I can probably send you some test code (same for anyone else who asks), but I'll have to check with my advisor first. The smallest I've made the test code is a bit under 300 lines. It's been run on Alpha 21264 EV67, Athlon C, Athlon XP, P4, and P-III, and one other Pentium-ish platform. At least two (I believe it's actually three) profilers have been run to find the bottleneck; it appears to be the floating point unit stalling for data.
Here are the timings. Note that these are just via "time" on GNU/Linux or a wall clock on Windows (or something -- I didn't do the Windows tests).
P4 dual Xeon 1.7GHz/gcc: 82 seconds P3 1000/msvc: 18 seconds Athlon C 600/msvc: 2 seconds P3 1000/msvc, using floats and sse:
2 seconds Alpha 667/gcc: 2 seconds Athlon XP 1900+ 0.88 seconds
I guess the Athlon's clock was closer to the P4's clock than I recalled in my original post. Either way, the slowdown on the Pentiums can be easily seen.
Incidently, you can get a nice new dual Alpha 21264 667 4u rackmount with 4GB ram and 18GB scsi (64 bit) for = $14,000 these days. With educational discount, you can buy a Compaq ES40 (with single cpu to start) for $20K. I have no idea what the used 21164 machines are selling for these days.
I don't have the same motivation for 64 bit machines (I need them for cycle servers with big memory), but I'm just as anxious for a commoditized 64 bit platform to emerge.
I, for one, am hoping to replace our Alphas with cpus from the AMD Hammer series. We're about to buy a bunch of P4-based machines despite the problems we've had with certain tight loops in scientific code performing 80 times slower than a similarly clocked Athlon (according to Athlon advertised "speed", not actual clock). No, I'm not exaggerating, and this has been verified independently -- the P4 cpu has some huge weak spots that really suck if you hit them. If Hammer were out and working properly, we probably wouldn't buy the P4 machines to hold us over.
We need 64 bit machines to accomodate massive memory for our research. I'm really hoping the Hammer can provide a relatively inexpensive and *commoditized* 64 bit platform for us to work on, compared to existing 64 bit (workstation/server) platforms. And I want it yesterday. Actually, I want it last year.
I have no idea what the editors or submitter meant, of course.
I believe that cpu caches are typically built from static ram. Cray has some beautiful "old" machines which used exclusively static ram, IIRC.
While IBM mentioned the application of this new transistor to static ram, I would guess it would be useful for any fast silicon with transistors (i.e. most =-). I'd appreciate any replies which knew more about this either way.
I risking death at the hands of Danish friends for posting this joke...
"The Danes aren't stupid. They aren't perverse, or ignorant, or backward. They live on a flat sandbar with few mineral resources in a cool sea,..."
I think you've just proved that the Danes are perverse, ignorant, *and* backward. They should have moved en masse to Iceland when they had the chance. Or taken over North America before the crummy Puritans arrived.;-)
Google is almost certainly *not* the best place to get this sort of data. The population that uses google is almost certainly not representative of the large population of computer users. And how do they count the accesses? Per access or per "machine"?
I suspect these numbers indicate that Windows users are generally complete nincompoops that require 91 times as many google searches to get the same data as a GNU/Linux user gets in 1 search. Mac users, bless their souls, rate much higher at 1/4 the intelligence of a GNU/Linux user.
Your humorous comment reminds me of a funny anecdote.
Some scientist found that photographic plates in his laboratory were being unintentionally exposed while still in drawers. He or she concluded that the lab was no place to store photographic plates. Later, a scientist named Roentgen noticed the same thing, tried to explain it, and discovered x-rays.
Suggesting that the "sole task" of evolution is to follow the path of least resistance is misleading. Entities in an evolutionary system are trying to survive long enough to reproduce. I find it difficult to make a serious, detailed connection between this goal and taking the path of least resistance.
We should probably try to avoid 'humanizing' evolution (...evollution knows...,...evolution does not cheat, It's sole task...). This only makes the theories commonly associated with evolution harder to understand.
I can't be sure, but I think you missed the parent post's point. You put a big solar array in some wide-open area near the equator, isolate the hydrogen and oxygen there, and ship the hydrogen and oxygen to coastal areas like Bellinham, WA that only have two seasons (cold rain and warm rain).
Even if solar power is a bad example, the idea is intersting. Distribution of energy is as much of a problem as power generation. For example, losses on distribution power lines are significant.
They suggest in the article, somewhat subtly, that the real problem is *chaning* fuel cells while on the plane. At that point, you might have some hydrogen mixing with oxygen if the packaging wasn't perfect.
Whomever wrote about the statistical difference was being either 1) pedantic or 2) a jerk. That's an inclusive-or, by the way.
Using historical data you can make estimates of probability. Because we don't have a *perfect* model for the events being described, we'll never know the desired p.d.f. and never have any "statistical truth" about the matter. If you wanted to make the complainer happy, you'd have to choose a statistical model, then devise and justify a fitting procedure for that model. Then you could argue that, under your model, some event had a certain probability. At this point, the complainer would tell you that your model was stupid, and that you were stupid, etc.
I'm glad you wrote all that so I won't have to. I'd like to add that MSWord was called the "Word Processor from Hello" in my old mathematics department. The equation editor is horrible. Other design flaws for large documents with lots of graphs, charts, tables, and equations is that Word stores everything in one file (last I checked). Maybe that doesn't matter on modern computers, but on a 486 you couldn't get above a few pages of graphs and stuff before things crawled.
I really wish the chem guys would get into LaTeX. I think there are some chem packages available these days. I'd love to see all the sciences using and supporting LaTeX, because there's nothing better for scientific papers. I can compile, view, and edit my LaTeX journal papers comfortably on my iPAQ. There are several good semi-WYSIWYG front ends, like LyX and Scientific Word (or Scientific Workplace, with Maple integration).
Heh, you've caught me. I've never studied with constructionists, and in fact have never heard many good words said about that philosophy. Philosophically, I'm definitely an Existence & Uniqueness guy.
I think that invoking Plato isn't really necessary to justify the non-constructionist view. It's always seemed clear to me that there is something constraining mathematical thought, and that something seems *fairly* universal among humans. For instance, anyone not bothered by certain consequences of the Axiom of Choice is clearly a martian and probably not from our Universe.;-)
These unspoken and unspeakable constraints are what drives the notion that we're discovering something and not creating it. It's not that our ideas exist, its that our new ideas are forced to come into agreement with existing principles in order to maintain consistency in mathematics.
Computer science is not entirely based upon constructive mathematics, as near as I can tell. The first example coming to mind is complexity theory. I believe a lot of complexity theory depends on existence proofs which do not provide a method of constructing the necessary objects. While computers are Turing machines with finite resources and useful algorithms run in finite time, computer hardware is merely a part of computer science. The humans in computer science, though, are often classically-trained mathematicians. In fact, I've never met a constructionist face-to-face, and only if you describe yourself as one have I ever encountered a constructionist.
Any mathematical system which limited itself to constructive techniques would be less rich than modern mathematics. While the constructionist approach is a useful paradigm, I see it as only a part of mathematical practice.
At any rate, I don't want you or anyone to take the creation vs. discovery description I wrote too literally. As you probalby noticed, I wasn't particularly careful and didn't define most of my terms. It was meant as an informal summary of a constrasting viewpoint, a viewpoint which allows that ideas may be new but reality already exists.
Agreed! I'd also *love* to see laws forcing EULAs to not lie about people's Constitutional rights. Same for FBI warnings on videotapes, and copyright notifications in books. Oops, way off topic now.
It is common to hire an inspector when buying a house. I suppose that might not hold for "new" houses, but the "new" house market does not dominate the general house market. So in general, most people *do* hire a professional to inspect the house, and part of that inspection examines the house's construction and past care. For instance, signs of rotting are likely to be found.
The best thing that could come of any such lawsuits is an end to puffery and ambigous claims. The US is a pansey when it comes to advertising laws. I'd love to see companies forced to market what they actually produce, rather than marketing what they *wished* they produced or what people *wished* they had produced.
The mathematicians I have met (I'm one of them) by-and-large feel that new math ideas are *discovered* instaed of *created*. The distinction is important. Truth and algorithms already exist, we're just trying to *find* them and sort through the crap. Just because no human has previously written down some piece of truth or an algorithm before you do, doesn't mean you invented that truth or algorithm.
We're all standing on the shoulders of reality, trying to decode what we see. John Carmack's comment about struggling with a problem in order to understand it seems very much in line with this view, and very much inline with the academic research process. Academics don't get research done just by sitting around, trying to be creative. We do research by repeatedly struggling with a problem until we figure out which defects in our brain prevented earlier understanding.
Red Hat bought Cygnus some time ago. Cygnus is a company that specializes in porting gcc to new platforms, among other things. Red Hat is likely to be one of several major forces behind bringing the GNU toolchain to Hammer.
I disagree. The first half sucked because *both* teams stunk up the place. Ronaldo and Ronaldinho couldn't keep their feet on the ball. I've never seen a Brazilian World Cup team play so sloppily. It wasn't until the first Brazilian goal that Brazil started playing well.
Germany did a decent job in the second half. Unfortunately, nobody told them that the goal is the big white thing with three poles and a net. As a result, the Germans kept kicking the ball over the endline anywhere *but* the goal. How did Germany manage to accrue something like 13 corner kicks and not score a single goal?
My conclusion is the both teams were to blame for the lackluster first half, and not Germany's style.
A second response: but you have a good point -- if they're using the same mod chips that other enthusiasts are using, then you might have more than 100 cars driving 10,000 miles. That said, you might (or might not =-) lack the systematic synthetic tests like crash tests.
To prevent confusion, by "*thoroughly* test" I would like to mean drive 1,000,000 miles in mixed city/highway conditions and a variety of weather, coupled with synthetic testing for extremes and corner-cases. That said, I probably have to live with something like testing 100 cars for 10,000 miles, at least one or two of which had special testing done. Even then I might be pressing my luck.
Not to disparage the enthusiast, but it seems very likely they have fewer than 100 cars with the same computer. =-)
Not to excuse the other guy, but I expect the real issue is whether you want to drive a car run by source code that hasn't been *thoroughly* tested. I can't be sure that the auto manufacturers *thoroughly* test their code, either. However, the manufacturer has more time between writing the code and releasing it than the mechanic did.
On the other hand if auto manufacturers were run the same way software manufacturers seem to be, I'd be less confident that any serious testing had been done at all. =-)
I think his point was that Democrats fail to take full advantage of Republican scandals, while Republicans have shown prowess at turning even little Democrat scandals into intolerable, torturous, nation-destroying scandals.
At least Watergate was interesting (as are the things you mention). The Lewinsky and Whitewater stuff went on for ages, and was dull. The Republicans got bored of their own scandal, and some eventually asked Newt to go away.
"Reagan's alleged crimes were done for the expressed purpose of helping those fighting for their freedom against communism."
When was the last time you saw America do something for the pure, unadultered sake of helping someone fight for "freedom", "against communism"? Our government doesn't give a damn about communism -- we attack *ANY AND ALL* governments that don't bow down to our whims. It's as if our government is the Microsoft of governments.
"Congress had no authority to prevent the commander in chief from giving military arms to whomever he wishes."
Bullshit. We have laws, as well as policies, governing how we distribute and sell arms. The commander-in-chief is not God.
"That power is vested in the President in the constitution."
You'll have to quote a section and paragraph number for me. I don't recall reading this in our Constitution. If you can show it to me, I'll back down immediately. And then I'll work on an amendment.
"Nixon. Some two bit espionage done by an underling, that very likely occured without his knowledge. What was he guilty of? covering up a stupid move by someone acting without his authority."
Oh, really? Maybe this is exactly what we'd like to know. Maybe this is why we're still interested in Watergate.
Heh, I suppose Clinton could have said "I don't recall" every time he was asked about something. That would have been *so* much better! =-)
More seriously, there's a big *practical* difference in allowing someone to withhold info about extramarital affairs, versus allowing someone to stonewall on arms deals with rebels in Iran. I contend that, practically speaking, a lie is not a lie is not a lie.
And heck, Congress lies to us (and stonewalls, as far as that goes) all the time. We should feel vindicated when someone lies to them!;-)
Slashdot Mirror
I can probably send you some test code (same for anyone else who asks), but I'll have to check with my advisor first. The smallest I've made the test code is a bit under 300 lines. It's been run on Alpha 21264 EV67, Athlon C, Athlon XP, P4, and P-III, and one other Pentium-ish platform. At least two (I believe it's actually three) profilers have been run to find the bottleneck; it appears to be the floating point unit stalling for data.
Here are the timings. Note that these are just via "time" on GNU/Linux or a wall clock on Windows (or something -- I didn't do the Windows tests).
P4 dual Xeon 1.7GHz/gcc: 82 seconds
P3 1000/msvc: 18 seconds
Athlon C 600/msvc: 2 seconds
P3 1000/msvc, using floats and sse:
2 seconds
Alpha 667/gcc: 2 seconds
Athlon XP 1900+ 0.88 seconds
I guess the Athlon's clock was closer to the P4's clock than I recalled in my original post. Either way, the slowdown on the Pentiums can be easily seen.
-Paul Komarek
Incidently, you can get a nice new dual Alpha 21264 667 4u rackmount with 4GB ram and 18GB scsi (64 bit) for = $14,000 these days. With educational discount, you can buy a Compaq ES40 (with single cpu to start) for $20K. I have no idea what the used 21164 machines are selling for these days.
I don't have the same motivation for 64 bit machines (I need them for cycle servers with big memory), but I'm just as anxious for a commoditized 64 bit platform to emerge.
-Paul Komarek
I, for one, am hoping to replace our Alphas with cpus from the AMD Hammer series. We're about to buy a bunch of P4-based machines despite the problems we've had with certain tight loops in scientific code performing 80 times slower than a similarly clocked Athlon (according to Athlon advertised "speed", not actual clock). No, I'm not exaggerating, and this has been verified independently -- the P4 cpu has some huge weak spots that really suck if you hit them. If Hammer were out and working properly, we probably wouldn't buy the P4 machines to hold us over.
We need 64 bit machines to accomodate massive memory for our research. I'm really hoping the Hammer can provide a relatively inexpensive and *commoditized* 64 bit platform for us to work on, compared to existing 64 bit (workstation/server) platforms. And I want it yesterday. Actually, I want it last year.
I have no idea what the editors or submitter meant, of course.
-Paul Komarek
I believe that cpu caches are typically built from static ram. Cray has some beautiful "old" machines which used exclusively static ram, IIRC.
While IBM mentioned the application of this new transistor to static ram, I would guess it would be useful for any fast silicon with transistors (i.e. most =-). I'd appreciate any replies which knew more about this either way.
-Paul Komarek
I risking death at the hands of Danish friends for posting this joke...
;-)
"The Danes aren't stupid. They aren't perverse, or ignorant, or backward. They live on a flat sandbar with few mineral resources in a cool sea,..."
I think you've just proved that the Danes are perverse, ignorant, *and* backward. They should have moved en masse to Iceland when they had the chance. Or taken over North America before the crummy Puritans arrived.
-Paul Komarek
Google is almost certainly *not* the best place to get this sort of data. The population that uses google is almost certainly not representative of the large population of computer users. And how do they count the accesses? Per access or per "machine"?
I suspect these numbers indicate that Windows users are generally complete nincompoops that require 91 times as many google searches to get the same data as a GNU/Linux user gets in 1 search. Mac users, bless their souls, rate much higher at 1/4 the intelligence of a GNU/Linux user.
-Paul Komarek
Your humorous comment reminds me of a funny anecdote.
Some scientist found that photographic plates in his laboratory were being unintentionally exposed while still in drawers. He or she concluded that the lab was no place to store photographic plates. Later, a scientist named Roentgen noticed the same thing, tried to explain it, and discovered x-rays.
-Paul
Suggesting that the "sole task" of evolution is to follow the path of least resistance is misleading. Entities in an evolutionary system are trying to survive long enough to reproduce. I find it difficult to make a serious, detailed connection between this goal and taking the path of least resistance.
...evolution does not cheat, It's sole task...). This only makes the theories commonly associated with evolution harder to understand.
We should probably try to avoid 'humanizing' evolution (...evollution knows...,
-Paul Komarek
I can't be sure, but I think you missed the parent post's point. You put a big solar array in some wide-open area near the equator, isolate the hydrogen and oxygen there, and ship the hydrogen and oxygen to coastal areas like Bellinham, WA that only have two seasons (cold rain and warm rain).
Even if solar power is a bad example, the idea is intersting. Distribution of energy is as much of a problem as power generation. For example, losses on distribution power lines are significant.
-Paul Komarek
-Paul Komarek
They suggest in the article, somewhat subtly, that the real problem is *chaning* fuel cells while on the plane. At that point, you might have some hydrogen mixing with oxygen if the packaging wasn't perfect.
-Paul
Whomever wrote about the statistical difference was being either 1) pedantic or 2) a jerk. That's an inclusive-or, by the way.
Using historical data you can make estimates of probability. Because we don't have a *perfect* model for the events being described, we'll never know the desired p.d.f. and never have any "statistical truth" about the matter. If you wanted to make the complainer happy, you'd have to choose a statistical model, then devise and justify a fitting procedure for that model. Then you could argue that, under your model, some event had a certain probability. At this point, the complainer would tell you that your model was stupid, and that you were stupid, etc.
-Paul Komarek
I'm glad you wrote all that so I won't have to. I'd like to add that MSWord was called the "Word Processor from Hello" in my old mathematics department. The equation editor is horrible. Other design flaws for large documents with lots of graphs, charts, tables, and equations is that Word stores everything in one file (last I checked). Maybe that doesn't matter on modern computers, but on a 486 you couldn't get above a few pages of graphs and stuff before things crawled.
I really wish the chem guys would get into LaTeX. I think there are some chem packages available these days. I'd love to see all the sciences using and supporting LaTeX, because there's nothing better for scientific papers. I can compile, view, and edit my LaTeX journal papers comfortably on my iPAQ. There are several good semi-WYSIWYG front ends, like LyX and Scientific Word (or Scientific Workplace, with Maple integration).
-Paul Komarek
Heh, you've caught me. I've never studied with constructionists, and in fact have never heard many good words said about that philosophy. Philosophically, I'm definitely an Existence & Uniqueness guy.
;-)
I think that invoking Plato isn't really necessary to justify the non-constructionist view. It's always seemed clear to me that there is something constraining mathematical thought, and that something seems *fairly* universal among humans. For instance, anyone not bothered by certain consequences of the Axiom of Choice is clearly a martian and probably not from our Universe.
These unspoken and unspeakable constraints are what drives the notion that we're discovering something and not creating it. It's not that our ideas exist, its that our new ideas are forced to come into agreement with existing principles in order to maintain consistency in mathematics.
Computer science is not entirely based upon constructive mathematics, as near as I can tell. The first example coming to mind is complexity theory. I believe a lot of complexity theory depends on existence proofs which do not provide a method of constructing the necessary objects. While computers are Turing machines with finite resources and useful algorithms run in finite time, computer hardware is merely a part of computer science. The humans in computer science, though, are often classically-trained mathematicians. In fact, I've never met a constructionist face-to-face, and only if you describe yourself as one have I ever encountered a constructionist.
Any mathematical system which limited itself to constructive techniques would be less rich than modern mathematics. While the constructionist approach is a useful paradigm, I see it as only a part of mathematical practice.
At any rate, I don't want you or anyone to take the creation vs. discovery description I wrote too literally. As you probalby noticed, I wasn't particularly careful and didn't define most of my terms. It was meant as an informal summary of a constrasting viewpoint, a viewpoint which allows that ideas may be new but reality already exists.
-Paul Komarek
Agreed! I'd also *love* to see laws forcing EULAs to not lie about people's Constitutional rights. Same for FBI warnings on videotapes, and copyright notifications in books. Oops, way off topic now.
-Paul Komarek
It is common to hire an inspector when buying a house. I suppose that might not hold for "new" houses, but the "new" house market does not dominate the general house market. So in general, most people *do* hire a professional to inspect the house, and part of that inspection examines the house's construction and past care. For instance, signs of rotting are likely to be found.
-Paul Komarek
The best thing that could come of any such lawsuits is an end to puffery and ambigous claims. The US is a pansey when it comes to advertising laws. I'd love to see companies forced to market what they actually produce, rather than marketing what they *wished* they produced or what people *wished* they had produced.
-Paul Komarek
To add another perspective:
The mathematicians I have met (I'm one of them) by-and-large feel that new math ideas are *discovered* instaed of *created*. The distinction is important. Truth and algorithms already exist, we're just trying to *find* them and sort through the crap. Just because no human has previously written down some piece of truth or an algorithm before you do, doesn't mean you invented that truth or algorithm.
We're all standing on the shoulders of reality, trying to decode what we see. John Carmack's comment about struggling with a problem in order to understand it seems very much in line with this view, and very much inline with the academic research process. Academics don't get research done just by sitting around, trying to be creative. We do research by repeatedly struggling with a problem until we figure out which defects in our brain prevented earlier understanding.
-Paul Komarek
Red Hat bought Cygnus some time ago. Cygnus is a company that specializes in porting gcc to new platforms, among other things. Red Hat is likely to be one of several major forces behind bringing the GNU toolchain to Hammer.
-Paul Komarek
I disagree. The first half sucked because *both* teams stunk up the place. Ronaldo and Ronaldinho couldn't keep their feet on the ball. I've never seen a Brazilian World Cup team play so sloppily. It wasn't until the first Brazilian goal that Brazil started playing well.
Germany did a decent job in the second half. Unfortunately, nobody told them that the goal is the big white thing with three poles and a net. As a result, the Germans kept kicking the ball over the endline anywhere *but* the goal. How did Germany manage to accrue something like 13 corner kicks and not score a single goal?
My conclusion is the both teams were to blame for the lackluster first half, and not Germany's style.
-Paul
A second response: but you have a good point -- if they're using the same mod chips that other enthusiasts are using, then you might have more than 100 cars driving 10,000 miles. That said, you might (or might not =-) lack the systematic synthetic tests like crash tests.
-Paul Komarek
To prevent confusion, by "*thoroughly* test" I would like to mean drive 1,000,000 miles in mixed city/highway conditions and a variety of weather, coupled with synthetic testing for extremes and corner-cases. That said, I probably have to live with something like testing 100 cars for 10,000 miles, at least one or two of which had special testing done. Even then I might be pressing my luck.
Not to disparage the enthusiast, but it seems very likely they have fewer than 100 cars with the same computer. =-)
-Paul Komarek
Not to excuse the other guy, but I expect the real issue is whether you want to drive a car run by source code that hasn't been *thoroughly* tested. I can't be sure that the auto manufacturers *thoroughly* test their code, either. However, the manufacturer has more time between writing the code and releasing it than the mechanic did.
On the other hand if auto manufacturers were run the same way software manufacturers seem to be, I'd be less confident that any serious testing had been done at all. =-)
-Paul komarek
I think his point was that Democrats fail to take full advantage of Republican scandals, while Republicans have shown prowess at turning even little Democrat scandals into intolerable, torturous, nation-destroying scandals.
At least Watergate was interesting (as are the things you mention). The Lewinsky and Whitewater stuff went on for ages, and was dull. The Republicans got bored of their own scandal, and some eventually asked Newt to go away.
-Paul Komarek
"Reagan's alleged crimes were done for the expressed purpose of helping those fighting for their freedom against communism."
When was the last time you saw America do something for the pure, unadultered sake of helping someone fight for "freedom", "against communism"? Our government doesn't give a damn about communism -- we attack *ANY AND ALL* governments that don't bow down to our whims. It's as if our government is the Microsoft of governments.
"Congress had no authority to prevent the commander in chief from giving military arms to whomever he wishes."
Bullshit. We have laws, as well as policies, governing how we distribute and sell arms. The commander-in-chief is not God.
"That power is vested in the President in the constitution."
You'll have to quote a section and paragraph number for me. I don't recall reading this in our Constitution. If you can show it to me, I'll back down immediately. And then I'll work on an amendment.
"Nixon. Some two bit espionage done by an underling, that very likely occured without his knowledge. What was he guilty of? covering up a stupid move by someone acting without his authority."
Oh, really? Maybe this is exactly what we'd like to know. Maybe this is why we're still interested in Watergate.
-Paul Komarek
Heh, I suppose Clinton could have said "I don't recall" every time he was asked about something. That would have been *so* much better! =-)
;-)
More seriously, there's a big *practical* difference in allowing someone to withhold info about extramarital affairs, versus allowing someone to stonewall on arms deals with rebels in Iran. I contend that, practically speaking, a lie is not a lie is not a lie.
And heck, Congress lies to us (and stonewalls, as far as that goes) all the time. We should feel vindicated when someone lies to them!
-Paul Komarek