[...] this is essentially the supply-side argument the U.S. Republican Party has been advocating for the past 40 years, that giving tax breaks to the producers increases economic activity and boosts the net wealth of everyone.
There's a huge distinction here though: leaving cash in the hands of producers gives them the freedom to invest that cash wherever they feel it is most beneficial, where subsidies force investment in a particular area that the government has selected (perhaps based on good intentions, but often based on less virtuous motives). This is a fundamental difference: are you trying to exploit local knowledge of where investment will be most beneficial, or are you centrally planning these investments?
I will also defer the analysis of how these approaches work in practice to others, but I think it's wrong to lump them together.
Republican presidents certainly don't have a very good track record for dealing with deficits, but the much-hyped surplus under Clinton was largely due to (1) him being forced to work with a Republican congress and (2) being lucky enough to be president during the Internet bubble (and not necessarily in that order). So the Clinton surplus is really a bipartisan anomaly, and not compelling evidence that Democrats are any better.
Bulldozers wont have on-die graphics like these Llano (Bobcat) CPU's until mid to late 2012 at the earliest.
True.
Responded too quickly... the "Bulldozers wont have on-die graphics [...] until [...] 2012" is true (this is the Trinity part, and was demo'd a couple of weeks ago, but it was announced that it won't be ready for production until 2012, I don't remember what was said about when in 2012).
The "Llano (Bobcat) CPU's" part is not true, Llano is most definitely not Bobcat.
Bulldozers wont have on-die graphics like these Llano (Bobcat) CPU's until mid to late 2012 at the earliest.
True.
What should be noted and what isnt well understood is that these "APU's" coming out from AMD are all Bobcat chips.
Not true. The E-series and C-series parts released in January (Ontario/Zacate) are Bobcat chips, built on TSMC 40nm process. The big deal with Llano (A-series) is that it's not Bobcat, it's an enhanced Phenom-derived core, built on GlobalFoundries new 32nm process. There is no such thing as a 32nm Bobcat at this point in time.
"AMD's performance target for Bobcat was 90% of the performance of K8 at the same clock speed"
Note the key clause *at the same clock speed*. To turn around and complain that a 1.6 GHz Bobcat isn't 90% of the performance of any of a set of desktop processors running anywhere from 2.7 to 3.6 GHz just shows that you didn't bother to understand the initial claim.
My thoughts exactly... the key word in the GP is "unconstitutional" (repeated three times). This is exactly the point of the original article: what do you think would happen if the DNS was under control of an organization that was not subject to a constitution that embodied a right to freedom of speech?
The article stresses the role of Bush administration policies without mentioning that many of these same policies were originated by the Clinton administration and were very much supported by Congress as well. Sort of like reading an article blaming Nixon for the Vietnam War without mentioning Kennedy or Johnson.
To be fair, the coverage is more balanced if you look at the whole series of which the article is part, but unless there's an article that focuses solely on the failure of Congressional oversight or one that discusses the origins of these policies in the Clinton administration then it's still not very even-handed. (I looked at the titles of the other articles and didn't see any that obviously fell into those categories, but maybe they're in there somewhere.)
I am not a Time's worshiper by any means but why is that people insist on measuring it against some Platonic ideal?
The issue is that the Times itself continues to pretend to ascribe to the journalistic ideal of objectiveness, and despite its decline it still has a wide enough readership to be influential. Plus, in the context of this discussion, I believe they would emphasize that objectivity when trying to claim superiority over newer media like blogs. If they just came right out and said "OK, we're biased, if you want the full story you should read some conservative blogs and split the difference" then there wouldn't be grounds for complaint.
More recently, there's also their refusal to acknowledge any
potential conflict of interest in their reporting on Caroline Kennedy's attempts to get herself appointed to the Senate.
I'm not saying that the NY Times (or old media in general) doesn't have a useful role to play, but if you think it's the role of impartial presenter of facts, that horse has already left the barn...
Gosh. You make it sound like AMD is being accused of a crime. Not a crime, but depending on your opinion of Intel, it could easily be taken as an insult...
The point I was making is that, even if there are only a few former Intel leaders at AMD, the fact remains that AMD was formed to compete against Intel on Intel's own turf, x86 compatibility. AMD's been around since 1969, well before x86, so to say they were "formed" for that purpose isn't right. That is where they've largely ended up though.
You cannot come up with a processor like that if all you think about is how to maintain a slight me-too superiority over Intel's technology. AMD must climb up the vision pole so they can see what's beyond Intel's backyard. I won't disagree with that. However I think where AMD has succeeded in the past it has been because they have done far more than "me too" with respect to Intel, within the narrower confines of commercially viable systems. And I think some of that is because they've hired non-Intel engineers from companies like DEC and IBM that have a much broader "system" view than just a "microprocessor" view. That's why I objected so strenuously to your initial claim.
So overall I don't disagree with your point that Intel and AMD are largely in the same market and as such are in the same boat as well. But I think your "monkey-see-monkey-do" characterization of AMD overlooks their record of innovations with respect to Intel (HyperTransport, x86-64, integrated memory controllers), and some of the factual details you're using to back that up (like being staffed with lots of former Intel executives, and being formed to compete against Intel) are just not true.
Many of their executives (e.g., Dirk Meyer) and lead engineers came from Intel and they only see the world through Intel glasses.
That's a pretty serious mischaracterization. Though Dirk Meyer worked at Intel for 3 years early in his career, he spent the next 9 years at DEC working on the early Alpha CPUs (see here). Many of AMD's other current & former top engineers on the CPU side are ex-DEC people as well (Mike Uhler, Rich Witek, Jim Keller in the past, etc.). IBM is the other main source of top AMD engineers (Rich Oehler, Chuck Moore, until recently Phil Hester, etc.).
There are some top folks from Intel too but they're definitely in the minority.
Notice that there are no Intel people mentioned here.
No, without a demand for these advances, competition would exist only to lower prices, but because this demand exists, the competition also includes innovation. If AMD weren't in the running, some other company or companies would be. Hurray for the market being properly represented.
What you're missing here is the concept of "barriers to entry". Between the legal hurdle of Intel's x86 implementation patents, the cost of the hundreds (or thousands) of engineers required to complete a modern microprocessor design, and the capital cost of manufacturing, it's not like a random person with a better idea can just go whip out a CPU and compete against Intel. AMD is one of the few companies that can play in this space. Most (maybe all) of the few others that have this capability (IBM, maybe Nvidia, ??) are constrained because of their dependence on Intel as a supplier and/or partner.
I think it's pretty clear from pre-Opteron days (have people forgotten those so quickly?) that without AMD we'd all be porting our software to Itanium in order to break the 4GB barrier.
If you're interested in the details, check out US patent application 20040156640 here.
It's brutally long (51 pages!) but provides a lot more details.
Basically each node has N (or N-1) receivers spaced slightly apart, along with a single transmitter. These receivers & transmitters are all in the same plane opposite a mirror. Every node can transmit simultaneously. The different angles at which the transmitters hit the mirror cause the beams to focus on a different receiver within each node's array.
Yes, it's not necessarily practical; as the article says understatedly, "Naturally, this algorithm performs much more computation per character than the straightforward serial algorithm".
It seems Hillis & Steele's original purpose was akin to the tone of this thread: just defending their highly parallel machine in the face of naysayers who had criticisms like "Well, what about parsing? Let's see you do *that* in parallel! Huh? Huh?".
Of course, in their case the naysayers seem to have been right:-). I'll leave the comparisons with the present situation to others (for now).
Starting to feel like a curmudgeon with posts like this, but it's already been done a long time ago. Hillis and Steele presented an algorithm for parsing in parallel in logarithmic time on the Connection Machine. The article is called "Data Parallel Algorithms", Communications of the ACM, December 1986.
For a group of characters (substring) in the middle of the file, you can locally build a table that maps whatever the incoming parser state is (at the beginning of the substring) to what the corresponding outgoing state would be at then end, and then that table lets you process the whole substring in unit time. I like to think of it as the parsing equivalent of a carry-lookahead adder. Probably best to read the article if you're curious.
True! Didn't mean to imply otherwise. I should have been clearer:
The use of the term "wheel of reincarnation" to refer to this phenomenon came from [...]
And of course the main contribution of this paper was the recognition of that phenomenon, not just the appropriation of a catchy phrase to describe it.
Yup, the idea is pushing 30 years old now, and came out of the earliest work on graphics processors. The term "wheel of reincarnation" came from "On the Design of Display Processors", T.H. Myer and I. E. Sutherland, Communications of the ACM, Vol 11, No. 6, June 1968.
Hard to say what you saw... certainly similar technology was around 15 years ago, but in a much much larger package.
Your mention of Stevie Wonder brings up what I think is an interesting little factoid: he's the link that got Kurzweil into music synthesis. He was an early user of Kurzweil's reading machines, and at some point complained to Kurzweil about the state of music synthesis technology, which inspired the whole Kurzweil line of music synthesizers.
Seems like he was parking in their parking lot, and refused to go even after they repeatedly asked him to leave, so I don't see why this isn't an open-and-shut trespassing case. (Of course, IANAL.)
I'll let others comment on the mention of "erotic services".
Wow, even for/. that's quite a load of stereotyping. As a former Vancouver resident, I think it's actually a pretty great place. Yea, there are neighborhoods (and residents) that are kind of seamy, but that shouldn't be too surprising, as it's a decent-sized city in its own right (pop. ~150,000). There are lots of very nice, honest, pleasant people who do not deserve to be tarred with your brush.
The truth is that many of the people in Vancouver are there because in spite of Oregon's high income tax that state cannot seem to get its act together and fund a decent public school system; so perhaps if you want to find some gross overgeneralization to apply to Vancouver residents, it would be that they are people that care enough about their children's education to put up with the often nasty commute over the I-5 bridge.
And for this crowd, it might be worth noting that Vancouver is the North American headquarters of iRiver (or however it's supposed to be capitalized), as well as having some other less cool but equally high-tech companies: a major HP printing division site, Kyocera, Wafertech ("by far, the largest semiconductor integrated circuit foundry in the United States", according to their web site), Xiotech, etc.
Are you retarded? Future versions of windows will run just fine on good ol' IA-32.
Don't be so sure... I recently attended a talk by Eric Traut, former Connectix CTO, now "Director, Virtual Machine Technologies" at Microsoft, in which he said that their future VMM (hypervisor) that at some post-Longhorn point will underlie all new versions of Windows will require both virtualization (Vanderpool/Pacifica) and 64-bit hardware support. (My guess is that the former is what matters most to them, but since there are practically* no 32-bit CPUs with virtualization support there's not much point in supporting that option.)
*I can't think of any off the top of my head, but if they are out there, there certainly aren't a significant number of them.
Note that patients need to have had sight in the past for this device to work. The visual cortex doesn't develop in people that were born blind, so their brain doesn't know what to do with these inputs. (Like in the movie "At First Sight".)
Reminds me of an old SNL Weekend Update line (Kevin Nealon, I think... paraphrased from memory):
"If you would like a transcript of tonight's broadcast, get a pen and write down everything I say really fast."
Calm down... I'm not disagreeing with the fallacy the original poster points out, just his or her rush to judgment. It's one thing to point out that UCLA data does not prove a national trend, and quite another to assert that UCLA has concocted this story "instead of admitting that the quality of their CS courses are dropping". The former is pedantic but true, while the latter is an equally unjustified conclusion and just snotty to boot.
The point of my posting is that it's not just a UCLA phenomenon, and a little bit of extra effort in looking at the article's references would have shown that (or if you're too lazy for that, you can take my word for it).
You simply can't take statistics from one university and assume that they're not indicative of a universal trend either. I teach computer engineering at a major public university in the midwestern US, and we are seeing trends exactly like UCLA. If you follow the link in TFA to the Taulbee survey, which encompasses all of North America, you'll see that the data there is consistent with UCLA's findings.
2) DP Matrix-Matrix multiplies. IBM added DP support to their VMX set for Cell (though at 10% the execution rate), check.
[...] ...clearly Cell is meant as a supercomputer first and a PS3 second.
I think you've refuted your own argument there: double precision floating point performance is critical for true supercomputing. (In supercomputing circles DP and SP are often referred to as "full precision" and "half precision", respectively, which should give you a better idea of how they view things.)
In contrast, SP is plenty of accuracy for things like rendering and game physics, since (very loosely speaking) as long as you're within a fraction of a pixel of the right answer you don't need any more accuracy.
I'd say the Cell architecture is very well suited for supercomputing as well as gaming, but the announced Cell implementation appears to me to be clearly targeted at the PS3. They'll have to come out with a "Cell HPC Edition" that has much better DP performance before they take over supercomputing. Not that I don't expect that they're working on that as we speak...
Slashdot Mirror
[...] this is essentially the supply-side argument the U.S. Republican Party has been advocating for the past 40 years, that giving tax breaks to the producers increases economic activity and boosts the net wealth of everyone.
There's a huge distinction here though: leaving cash in the hands of producers gives them the freedom to invest that cash wherever they feel it is most beneficial, where subsidies force investment in a particular area that the government has selected (perhaps based on good intentions, but often based on less virtuous motives). This is a fundamental difference: are you trying to exploit local knowledge of where investment will be most beneficial, or are you centrally planning these investments?
I will also defer the analysis of how these approaches work in practice to others, but I think it's wrong to lump them together.
Republican presidents certainly don't have a very good track record for dealing with deficits, but the much-hyped surplus under Clinton was largely due to (1) him being forced to work with a Republican congress and (2) being lucky enough to be president during the Internet bubble (and not necessarily in that order). So the Clinton surplus is really a bipartisan anomaly, and not compelling evidence that Democrats are any better.
Basically they all suck.
Responded too quickly... the "Bulldozers wont have on-die graphics [...] until [...] 2012" is true (this is the Trinity part, and was demo'd a couple of weeks ago, but it was announced that it won't be ready for production until 2012, I don't remember what was said about when in 2012). The "Llano (Bobcat) CPU's" part is not true, Llano is most definitely not Bobcat.
True.
Not true. The E-series and C-series parts released in January (Ontario/Zacate) are Bobcat chips, built on TSMC 40nm process. The big deal with Llano (A-series) is that it's not Bobcat, it's an enhanced Phenom-derived core, built on GlobalFoundries new 32nm process. There is no such thing as a 32nm Bobcat at this point in time.
"AMD's performance target for Bobcat was 90% of the performance of K8 at the same clock speed"
Note the key clause *at the same clock speed*. To turn around and complain that a 1.6 GHz Bobcat isn't 90% of the performance of any of a set of desktop processors running anywhere from 2.7 to 3.6 GHz just shows that you didn't bother to understand the initial claim.
My thoughts exactly... the key word in the GP is "unconstitutional" (repeated three times). This is exactly the point of the original article: what do you think would happen if the DNS was under control of an organization that was not subject to a constitution that embodied a right to freedom of speech?
What context?
The article stresses the role of Bush administration policies without mentioning that many of these same policies were originated by the Clinton administration and were very much supported by Congress as well. Sort of like reading an article blaming Nixon for the Vietnam War without mentioning Kennedy or Johnson.
To be fair, the coverage is more balanced if you look at the whole series of which the article is part, but unless there's an article that focuses solely on the failure of Congressional oversight or one that discusses the origins of these policies in the Clinton administration then it's still not very even-handed. (I looked at the titles of the other articles and didn't see any that obviously fell into those categories, but maybe they're in there somewhere.)
I am not a Time's worshiper by any means but why is that people insist on measuring it against some Platonic ideal?
The issue is that the Times itself continues to pretend to ascribe to the journalistic ideal of objectiveness, and despite its decline it still has a wide enough readership to be influential. Plus, in the context of this discussion, I believe they would emphasize that objectivity when trying to claim superiority over newer media like blogs. If they just came right out and said "OK, we're biased, if you want the full story you should read some conservative blogs and split the difference" then there wouldn't be grounds for complaint.
That's still no excuse for a single story to fail so blatantly in providing adequate context.
And this isn't the only instance of one-sided reporting; the contrast of the pieces that happily repeated rumors and insinuations regarding McCain's relationship with a former lobbyist and McCain's wife's past with their refusal to pursue or even acknowledge John Edwards's affair until he did because they don't want to dignify rumors and insinuations is pretty telling in my opinion.
More recently, there's also their refusal to acknowledge any potential conflict of interest in their reporting on Caroline Kennedy's attempts to get herself appointed to the Senate.
I'm not saying that the NY Times (or old media in general) doesn't have a useful role to play, but if you think it's the role of impartial presenter of facts, that horse has already left the barn...
So overall I don't disagree with your point that Intel and AMD are largely in the same market and as such are in the same boat as well. But I think your "monkey-see-monkey-do" characterization of AMD overlooks their record of innovations with respect to Intel (HyperTransport, x86-64, integrated memory controllers), and some of the factual details you're using to back that up (like being staffed with lots of former Intel executives, and being formed to compete against Intel) are just not true.
That's a pretty serious mischaracterization. Though Dirk Meyer worked at Intel for 3 years early in his career, he spent the next 9 years at DEC working on the early Alpha CPUs (see here). Many of AMD's other current & former top engineers on the CPU side are ex-DEC people as well (Mike Uhler, Rich Witek, Jim Keller in the past, etc.). IBM is the other main source of top AMD engineers (Rich Oehler, Chuck Moore, until recently Phil Hester, etc.).
There are some top folks from Intel too but they're definitely in the minority.
Notice that there are no Intel people mentioned here.
What you're missing here is the concept of "barriers to entry". Between the legal hurdle of Intel's x86 implementation patents, the cost of the hundreds (or thousands) of engineers required to complete a modern microprocessor design, and the capital cost of manufacturing, it's not like a random person with a better idea can just go whip out a CPU and compete against Intel. AMD is one of the few companies that can play in this space. Most (maybe all) of the few others that have this capability (IBM, maybe Nvidia, ??) are constrained because of their dependence on Intel as a supplier and/or partner.
I think it's pretty clear from pre-Opteron days (have people forgotten those so quickly?) that without AMD we'd all be porting our software to Itanium in order to break the 4GB barrier.
It's brutally long (51 pages!) but provides a lot more details.
Basically each node has N (or N-1) receivers spaced slightly apart, along with a single transmitter. These receivers & transmitters are all in the same plane opposite a mirror. Every node can transmit simultaneously. The different angles at which the transmitters hit the mirror cause the beams to focus on a different receiver within each node's array.
At least that's one possible embodiment...
Yes, it's not necessarily practical; as the article says understatedly, "Naturally, this algorithm performs much more computation per character than the straightforward serial algorithm".
:-). I'll leave the comparisons with the present situation to others (for now).
It seems Hillis & Steele's original purpose was akin to the tone of this thread: just defending their highly parallel machine in the face of naysayers who had criticisms like "Well, what about parsing? Let's see you do *that* in parallel! Huh? Huh?".
Of course, in their case the naysayers seem to have been right
Looks like if you don't have ACM digital library access and don't feel like trudging to a library you can find a copy here: http://cva.stanford.edu/classes/cs99s/papers/hilli s-steele-data-parallel-algorithms.pdf
For a group of characters (substring) in the middle of the file, you can locally build a table that maps whatever the incoming parser state is (at the beginning of the substring) to what the corresponding outgoing state would be at then end, and then that table lets you process the whole substring in unit time. I like to think of it as the parsing equivalent of a carry-lookahead adder. Probably best to read the article if you're curious.
The use of the term "wheel of reincarnation" to refer to this phenomenon came from [...]
And of course the main contribution of this paper was the recognition of that phenomenon, not just the appropriation of a catchy phrase to describe it.
http://www.cap-lore.com/Hardware/Wheel.html
Your mention of Stevie Wonder brings up what I think is an interesting little factoid: he's the link that got Kurzweil into music synthesis. He was an early user of Kurzweil's reading machines, and at some point complained to Kurzweil about the state of music synthesis technology, which inspired the whole Kurzweil line of music synthesizers.
More details here: http://www.kurzweilai.net/bios/bio0005.html?printa ble=1
Seems like he was parking in their parking lot, and refused to go even after they repeatedly asked him to leave, so I don't see why this isn't an open-and-shut trespassing case. (Of course, IANAL.)
I'll let others comment on the mention of "erotic services".
The truth is that many of the people in Vancouver are there because in spite of Oregon's high income tax that state cannot seem to get its act together and fund a decent public school system; so perhaps if you want to find some gross overgeneralization to apply to Vancouver residents, it would be that they are people that care enough about their children's education to put up with the often nasty commute over the I-5 bridge.
And for this crowd, it might be worth noting that Vancouver is the North American headquarters of iRiver (or however it's supposed to be capitalized), as well as having some other less cool but equally high-tech companies: a major HP printing division site, Kyocera, Wafertech ("by far, the largest semiconductor integrated circuit foundry in the United States", according to their web site), Xiotech, etc.
*I can't think of any off the top of my head, but if they are out there, there certainly aren't a significant number of them.
Pretty cool nonetheless.
Reminds me of an old SNL Weekend Update line (Kevin Nealon, I think... paraphrased from memory): "If you would like a transcript of tonight's broadcast, get a pen and write down everything I say really fast."
The point of my posting is that it's not just a UCLA phenomenon, and a little bit of extra effort in looking at the article's references would have shown that (or if you're too lazy for that, you can take my word for it).
You simply can't take statistics from one university and assume that they're not indicative of a universal trend either. I teach computer engineering at a major public university in the midwestern US, and we are seeing trends exactly like UCLA. If you follow the link in TFA to the Taulbee survey, which encompasses all of North America, you'll see that the data there is consistent with UCLA's findings.
In contrast, SP is plenty of accuracy for things like rendering and game physics, since (very loosely speaking) as long as you're within a fraction of a pixel of the right answer you don't need any more accuracy.
I'd say the Cell architecture is very well suited for supercomputing as well as gaming, but the announced Cell implementation appears to me to be clearly targeted at the PS3. They'll have to come out with a "Cell HPC Edition" that has much better DP performance before they take over supercomputing. Not that I don't expect that they're working on that as we speak...