it were an ex post facto issue to change the information gathering rules after the fact because of the impact on other defendants, all that would mean is that such evidence would be inadmissible against affected defendants, not that Congress' removal of the criminal penalties for the illegal surveillance would not be applicable retrospectively.
I think there is a reason they have a RAM stack that the processor attaches directly to: I'm not sure how big it is supposed to be, but I would imagine its pretty big.
Also, other stories on the announcement, (like this one) highlight that the main initial target is the "mega data center", saying that Intel pointed specifically to Google and YouTube as examples.
Prior to the 486, much of what you said would, I believe, have been true of CPUs and FPUs, too; CPUs handled instructions FPUs didn't, and vice versa, applications were written to use FPUs where appropriate, etc. There were different FPUs with different instruction sets available to complement the same CPUs.
Then Intel rolled the FPU up into the CPU chip with the 486DX, keeping the functionality pretty much the same as their standalone FPU, and that was that. Sure, for a while, the low end 486SX without the FPU that could still take an external FPU was around, too, but from the standalone FPUs days were numbered.
The real problem isn't, I think, that CPUs and GPUs do different things, its that the people who pay money for premium GPUs are likely to want to upgrade in the life of the CPU, and aren't going to want to pay more to lose that freedom unless they are going to get a big performance boost out of the gate; this wasn't an issue with FPUs.
OTOH, if putting the GPU functionality on the CPU chip gives a big performance boost, that would be an incentive to adopt that kind of architecture.
I think the question was about compilers designed to produce code optimized for multiple CPU or multiple core systems, not about compilers that themselves were made to get a huge performance boost running on multiple cores.
"should be" =/= "is". Yes, its interesting that Microsoft isn't pursuing this as a core feature, OTOH, the Xbox 360 was the first of this round of consoles out, and therefore the least able to be influenced by expected features in the others: the fact is that the response to the Wii's inclusion of this feature can reasonably be expected to influence expectations in the market, and even if the some of the people who pay intense attention to the business end of things see Sony as playing "copycat" (which they clearly are), in the long run, it may be having the feature is valuable because of something Microsoft didn't anticipate when it set the core features for the Xbox 360.
Raytracing alone isn't the be-all end-all, but if you can do what, say, POV-Ray does (which can augment raytracing with an radiosity-based rendering which can really improve lighting realism), but in realtime at a reasonable framerate with fullscreen coverage, that would be great.
I'm not sure 1 TFLOP is enough to do that and do everything else you need to do for a game, though.
I've seen quite a number of languages, mostly kinda fringey academic languages now, that focus on simple, efficient, and safe distribution of tasks across processors. Presumably, if hardware demanding that becomes more common, ideas will be ripped from them and blended with features from currently popular languages to create new languages appropriate to the new needs, just as increasing demand for features which fit well with the OO paradigm saw features from OO languages grabbed and tacked onto, among other existing non-OO languages, C, in several different forms...
No, as to all of those scientific laws that propose a mechanism which explains and predicts empirical observations; Moore's law is a fairly consistent empirical observation for which no explanation is offered or, more importantly, testably supported.
Without any understanding of the particular processes which might be producing this observed behavior, we don't have a lot of basis for guessing whether, as the context in which technology develops changes, Moore's law will continue to hold true or not.
Seriously, just wait. Don't upgrade until this sucker comes out, otherwise what ever you buy will be obsolete fast.
I generally upgrade every 2-3 years, anyhow, and I'm due for one on my "desktop" (actually, a floortop, but anyhow) about now, anyway, so I figure I can do this one and the next one without any worry that I might want to buy one of these in 5 years, since that'd be the natural time to upgrade again, anyway.
You know, that would be a more useful syscall if its return was more like is_computer_on(), immediately preceding. I do wonder, not having much exposure to BeOS, if those are jokes in the documentation, or in the actual sytem libraries.
Speaking of that, I wonder if development in this direction might lead, in the next decade or so, separate GPUs going the way of separate FPUs...
Er, I wasn't pointing to any particular multiplier. I was pointing out that, even if you are right that, when released, these would be prohibitively expensive for most purchasers, that history suggests that processors go from "prohibitively expensive for most user", to "common", to "you really need to upgrade that old piece of crap" pretty quickly.
As intel already has apparently designed and produced prototype 80 core chips, are you telling me that somebody has shown how to design effective systems with them? Certainly, Intel has already talked (as recounted in TFA) about some of the approaches to the problems you point to. Its like you are ignoring that and sticking your fingers in your ears and screaming "Can't be done! Can't be done! Can't be done!"
For Windows, and for Linux.
Dunno. Probably.
Yeah, well its possible that the returns from investment in technology to increase clock speed is less than the return on a similar investment in compiler technology and multicore technology. The clock speed games been worked for decades, and modern processors are on the order of 1,000 times the clock speed of the first PCs. The easy returns in that avenue may already have been acheived.
It might be, but its pretty immature of you to assume that that is what they are thinking.
And...so? Does this announcement imply that Intel isn't researching other things?
I'm not sure this is 80 general-purpose processing cores: the article claims that there are "80 floating point cores". Clearly, the big selling points of the chip are, in Intel's view, its data transfer at 1 TB/sec, and its floating point speed at 1 TFLOP.
An 80-core chip with RAM attached directly to the processor chip, as TFA discusses, is going to have an advantage in transferring data between cores, and plus it'll probably be a lot smaller. Than 40 dual core (or 20 quad core) chips.
And I know of know game that gives you 1/6 of the total gameplay free, either. Both movies and games often have selected promotional material available as a means of drumming up interest, yes.
Who cares? No one was suggesting not doing what research can be done, and no one was asking for your sympathy for anything. I was answering your question of why people might be happy with spending money on movies and not happy with games even though the $/hour for the former was generally greater than the latter, and the latter offered more interactivity.
So that means the temperatures were that high before manufacturing and automobiles?
Yes, temperatures were this high before manufacturing and automobiles. What is concerning is not so much the current temperature, but the trend of rapid increase that has been strong since about the Industrial Revolution and appears unprecedented.
And will, if continued, in not too long reach temperatures that are, themselves, sources of grave concern. At which point, it may well be impossible to prevent catastrophic results.