Guest commentary by Figen Mekik – Grand Valley State University
“But Figen, humid air feels oppressive, heavy!” students told me, almost in unison. A very treasured moment indeed. I just got a glimpse of probably a long held misconception: water vapor is heavier than dry air. So, we took out our periodic tables and calculators, and went on to calculate the molecular weight of H2O and how it compares to that of N2 and O2 (most of the atmosphere). Happy that I corrected a major fallacy, I didn’t see the rest coming.
Apparently, there are many other sinister fallacies lurking just underneath the surface of the heavy wet air idea. One student asked “is the formula for water vapor the same as for liquid water?” and was astonished to find out that it is always H2O regardless of phase, even in ice! I said “we like to keep things simple in science” and a couple of ladies giggled “as if!”
Then another admitted that he always thought water split into H2 and O2 upon evaporation which would make wet air heavy. Another student answered him with “No way man. When water vapor condenses to liquid, the molecules get bigger which is why liquid water is heavier than vapor.” So we had a long discussion about molecular dynamics of evaporation and condensation. Also, once I helped the students realize the stark contrast between what they think they know (water vapor is heavy) and something else they know from the Weather Channel (low pressure means rain), the cognitive dissonance (the psychological tension created by conflicting knowledge) drove them to question both “bits of knowledge” and to adjust their ideas. By the end of the hour, they were saying this is SOOO weird, humid air rises. Who knew!
Here are some other common and very tenacious misconceptions:
[1] Seasons are caused by cyclical changes in Earth’s proximity to the Sun. The main causes underlying this one likely are that [a] intuitively it makes sense and [b] textbooks frequently exaggerate the eccentricity of Earth’s orbit to the extreme that such an idea is logical. The problem is this misconception is extremely popular, from kindergarten to high school physics teachers. A very confused young man once told me openly “Well, my third grade teacher told me that the Earth’s axis is tilted and that is why we get different seasons and it’s winter in the northern hemisphere, when it’s summer in the southern hemisphere. My high school earth science teacher told me during the summer we are closer to the Sun and summers are hot everywhere. Now you are saying my grade school teacher was right all along. And there is all this hype about sunspot activity being the real cause behind global warming. Since the Sun causes our seasons for whatever reason, that sounds believable to me. But you say it’s CO2 in the atmosphere causing global warming. How do I know I can trust you?”
He has a point! And it is very difficult to address the inconsistencies in his education convincingly. I could have told him about my PhD and that I am a climate scientist, but that really doesn’t have much currency in such situations. So I acknowledged that he has a valid point and devoted the next month to demonstrations and data and error margin analysis to empower the students to the point that they could understand the science for themselves. We couldn’t cover coastal geology that semester because we ran out of time, but I think it was worth it anyway.
[2] The hole in the ozone layer and atmospheric pollution (including but not limited to aerosols) cause global warming. Like the previous one, this one is also very tenacious and difficult to dispel because it is often presented this way in the media and most primary and secondary school teachers share the same fallacy. Perhaps one of the underlying faulty notions here is that the Earth receives heat from the Sun, instead of radiation. So, the thinking here is that the ozone layer shields our planet from the Sun’s harmful rays and its heat. And because there is a hole in the ozone layer, the extra heat seeps in and gets stuck under the ozone layer causing the greenhouse effect. I know, yikes!! I try to dispel this misconception by explaining that though the sun is indeed quite hot, there is all this empty space between the Sun and our planet and heat travels to Earth as infrared radiation from the sun, but the Sun’s output of infrared is only a fraction of its output as visible light. Energy from the sun mostly reaches us as visible light and ultraviolet radiation. (Minor edit to remove confusion with sensible heat and radiation. Sorry about that!).
However, the notion that global warming and ozone depletion are linked is not entirely wrong. As was discussed earlier on RealClimate (Ozone depletion and global warming), original CFC’s as well as ozone itself are powerful greenhouse gases and stratospheric cooling caused by the increase in atmospheric CO2 actually accelerates ozone loss there. Even the replacement gases to be used in lieu of CFCs may have significant greenhouse warming potential. BUT, ozone depletion (“the hole in the ozone layer”) does not cause global warming.
This discussion eventually lends its way to a discussion of aerosols (see Aerosols: the Last Frontier) and although aerosols tend to scatter or absorb incoming solar radiation (hence a warming effect), their net effect is in the direction of cooling because they have a positive influence on the nucleation of clouds which increases our planet’s albedo (ability to reflect light).
[3] The greenhouse effect and global warming are the same thing. This is another yikes!! Perhaps the root of the problem here is that the discussion of the greenhouse effect in the classroom is often tightly linked with that of global warming. It needs to be explicitly pointed out to students that without the greenhouse effect our planet’s surface would be about 30 degrees C cooler and with wild differences in temperature between night and day. Not exactly habitable. But anthropogenic global warming is caused by the human-induced increase of greenhouse gases in the atmosphere since the Industrial Revolution, particularly CO2. Most of the past changes in climate on glacial-interglacial timescales can be explained by invoking changes in solar activity and greenhouse gas concentrations in the atmosphere, sure. But the warming we have been experiencing in the last few decades cannot be explained if we do not include the effect of greenhouse gases released by human activities (see the IPCC 4th Assessment SPM, and Avery and Singer: Unstoppable Hot Air, just to name a couple).
[4] Toilets flush in opposite directions in the northern and southern hemispheres. This one is kind of a pedagogically useful misconception because although it is absolutely wrong, the idea behind it is correct and it is primarily a matter of scale. Having said that, I find the Coriolis effect to be one of the most challenging topics for students to grasp as soon as we move beyond its initial descriptive definition. There is often lots of confusion between “to the right” and “to the east” in the northern hemisphere. Plus when we add another dimension to the mix (vertical) in discussing tropical hurricanes, this becomes a serious barrier to understanding. So, I try to avoid any directional terms, like east or west as well as clockwise or counter-clockwise. Not because students are too young to know a non-digital traditional clock, but because from satellite images hurricanes look like they are rotating counter-clockwise. Really can’t argue with what the students are seeing for themselves. But if we keep the terms simple, “moving objects in the northern hemisphere are deflected to the right within the frame of reference of the moving object,” it becomes a little easier to understand, though still challenging. Another challenge here is that the Coriolis effect comes across as a force and it is difficult for students who have not had physics to distinguish between a force and a deflection (an effect).
Perhaps you are now thinking “this may be true in some university in west Michigan but surely in other, more prestigious universities the students know better!” If only this were true. A Private Universe is a video documenting lingering misconceptions among Harvard graduates about the causes behind seasons and lunar phases. The problem is misconceptions are hard to detect because most students are adept at answering questions with exactly what the teacher wants to hear and with correct terminology but without any real understanding of the science. After nine years of collegiate teaching I now know to encourage a casual “say whatever is on your mind” attitude with students. This way, I am hoping to get them to inadvertently voice their misconceptions so I can address them.
And one may be tempted to think this is solely an American problem because the American system of education has been exposed to some serious criticism of late. Again, not so! It’s a global problem. Here are some examples from a couple of quick Google searches. Greek kindergarten teachers harbor deeply rooted confusion about the “ozone hole” and the “greenhouse effect;” while Greek primary school teachers think the ozone hole causes climate change. Australian university students believe a large portion of the ozone hole is over Australia and that the high rate of skin cancer is largely caused by this hole. Junior high school students in Israel seem to understand various processes within the hydrologic cycle, but believe its beginning point is the ocean and the end point is groundwater. And some Turkish in-service physics teachers believe that the moon does not rise and set while Turkish pre-service science teachers think summer is warmer than winter because the Earth is closer to the sun in the summer time.
How about you? Take this quiz to see where you stand ;) Update: Apparently the quiz has been taken off line…
I think, however, there may be some room for improvement in the wording and explanations in this quiz because some questions are very obscure, ambiguous and Chicago-centric. I would like to know what commenters think about it.
Where do misconceptions come from? Personal experiences and intuitive understanding play a large part in fostering misconceptions, and most false notions are reinforced through school and the media. I would like to share with you this delightful and brief story of how personal experiences color the judgment of a bunch of 4th graders about the nature of heat. They have a wise science teacher who broaches the topic with a question: “can you give me an example of something that is hot?” She is expecting answers like the Sun, or a stove or maybe even Britney Spears. But the students say sweaters, hats, and coats. One says “rugs are wicked hot.” The teacher says “when I touch your sweater it doesn’t feel hot.” The students say “Ooh, it’s a matter of time. With time it can be 200 degrees!” Hmmm.. Can you blame them? They spent at least nine years in cold Massachusetts winters and their parents and teachers always told them to put on their warm clothes.
Like this example, some of the problem underlying misconceptions stems from language. “Warm clothes” implies clothes that emit heat, “greenhouse gas” suggests greenhouses are warm because of their gas content, “the rise and set of the sun” suggests the sun is moving across the sky, not the earth is rotating on its axis, and “the theory of relativity” implies all things are relative when actually the theory is based on the constancy of the speed of light.
Let’s go back to our 4th grade class to see how this very experienced teacher addressed the problem. She could just come right out and say “that’s ridiculous, you’re clothes don’t emit heat, they trap the heat your bodies emit.” That would certainly save time to cover more content; instead she decides to do something else (e.g. concept/inquiry based learning for the educators out there). She says “Tomorrow I want everyone to bring something hot from home.” The next day sweaters, scarves, hats and even a down sleeping bag arrive. The teacher puts a thermometer into each one and they wait until the next day for them to get hot on the inside. The students are convinced the down sleeping bag will be 400 degrees! They rush in the next morning and quickly check their thermometers. 68 degrees! They’re shocked. But convinced? Not a chance! They are not going to dismiss 9 years of personal experience just like that. “Cold air got in there!” says one little girl. “When I sit in the car with the windows up, it gets hot. We need to hide our clothes.” So sweaters and hats get put into drawers and closets with their thermometers snuggly in them. Another night goes by. The next day they rush in and check their thermometers again. Again 68 degrees! Except one student has 69 degrees. They all applaud. Still not convinced, after all there has been indication in the right direction! Several nights go by like this. Finally serious doubt begins to ensue. So the teacher says “I want everyone who believes clothes are hot to walk to this corner” and she points left; “and the ones who think clothes trap the heat our bodies emit to this corner” and she points right. Most of the students go to the right but three stubborn ones go to the left. Guess you will always have the denialists! But no matter what, these students experienced two things more important than heat: the scientific method in action and sometimes the way something feels is only that and not reality.
So, are misconceptions barriers to understanding or helpful pedagogical tools? That will largely depend on the individual teacher’s (professor’s) style and interests. But the important thing is to [1] challenge misconceptions, [2] demonstrate their faultiness through carefully devised experiments (ideally by the students), [3] help develop multiple working hypotheses to understand the meaning of the results of these experiments, [4] devise more experiments to test and retest each hypothesis, and [5] NEVER let a student leave the classroom with a diagnosed misconception uncorrected. And, perhaps the most effective method for eradicating misconceptions at every level is going to be investing large quantities of time, money and effort into educating primary and secondary school educators. NSF has many programs that fund such efforts, but much more effort is clearly needed on a global scale.
Disclaimer: I am not an educational psychologist. I am simply a college professor and ocean/climate scientist enjoying a rich and intense teaching career in the Geology Department at GVSU. Also, my anecdotes and all my quotations are intentionally fictionalized to protect the confidentiality of students. The ideas expressed in the quotes are amalgamations of multiple repeated ideas expressed to me from students, professors and colleagues alike since I started graduate school in 1991 at Middle East Technical University in Ankara, Turkey; and the misconceptions I mention are not unique to any of my students but are listed in over 7000 published misconceptions about science.
Re: #194
According to my understanding, you do not get a Coriolis effect on a rotating cylinder. The equations of motion seem to back this up.
Re # 197
[ interesting to see inteligent people speculate about a subject outside their expertise. ]
Seems to me everyone does that every day – no one is an authority on everything, and most people (including scientists) are authorities on only a very small body of knowledge. And educated people are curious, and want to learn more, so they frequently delve further and further from their areas of expertise; in fact, research (and teaching) usually forces you to do this.
[A little ironic to see misconseptions from scientists in a thread about laypeoples misconseptions about science.]
Ironic perhaps, but honest. I don’t hide from my students the fact that I don’t know everything. The problem is, most teachers (including college professors) must teach topics outside their narrow fields of expertise (sometimes far outside). Anyone who has ever reviewed a draft version of a textbook knows the author, supposedly an authority on the subject, often makes glaring mistakes; unfortunately, some of those mistakes make it into the final version of the book and are taught and learned in the classroom (sometimes with additional mistakes introduced).
[It might not be that strange that people have false ideas about science?]
I don’t think anyone claimed it was strange – humorous at times, certainly frustrating at times, but not strange. The question is, how do educators (or educated people) go about correcting those misconceptions? Obviously, the first step is to correct their own misconceptions (and areas of ignorance), then go about figuring out how to do a better job of teaching the subject to someone else.
And if topics such as the Coriolis effect are not clearly (or correctly) explained in textbooks, those of us who rely on those books for background knowledge will come away with misconceptions. I don’t have any real trouble understanding the mathematical explanation of the Coriolis effect – I can plug numbers into an equation and usually get the correct answer for the deflection. But, I don’t understand how the equation was derived. So, I’m still waiting for a clear and correct explanation (or analogy) that doesn’t rely on the math – that is what I need to help students understand the Coriolis effect. Simply giving them the equation doesn’t really educate them.
By the way, the analogy of a moving ball (or line drawn) on a rotating disk following an arc is flawed in my mind: First, your hand (or the ball leaving your hand) is in fixed reference plane, whereas an ocean current or rocket heading north from the equator is in the rotating reference plane of the earth. Secondly, a record player turntable turns clockwise, whereas the earth (viewed from above) turns counterclockwise. If you rotate a globe counterclockwise while drawing a straight line from the equator to the north pole, the line will inscribe an arc deflected to the left, opposite that of the Coriolis effect.
I’m open to a clear conceptual explanation or analogy of the Coriolis effect.
In defense of my questionable knowledge about the Coriolis effect, I quote here from a popular oceanography textbook, Ocean Circulation (Open University, Pergamon Press in association with The Open University, Milton Keynes, England, 1989):
Chapter 1, page 7: “…When it leaves the launcher, the missile [fired northward from the equator] is moving eastwards at the same velocity as the Earth’s surface was well as moving northwards at its firing velocity. As the missile travels north, the Earth is turning eastward beneath it. Initially, because it has the same eastwards velocity as the surface of the Earth, the missle appears to travel in a straight line. However, the eastwards velocity at the surface of the Earth is greatest at the Equator and decreases towards the poles, so as the missle travels progressively northwards, the eastwards velocity of the Earth beneath it becomes less and less. As a result, in relation to the Earth, the missile is moving not only northwards but also eastwards, at a progressively greater rate… This apparent deflection of objects that are moving over the surface of the Earth without being frictionally bound to it – be they missiles, parcels f water or parcels of air- is explained in terms of an apparent force known as the Coriolis force. ”
The book then presents a sidebar question about the Coriolis force for a hypothetical cylindrical Earth rotating about its axis. The answer at the back of the book is that “For the hypothetical cylindrical earth, the eastwards velocity of the surface would be the same whatever the distance from the poles, and so there would be no apparent deflection and no Coriolis force.”
This is how I have been explaining the Coriolis effect to biology students for over 20 years – am I explaining it correctly? Or not?
Re 202: I couldn’t have said it better. I teach 3-4 courses per semester and though all are related to earth science the topics are wide and mixed. As for me personally, I hope it is clear to all readers that I am in no way implying I know everything, and if I were I would be laughably arrogant and wrong. Wasn’t it Socrates who said the more you learn, the more you realize how ignorant you are. And that’s the way it should be for a scientist and science teacher/professor. Everyone has misconceptions and ignorance. Everyone. The point is to be honest about it and learn from it and next time teach it better. That’s why I like RealClimate so much because thanks to the efforts of the moderators, scientific ideas are open to discussion without personal ridicule.
Re 203: I am also desperately asking the same question, am I teaching this correctly? I am not all together convinced that your rocket example or the classic if we were flying in a plane problem is all that erroneous. I am probably really oversimplifying but I thought if you stick to a latitude, you don’t need to correct the route of your flight, but if you are travelling meridionally, you would have to correct your route and take into account the Coriolis effect. Is this right?
Re #198: [A fun experiment is to pull the plug and look at the vortex, stir the water in the other direction and look at the new (hopefully) vortex in the other direction.]
Except that what happens is that within a minute or two, the new vortex will slow down, stop, and restart in the original direction. Why?
203, I belive your book is correct. It doesn’t explain the coriolis force for east/west moment though.
202, good post, I have some comments though but no time.
Re: Coriolis Effect
This much is clear: understanding the Coriolis effect is not so simple!
Re: #202
I completely agree with your statements about self-knowledge of one’s own ignorance being essential for true advancement. But I disagree with your claim that if you draw a straight line (“straight” in the sense of being a geodesic on a sphere) from equator to north pole while the sphere rotates eastward underneath your pen, the arc will curve left; in fact it will curve to the right. Try it!
I derived what I believe to be a rigorous mathematical formulation of the Coriolis effect on a rotating sphere, and I’ve posted it on my blog. I get the “correct” results — deflection to the right in the northern hemisphere, to the left in the southern, and all the right proportionality factors. It’s also clear that there is no force and no real deflection; it seems to me that only the term “Coriolis effect” is strictly applicable. And I get no Coriolis effect on a spinning cylinder.
#197
In the original blog post by me and all the comments about Coriolis on this thread we keep saying deflection is to the right in the northern hemisphere, and to not confuse this with clockwise, counterclockwise, east or west. Let’s not keep repeating the same point.
Re 185.
I don’t buy that we can make up so much with conservation alone. I am also not convinced that burning biofuels such as ethanol and biodiesel are carbon neutral. In the case of ethanol you are taking a food crop such as sugar and corn that are grown anyway and using it for fuel. Ethanol crops can and frequently do cause land use changes from forest to cropland and requires fertilization to grow it and energy to maintain and manufacter it. (Not to mention that corn ethanol produces less energy than is required to create it.) Biodiesel is also process dependent in whether you get more out than you put in and depends on the feedstock if it is carbon neutral.(needs to be algae based and based around a C02 producer which we have plenty of.) All require large amounts of water to manufacture. Unfortunatley the algae method for biodiesel is still in the prototype stage and they have yet to produce the fuel in volume. Ethanol/Biofuel have more to do with energy independence that GHG reduction. (Still a valuable goal though.)
OTOH Of all the forms of solar power I find the algae biodiesel method to be the most promising for transportation.
For conservation, I do not see that in the cards that humans will reduce power consumption. The fact that you sit on a computer connected to the Internet proves that. Certain economic and politcal objectives oppose energy conservation. (From simple ones to safety lighting to goals such as business persistance during disasters.)
Regarding Coriolis force and why it matters: http://oceanmotion.org/html/background/western-boundary-currents.htm
The formation of the mid-latitude gyres in both hemispheres is also a function of the balance between the Coriolis acceleration and the pressure gradient. The result of wind blowing over water in the Northern Hemisphere is net (Ekman) transport of water to the right of the wind direction, thus when winds blow south along the California coast, the water is transported offshore and the result is strong upwelling that brings nutrients to the surface. See http://oceancurrents.rsmas.miami.edu/ocean-gyres.html , as well as http://oceanworld.tamu.edu/students/currents/currents4.htm
This is the point that Carl Wunsch has made; shutdown of the Gulf Stream is not going to happen due to a warming and freshening North Atlantic (but the formation of deep water in the Labrador sea is another issue).
Re 207 Tamino…I did try it (drawing a straight line on a globe whose surface is rotating east) and I keep getting a deflection to the left. I first tried it while drinking a cup of coffee and preparing to eat a spherical chocolate truffle candy; I had no globe handy, so I used a Sharpie ink marker and drew a line on the foil wrapper while I rotated the truffle. The line wasn’t a smooth arc because the wrapper was crinkled, but it was deflected to the left. I tried it again at home using a large rubber ball – same thing. If, after all of this, I understand the Coriolis effect correctly (confirmed by your and Fredrik’s explanations) my hand would have to rotate east with the rotating globe in order to get a rightward deflection. Yes? In fact, my hand is in a fixed frame of reference relative to the rotating globe. A little thought experiment would seem to confirm this:
Instead of drawing a continuous line on the rotating globe, I move the globe and my pen in discrete increments – each time I rotate the globe 10 degrees of longitude to the east, and move my pen 10 degrees of latitude north (in a straight line). Starting from the equator on the prime meridian, after three increments, my pen touches down at 30′ N, 30′ west. After three more increments, my pen touches down at 60′ north, 60′ west. Connecting the dots I get an arc deflected to the left.
Re 210 Ike, Thanks for the link. I also teach, albeit in a very superficial manner, about the Ekman spiral. I must confess, I understand that (the progressive deflection of a current to the right with increasing depth, with the deepest component of the current moving, in theory, at 180 degrees relative to the surface water) even less than the Coriolis effect.
http://en.wikipedia.org/wiki/Ekman_spiral
Re: #211 (Chuck Booth)
On this we agree: following the procedure you outline, the line will intersect [0 deg.Lat, 0 deg.Lon] (the starting point), [10 deg.N, 10 deg.W], [20 deg.N, 20 deg.W], [30 deg.N, 30 deg.W], etc.
That line is initially (at the equator) directed eastward. But this is not the Coriolis deflection, it’s just the initial direction of movement. Connect the dots, and the arc most assuredly *curves* to the right. The Coriolis deflection is the curvature of the line (to the right), not its initial direction (to the left).
Thanks Ike, that’s really helpful!
I know this is way too late, but way back at comment #21, Brian said
They also say that the total mass after a chemical reaction is exactly the same as it is before. This is just wrong. Chemical reactions can be exothermic or endothermic, and convert some mass to heat or vice versa.
Brian is just wrong, in chemical reactions the mass remains the same, the energy is stored in the molecules. With nuclear reactions some mass is converted to energy.
This thread reminds me of The Jaywalk All-stars on Leno. The sad thing about it is Jay Leno claims that it is very easy to find people who can’t answer simple questions like name the vice president and when was the war of 1812. But even sadder, if he asked simple science questions the vast majority of people (my guess is 80-90%) would be clueless.
Re #209: [I do not see that in the cards that humans will reduce power consumption. The fact that you sit on a computer connected to the Internet proves that.]
I don’t see how that constitutes proof. Indeed, for me it’s just the opposite. I can sit at this computer and do useful work (at least my employers seem to think so, since they keep paying me :-)) without having to spend time and energy commuting. That saves maybe 5 gallons of gas per week. Then there are other savings: on-line banking & bill paying saves paper, stamps, and trips to the bank to deposit paper checks…
Then think about the decreases in computer power consumption. Of course for years this was just less per megaflop or megabyte, masked by the fact that you kept getting more of them, but lately it’s becoming an absolute decrease. A few years ago I used a tower machine, with cooling fans that’d compete with a vacuum cleaner and a massive CRT display. Today I use a notebook & LCD display, and get better performance with much less power consumption.
That same process, getting the same or better performance from less power, could be done in many fields, from electric lights to automobiles, if anyone cared enough to bother. That it mostly hasn’t is a reflection of the fact that, thanks to the offloading of environmental costs, energy has been so cheap that people have only bothered in special cases, such as notebooks and other mobile electronics.
Re # 214 [Chemical reactions can be exothermic or endothermic, and convert some mass to heat or vice versa.]
You might want to rethink that statement.
Re: #214 (Wang Dang)
No, Brian is right. If the chemical reaction causes the emission of radiation, the mass of the chemicals left behind is reduced proportional to the energy radiated away, according toE=mc^2 — just as with nuclear reactions.
Re: Correction to #212 (Tamino)
I mistakenly said, “That line is initially (at the equator) directed eastward.” I should have said, “That line is initially (at the equator) directed northwest.”
Re 214: […when was the war of 1812.]
Sometimes those simple questions can be deceptive, though. So when was the War of 1812? Now if you give the simple answer, “1812, of course”, you’d be only about 20% right. It started in June of 1812, and lasted until January/February of 1815. (The ending date varies because its best-known engagement, the Battle of New Orleans, was actually fought after the peace treaty had been signed.) See? Not such a simple question, was it?
Re#216
Thanks Chuck, I forgot to put the quote around the paragraph from Brian’s statement in #21. The sentence you quoted is the one from Brian that I was trying to correct.
“This, my leige, is how we know the earth to be banana-shaped.”
James,
Actually computer consumption has been increasing. Case in point. CPU power consumption for say a P75 was around 8-9 Watts and required modest cooling, to processors that consume from 65 to 130 Watts on current buld processes. (There are more effiecent CPUs but they are geared to mobile and embedded applications.) Granted they get more performance per watt, the power consumption has still increased significantly. (Even your laptop uses alot more power than a model of 5-10 years ago even with greater performance per watt.) In most cases the increased power effiency in new processes are used to drive increased performance per watt, not to lower absulute power usage. (Excpetions are only mobile electronics.) Video coprocessors are even worse with current state of the art video subsystems consuming 200Watts or more. Most folks that use PCs alot also leave the things on 24/7. (Folding or SETI @Home) In any case you end up with more load on the electric grid due to PCs that require more power.
PCs are also a poor measure of effiecency as silicon fabrication is certainly not grean , and is very energy,resource, and manpower intensive to manufacture. (Takes 10-100’s of man-years to design a CPU!) Lastly all of those obselete PCs that are in the trash are not that great either.
I was mainly speaking of the Internet, which is not power effiecient at all. It requires PoPs all over the place, each with regulated and backup environment and power. (Average power for a medium size core router is 8000Watts with heat output of 8000BTU/hr and the biggest ones for big hubs can consume 13.2 Kwatts per 7 foot rack! Again you don’t just have one there are layers to good network design!) In addition most datacenter’s are under power consumption crunches as the outgrow their electricity and UPS capacities regularly due to increased power density from hosted servers. This server you use for this website is not all that cheap on power you know.(Average 1U server consumes 350-500Watts and average Datacenters have 100s to 1000s of them to be profitable) Neither for that matter is the clusters everyone uses to run simulations as they are racks of a bunch of 1U servers all of which require backup power and regulated environment on backup power.
Lastly Internet bandwidth demand growth (Internet growth about an order of mangitude every 2-3 years.) is much higher than power demand growth and will not level off, and in effect it drives a portion electricty demand growth. (Increased bandwidth capacity -> more equipment -> more fiber laid in the ground -> more cooling for COs, DCs and PoPs -> more backup UPS power -> more electric power demand -> more oil burned to lay fiber and power electronics) Increased demand is driven by all of us tapping on our keyboards, watching video, bit torrent, etc.etc.
Yes, I baited you some as we have sparred about this in the past. :)
Your job is unusual in that you get to stay put and get paid. (How nice!) I have to go all over the place to do my job even though I “telecommute” to work. They certainly would not pay me if I sat in one place all of the time! (I live in Florida and the company is in Maryland.)
I said I don’t think conservation will achieve the goals that were put forth in a previous post of 50% of a 60% reduction in carbon footprint in the US. It will do some but not near that much and this example is just one of the reasons why. We like the conveniance and comforts provided by abundant power and will not want to do without them and folks are always thinking up new ways to get electricity to do more and more tasks that we don’t want to do. (TV remote sound familiar?)
Re Brian’s comment about mass and energy:
What about the vice versa part?
There is some truth to what you say, especially if you take the points in isolation, and not as part of a whole. You get several trends working in different directions: higher efficiency because more performance per watt, yet more power demand because more people are buying computers. More internet bandwidth using more power because more people are finding more things to do on-line, but a lot of those things replacing more energy-intensive alternatives. (What’s the energy cost of say downloading a movie vs going to a video rental store vs going to an actual movie theater?) It’d take someone more knowledgeable than me to say where all those factors balance out at any given time, but I think trends are towards lower total per capita consumption.
You do need to remember that we’re still in the midst of an ongoing process. At first the emphasis was on geting the damn things to work reliably, then increasing performance regardless of energy cost. Now pure performance is giving way to power efficiency. Remember when the main selling point in computer ads was clock speed? Nowadays that’s often relegated to the fine print. Limiting power consumption has become a selling point in the commercial world.
As for conservation, a lot depends on how much incentive there is. When there’s no apparent cost associated with leaving your PC on 24/7 (or leaving outside lights on all the time, or whatever), of course a lot of people are going to do just that. Raise the perception of cost, though, and that changes behavior.
Re # 212, 218 Tamino,
Correction noted, but you’ve still lost me. I fail to see how a line connecting a series of ink marks on a globe that start at 0 lat, 0 long. and end at, say, 60 deg. N, 60 deg. W, can be described as curving to the right. If an imaginary person knowing nothing about the Coriolis effect had been standing on the prime meridan at 40 deg. N (Mediterranean coast of Spain?) and learned there was a giant pen heading north that would crush him, he would no doubt have been pleasantly surprised to find that the path of the killer pen seemed to be deflected to the west, out into the North Atlantic. (And had he known about the Coriolis effect, he might have expected the giant pen to veer east and touch down somewhere in Turkey; but, then, he likely wasn’t aware that the pen was in a fixed, external reference frame, and not rotating with the globe.)
So, where have I gone wrong?
Actually everything I said was true.
Current state of the art for CPUs is going multi-core packaging or silicon die (As an example AMD does multi-cores per die, where Intel does multi-cores per die and then multi-die per package.) with individual cores driving towards power effieciency with more performance realized with multiple cores per socket. While it is true that some higher performance individual cores are getting into the 25 – 50 watt range. With the current rage of quad core and soon 8 core chips, you get power ranges still in the 100 to 200 watt range for CPU processing. (Although the performance increases per watt.)
You need to remember that it is ALWAYS an ongoing process that never stands still! (As in all science with the expetion that digital technology grows by orders of magnitude.) The only reason that the clock speed race was called off was the fact that no one could continue to ramp the clock speed (In volume) due to the current silicon processes having high switching leakage current and high static leakage current that was inhibiting clock speed increase past the magic 3 to 4 Ghz barrier. That “barrier” is still hard to get through today and the chip companies are always trying to one up the other with performance increases so they tried a different tack with the above mentioned multi-core and each individual core getting more IPC (wider paths which also has limits.)
As far as downloading a movie or driving to watch one, that has to do with individual taste. I like going to the theater for the experience. If I paid for a “movie room” in my home it still would not be the same. Cost can regulate behavior to some degree but there will always be folks that buy everything online (Like you.) and folks like me who like to go get it. Just a matter of taste.
Re #217
tamino, thank you for the correction. I was wrong, but only because my chemistry professors lied to me many years ago.
I apologize to Brian.
http://www.newton.dep.anl.gov/askasci/chem03/chem03641.htm
Re #225, Although overall modern CPU’s are moving to ever smaller die sizes (0.65 and 0.45 micron are the current targets attained) and more cores on the same die etc I believe that low power versions are coming base around TDP (thermal envelope) of around 35 to 85 Watts. Under load they will obviously consume more but far less than previous CPU’s have.
Issues relating to graphic cards and more than one of them worry me as these technologies are now finding their way into games consoles to. Recently a 1 KW power supply was released for PC’s in order to power quad graphic cards configurations. I have a 550 Watt one in my machine.
How many playstation 3’s will be sold world wide and how many high definition LCD/Plasma TV’s ? A large LCD TC consumes over 300 Watt of power, 5x more than a 32″ CRT TV.
it is a worry.
Re: #224 (Chuck Booth)
I hope the Spaniard would hear from reliable news reports that the giant pen wasn’t heading north at all. Earthbound observers will testify that it was never heading north, it left the equator heading northwest. Space-based observers may think it’s heading north, but they’ll also say that the Spaniard is heading east, so the two are not on a collision course.
Now let the pen go all the way from equator to pole. From the viewpoint of an earthbound observer it starts out northwest, which is most definitely not in the direction of the pole. But its path curves to the right — the more so the further north it gets — as it must in order to reach the pole.
The apparent deflection is only due to the observer’s point of view; space-based observers (in an inertial frame) see the pen go straight (in a geodesic sense) from equator to pole.
Of course the best way to see this is to find a globe with latitude/longitude lines, put dots on the points [0,0], [10N,10W], [20N,20W], etc. all the way to the north pole, and connect the dots.
Re #225: [With the current rage of quad core and soon 8 core chips, you get power ranges still in the 100 to 200 watt range for CPU processing.]
While on the other hand you see more and more people who, like me, realize that they don’t need multi-core CPUs for what they do, and would rather have portability & longer battery life. The multi-core CPUs and high-end graphics processors go to a small minority.
[The only reason that the clock speed race was called off was the fact that no one could continue to ramp the clock speed…]
Not entirely. There’s also the fact that a lot of that clock speed was wasted due to memory bandwidth limitations, so your 3 GHz CPU was often stalling on memory 3 clocks out of 4. Consider that the BlueGene supercomputers were designed with 700 MHz CPUs for just these reasons, memory bandwidth & power consumption.
[As far as downloading a movie or driving to watch one, that has to do with individual taste.]
It’s individual taste only if one has both options. How many more people would be driving to movies if they couldn’t download or rent videos? Frankly, I have no idea. What I do know is that having such options makes it possible to arrange life so as to use much less energy. And at least in my case, save time and improve perceived quality of life too.
Re 229.
Actually you are wrong.
Multi-core CPU’s are now maintsream components in both desktop and laptop computers. Also 40 to 50 Watt video subsystems are also mainstream for desktops with laptops having 10 to 20 watt accelerators. (The only exception being intel integrated graphics which are very crappy and are only good for surfing, no games or opengl rendering works and use about 5-10 watts max.) Core2 Duo and Turion X2 processors are definitly the “sweet spot” for selling chips. Quad cores are supposed to hit 2008. I suppose you could get a Via C7 or some other ultraportable which only uses a watt or so to pound out your email though. (Good luck using that with power hungry apps though.) In any case it is what is sold the most, not what you think you like. I also value portability as I have to travel all around to do my job, and my corporate provided cheapie “mainstream” IBM laptop has a core2 duo in it and gets about 8 hours with two battery packs which is good enough for me.
Mainstream in called “mainstream” for the express purpose as that is where the most volume and sales dollars are. (Like the guy said above think about all of those big LCD TVs, xbox 360’s and PS3s!) If what is sold draws more power in absolute terms than what was sold before that constitutes increased power demand even though the computers are much more powerful and have more MIPS/Watt.
You are also mixing diferent types of applications, from PCs to supercomputers which have radically different design philosphies. IBM picked an intial 700Mhz clock speed for blue gene as it a seriously multi-core device using altivec instructions for vector math which is easily parrelizable BTW. They needed a certain Giga/Flop per core target figure that maximized performance/watt and they got it with power pc440 at 700Mhz. It has 1024 CPUs (2048 cores) with a NUMA memory acrchitecture which totaly renders irrelevant memory bottlenecks as each processor module has 4MB L3 cache and each quad of cores has a path to 512MB of local dram. (That causes memory bandwidth to scale linearly per each quad of cpu cores, total GB/sec = X You can bet to compete their next “blue gene” will have vastly greater power, memory bandwidth, power draw, cpu clockspeed, and number of CPUs.
Just for a counterpoint the Cray Redstorm uses 2,000 to 120,000 cores (96 per cabinet) single/double/quad core AMD opteron cpus. Running from 1.6 to 3Ghz, with each CPU core providing 10 to 12GB of memory bandwidth, scaling from 960GB/sec memory bandwidth/cabinet. The sandia red storm uses 2MW of power for 10,000 2.0Ghz CPUs for 40Tflops, 55TB/s memeory bandwidth, with an eye to triple that performance near term.
In any case modern CPU designs are increasingly effective at hiding the latency to main memory by using large caches, seperate on die connections to memeory (AMD), doubling memory bus width , prefetching data to cache before it is needed, doing out of order reads on the memory bus and busting writes to maximize BW utilization, etc. etc.
Also the clock speed war was killed becuase Intel could not ramp the Pentium 4 design beyond 3-4 Ghz in volume without turning them into good space heaters. (They still had a good crack at it though. The P4 design was intended to reach 10 to 20 Ghz and ran into process leakage problems not a desire to lower power requirements. Same with AMD, once they could no longer scale the clockspeeds of their designs at will they were the first to emphasize performance/watt and power consumption as a means to compete as they were beaten by intel on the clock speed front. (just now getting to 3 Ghz.)
Your last paragraph fine. If staying at home increases your quality of life, well more power to you. If I am not mistaken you leave in the moutains a good ways from town? (You said that in a post somewhere.) I don’t and it would not take near as much time and energy for me to go out as it would for you.
I have already said your job is definitly not the norm as most employers want to see your face, or move you all around (like me) in order to pay you!
[[Correction noted, but you’ve still lost me. I fail to see how a line connecting a series of ink marks on a globe that start at 0 lat, 0 long. and end at, say, 60 deg. N, 60 deg. W, can be described as curving to the right. ]]
If you’re facing west.
Jim wrote in #221: “I said I don’t think conservation will achieve the goals that were put forth in a previous post of 50% of a 60% reduction in carbon footprint in the US.”
Jim, did you read the section of the freely downloadable American Solar Energy Society report that projected the carbon emission reductions that could be achieved with application of existing efficiency technologies?
If you did read it, can you cite specific flaws in, or offer specific rebuttals to, its projections?
If you did not read it, why not? Are you so confident in your own beliefs and assumptions that you already know that it must be wrong, so there is no need to read it?
Re 232.
No I haven’t read it. I am going too as soon as I can, and quit trying to bait me into saying something snotty. That kind of language only works on the unwashed after all. In any case projections are not reality. (For exmaple, look at the projected power savings for the new DST in the US that saved nothing.)
I will delve into electronics a bit as I can’t help myself as I am an EE after all. :) In one of my past jobs (I have went from the electric power industry to, lightweight ASIC/VHDL/Verilog, to heavy circuit design, to something a lot easier, Internet backbone architect, and I got to keep my hair :) ). I designed portable eqiupment for where there was profitablity in the extra time and effort required to shave milliwatts out of every part of the system design. For the control eqiupment and everyday stuff, all that was desired was for the equipment to perform reliably (Case in point, TTL is still used everywhere) , and the “management” would not sanction the extra time and effort. (And material costs as well as low power ICs, and basic components come with a premium as they had to do the same thing for the IC as we have to do with the circuit.) That extra effort is not a small undertaking, it is a significant portion of overall design costs, and will not be performed without some expectation of a return on the investment. Lastly, in some cases a reduction in power is not possible if the driving design goal is a performance target. (ie. MIPS in a certain amount of floor space.)
But I do know this, us science folks don’t know much about the people we design things for. Lots of toys, gadgets and specificly software and OS are written or designed by technical people for technical people, not for the mass of folks that are simply wired different. (Vista or Linux ring a bell?) I am also going by the current facts of the real power demand growth and power shortfalls we have already, and a proliferation of ever more power hungry devices to use that power. (eg. entertainment equipment such as xbox360, ps3, and PCs.) As long as people do not wish to do a menial task they will design a new machine to perform it for them. That is just basic human nature and there is ample proof all around you as we are all guilty of it. (Cars, Dishwashers, calculators, vacuum cleaners, washing machines, remote controls, etc.etc.)
Jim
Re # 232 I’m looking at a globe (actually, a large rubber ball) with a series of ink marks heading north and west (up and left) from the ball’s equator. Although my pen moved straight north as it laid down the line of dots, an observer on my “globe” would see the pen heading northwest, as Tamino pointed out. By my estimation, the pen will eventually touch down at 90 deg. north, 90 deg. west (I need a real globe with lat. and long. lines to verify this).
By the way, in case this point has gotten lost, I originally used this example to point out what I consider to be a flawed analogy for demonstrating the Coriolis effect (likewise, inscribing a line on a record rotating clockwise on a record turntable is also a flawed analogy, I think- I have seen both used in books or on websites). A proper analogy should account for the fact that the “pen” must rotate “east” in the same frame of reference as the earth, not reside in a fixed (intertial) frame of reference as the earth rotates under it.
I’ll have to go out and buy a globe and check this out – it’s hard guessing at lat. and long. lines on a green rubber ball.
Re #230: [Actually you are wrong.]
Well, it happens, but about what, exactly?
[Multi-core CPU’s are now maintsream components in both desktop and laptop computers.]
Did I say otherwise? What I said is that lots of people find they don’t need the most powerful machine available, like many people I work with, who’ve gone to notebooks from desktop machines. As for relative power consumption, given the lower level of heat & fan noise, I have to think the laptop is using much less power. Humm… OK, doing a “cat /proc/acpi/battery/BATT0/info” tells me that each battery pack in my machine has a design capacity of 59.2 Watt-hours. I can work about 2-3 hours per pack, so I must be using about 20-30 watts on average, no? Since my last tower machine had IIRC a 350 watt power supply, it would seem I’ve cut my power consumption considerably. Factor in the difference between 20″ CRT and LCD, too.
[Like the guy said above think about all of those big LCD TVs…]
Fine, but compare their power consumption to an older TV (even going back to vacuum tubes), or an LCD screen with a CRT. Once again, the technology works both ways: cut power going to solid state & LCD, increase it going to bigger screens. What’s the net effect?
You’re also wrong on a lot of details of the BlueGene architecture (about half the code I write these days is targeted to BlueGene), and I’m pretty sure you’ll find the next generation using even less power per CPU than the current one.
[I have already said your job is definitly not the norm as most employers want to see your face, or move you all around (like me) in order to pay you!]
You miss my point there, though. Add up all the miles not driven by the current telecommuting population, and I bet it more than cancels out the increase in energy use in internet infrastructure. As for you and your employer, sure, they may want to see your face now, but how badly will they want to go on seeing it when doing so puts them over their mandatory CO2 targets?
Once again, there are two factors at work. One is the possibility of achieving various energy reductions, the other is giving people the incentive to use them. Given the first, people may or may not choose to use it. Without it, they can’t.
lets do that. My argument is that the net affect of all of these advances is power consumption growth, not a net reduction, you have yet to say anything that refutes that.
[[Multi-core CPU’s are now maintsream components in both desktop and laptop computers.]
[Did I say otherwise? What I said is that lots of people find they don’t need the most powerful machine available, like many people I work with, who’ve gone to notebooks from desktop machines]]
In a sense yes. Stating that simply going to notebooks from desktops saves power. The answer is it depends. Not to long ago my PC had a 120Watt power supply, now laptops have 120Watt power supplies. Yet current mainstream desktops use about 300Watts. So if you switch from a new computer to an new laptop then yes you save power. However switching from an older computer to a new laptop the power savings are not so clear. In any case successive iterations of the SAME devices do not use less power which was the main thrust of my argument.
My point being mainstream, is that these multi-core machines are not the fastest available, they are what is on special at walmart/bestbuy and as such get sold the most, hence power consumption grows.
[Once again, the technology works both ways: cut power going to solid state & LCD, increase it going to bigger screens. What’s the net effect?]
The net effect is to use more power. Some examples:
Let’s use LANL blue gene verus Sandia RedStorm verus ENIAC as a good example:
Blue Gene Power input 26.7KW max 7 Kw Min per rack
with a requirement of 8 ton cooling per rack. (8 tons is approx 30Kw power draw.) So each rack is requiring 56.7 KW max power. LANL had a final buildout of 64 cabinets with a total aggregate power draw of 56.4Kw x 64 = 3.609 MW. Red Storm at Sandia uses 24 cabinets each that draw 22Kw Max with similar cooling requirements for an aggregate of approx 3MW. ENIAC pulled 150Kwatts with no AC units.(All power was Peak power draw.) Granted the computing power of the newer systems utterly dwarfs ENIAC, my point has been made.
Even when we recover effieciecy in indivdual components those are more than made up with an increased performance target for each successive generation.
Another example. My 6 year old 32 CRT Inch Zenith HDTV pulled 231 Watts, My samsung 42 inch HDTV monitor pulls 228 Watts. (A wash if you ask me.) Both are average models that costed about a 1200-1300 bucks new.
Another example
PS2 drew about 40-45 watts, PS3 draws around 200Watts, with total peak power 380Watts. Xbox360 uses around 160Watts, with 500 Watts peak. (I know that they never get near peak for consumer systems but gotta put them in there for accuracy.)
I have made my point using easily verifiable facts. (Except my tvs of course.) Performance/watt always increases yet the net affect is a steady increase in power draw as now the designers use more and more of that performance. (In almost all electronics that is true.) For mobiles old laptops drew less than new ones do. Get an old one and see for yourself. The current trends are clear, newer devices do more or are more powerful, have more performance/watt, and draw more power than the previous generation. Thus creating greater power demand. Going back to tubes is irrelevant as that stuff died away in the early 70s, and as AGW folks everywhere say it is the las 30 years that show the warming trend, so it is the last 30 years of silicon electronics that also matters.
[You’re also wrong on a lot of details of the BlueGene architecture]
Really? Even though I got those details from IBM?
What details did I get wrong?
See for yourself:
One Data sheet from IBM
http://www03.ibm.com/servers/deepcomputing/pdf/bluegene_spec_sheet.pdf
And the watered down one on wikipedia
http://en.wikipedia.org/wiki/Blue_Gene
[I’m pretty sure you’ll find the next generation using even less power per CPU than the current one.]
Maybe if that is there design goal than yes. If IBM decides to use Power 5 or Power 6 CPUs in Blue Gene, then you are wrong. And as of right now they have to do something as their competition can do 64bit CPUs and they use 32 bit CPUs which means power 5, PowerPC 970, or Power 6. If they use Cell, it still pulls more power than the current PPC 440s in there. Make no mistake, without altivec for vector math programs, those 440’s would not be in there as their other performance areas are very weak.(Integer ops)
Just becuase you write code doesn’t mean squat. I have written assembly and high level code for a lot of stuff too, and if you code to the metal with power assembly language then yes, you SHOULD know what you are talking about, but if you code in fortran or some other high level language, you have an abstraction layer that does the details for you and you are writing a logic algorithm, not moving bits or understanding (Or needing to understand) the physical hardware that well. Lots of programmers don’t know what mov aex bex means!
You totally missed my point. Writing computer code is about the only thing you can do in isolation from your employer. If you have to interact with people in any meaningful way, you can’t do that effectively with telecommuting. (Nothing beats a look in the eye and a good handshake!) In any case the telecommuting popluation is very very very very small.
205, “Except that what happens is that within a minute or two, the new vortex will slow down, stop, and restart in the original direction. Why?”
This doesn’t happens for me when I have tried the experiment. Did it happen for you?
207, “This much is clear: understanding the Coriolis effect is not so simple!”
I would say that it is easy to derive the coriolis acceleration mathematicaly. Easier than you have done in my opinion.
To intutivility understand the details of it is not that simple but I think the explaination in Chuck’s book is clear and simple enough. It is a quantiative reasoning though.
” It’s also clear that there is no force and no real deflection; it seems to me that only the term “Coriolis effect” is strictly applicable. And I get no Coriolis effect on a spinning cylinder.”
Yes, the force doesn’t exist if we look at the earth from space, but do we do that? We live on the earth and thus rotate togheter with it and the force exist in our world. You feel the centrifugal force when you sit on a merry go round. It is nothing imaginary with it. I have performed the throwing ball on a rotation disk experiment as seen in a movie above and the coriolis force definitely existed in my world then.
Why the missconceptions and errors? I belive the main reason is that people try to explain difficult subjects for people without adequate knoweledge, often a lack of mathematical skill. Trying to explain a phenomena in an intuative way without using math in short. This is very good if someone managed to do it but it often fail to explain the phenomena and it sometimes lead to errors.
Aerodynamics and flight theory is a good example of this happens. It exist a well known theory but it involves difficult concepts and mathematics. People have try to explain why an airplane fly without the mathematics and the result is often a disaster like for example the equal transit time claim and claims that either of Bernuelli or Newton is correct about flight when the truths is that they actually are the same theory.
You can found a lot of books claiming that they explain flight in the correct best way but they often include large errors and incorrect logic.
The reason is that the real theory is difficult so we try to explain it instead.
The coriolis is simular though actually easier. People dont understand the math and try to explain it without math and the result is not always good.
Re: #238 (Frederick)
I’m sure you’ve seen far simpler derivations of the Coriolis effect in 3-D space, or on a flat 2-D surface. But the complication of movement constrained to the surface of a sphere adds a few layers of complexity to the “traditional” derivation.
Certainly the mathematical framework in which I chose to work is far more difficult. But if that framework is already familiar then the equations of motion simply “fall right out” of the formalism. So I’d agree that the mathematical framework is far more complicated, but once that’s mastered the derivation of the answer to this problem is far simpler. That seems to me to be a common theme in physics; as the math gets more advanced, the physical derivation gets simpler.
Still, I admit I used mathematics not familiar to many people, so my derivation isn’t very accessible. I was just trying to get the correct answer.
I disagree. You don’t feel a centrifugal force at all; what you feel is the centripetal force required to make you orbit around the center of the merry-go-round rather than move in a straight line.
As for the ball thrown by a rotating observer, your perception of the Coriolis “force” is entirely imaginary. The word “force” has a very precise meaning: it produces acceleration in an inertial frame. If we allow any reference frame at all to define force, then I can make the “force” on any object, at any moment, anything I want it to be.
Re: #239
I agree that attempting to explain essentially mathematical phenomena in non-mathematical terms can be a source of misconceptions and errors. Attempts to explain the theory of relativity without mathematics illustrate the extreme difficulty of doing so.
But I would emphasize that another source is the imprecise use of terminology. Words like “force” and “acceleration” have very precise meaning in physics, but when we misuse them to describe apparent forces and accelerations which are really only due to the reference frame of the observer, it can lead to a great deal of misconception and error.
Re #236: [Not to long ago my PC had a 120Watt power supply, now laptops have 120Watt power supplies.]
Yet can run for several hours off a battery holding 60-80 Wh? I think maybe you’re confusing peak draw with average power consumption. Sure, if I test a fairly large simulation on a laptop, the CPU will kick up its clock speed, the cooling fans will come on, and I’ll be drawing a lot of power. The thing is, I don’t do that all that often. Most of the time I’m just looking at the screen, or typing, and the machine is drawing a small fraction of its max power. This wasn’t true of older machines: it was the need for power conservation in mobile computing that created that ability, which then moves into other machines.
[What details did I get wrong?]
To start with: [it a seriously multi-core device using altivec instructions for vector math which is easily parrelizable BTW.]
The BlueGene has no more vector math capability than the Pentium/AMD processor in your PC. The FPUs in each core can process two 64-bit doubles simultaneously (Intel can do 2 doubles or 4 floats), but whether that capability is usable depends on the code.
[It has 1024 CPUs (2048 cores)…]
Depends on the size of the machine. The one I mostly use has 4K CPUs, BG/W I think is 56K, the LLNL machine a full 128K.
[…with a NUMA memory acrchitecture which totaly renders irrelevant memory bottlenecks as each processor module has 4MB L3 cache and each quad of cores has a path to 512MB of local dram.]
Not at all. BlueGene is essentially a message-passing architecture, with each CPU having its own memory. It’s a bit more complicated, in that CPUs are organized in pairs, each pair having access to 1 GByte. In coprocessor mode, one CPU does computation, the other handles communication tasks, sharing the memory. In the other mode, each CPU has access to half the memory, and handles its own computations.
Except for the communication mode, the memory access and its bottlenecks are just like those in a single PC – in fact, you can look at BG as just a big, optimized Beowulf cluster. That and power consumption are the reasons it runs a 700 MHz clock. It’s also the reason why, if you do serious profiling & performance analysis on a large problem (too big to fit into cache) you’ll find that no matter how fast your CPU, execution speed is pretty much limited by memory access times.
[Lots of programmers don’t know what mov aex bex means!]
I guess I’m one of them. Try MOV EAX, EBX :-)
“But the complication of movement constrained to the surface of a sphere adds a few layers of complexity to the “traditional” derivation.”
It is just to constrain the equation to the surface. No large step. I am familiar with your notation. I was talking more to most of the people that probably dont understand your notation. “where gamma are the Christoffel symbols for the space” is going to make 99.9 % of the population look as question marks.
About if the forces exist or not. You feel a force an a merry go round or in an airplane doing a loop. It must be included if you want to make calculations in the acceleration frame as is often done. But all this is about semantics because we both talk about the completely same thing.
“The word “force” has a very precise meaning: it produces acceleration in an inertial frame. If we allow any reference frame at all to define force, then I can make the “force” on any object, at any moment, anything I want it to be.”
I have to disagree here. The word force is used for the coriolis and centrifugal force in mechanics. The [tex]C(dot(q)q)q [tex] term in lagrangian mechanics is always (as I have seen) called centrifugal and coriolis forces. I am working with wind turbines and people talk about centrifugal forces in the blade all the time.
We live on earth and most people are never going to leave the earth. So why go outside the reference frame of the eart to perform calculations?
test [tex]$C(dot(q)q)q $[tex]
http://en.wikipedia.org/wiki/Fictitious_force
Good article about the subject.
How do you use tex notation in posts?
[mov aex bex]
That was at 1 in the morning after a very long day! it would work though just got to get creative with register renaming!
[The BlueGene has no more vector math capability than the Pentium/AMD processor in your PC. The FPUs in each core can process two 64-bit doubles simultaneously (Intel can do 2 doubles or 4 floats), but whether that capability is usable depends on the code.]
Not true. The original vanilla 440 does not even have an FPU, Blue gene CPUs possess a special FPU grafted onto the CPU core via the APU interface(FPU2) that is specialized to handle math and SIMD to accelerate througput. Altivec is just an implementation of SIMD I was trying not confuse anyone by using a well known trade-name. I don’t write code for altivec or FPU math for blue gene so I don’t know if those instructions are the same.
In any case, yes Intel/AMD also have SIMD (SSE/2), the register scarcity and stack based FPU archtectures hamper their performance considerably, whereas the PowerPC does not have to lug around the x86 ISA and as such its performance in FPU calcs is much greater.
The 1024 cpu figure was per rack as the machine size varies with budget. It was still not wrong.
So much wisdom and such an obvious flaw. Of course proximity to the sun can’t cause seasons otherwise north and south would have summer at the same time. You can force water to rotate the other way going down a drain buy swirling it with your hand befor pulling the plug so no cosmic force is behind that one. However the little rant about humid air implies that ice is lighter than air since you say water vapor is lighter than air and water expands when it freezes therefore ice would be the lightest of all. A water molecule is lighter than an oxygen molecule or nitrogen or CO2. However humid air consists of dry air PLUS water vapor and since air is very rare fitting the water molecules in between the various air molecules is no problem. The two combined in a given volume must outweigh the individual components. In much the same way a gallon of sea water is heavier than a gallon of fresh water simply because it has more “stuff” in a given volume. Finally the students expressed the sentiment that humid air “feels” heavy and it does. This is tied more to density than weight. Oil is lighter than water but because of its’ much higher viscosity it feels much thicker than water and the fact that oil floats would seem counter-intuitive. This reminds me of a global warming denier I heard who said it was impossible for melting ice to raise sea levels since a glass full of ice didn’t over-flow when the ice melted. He ignored the fact the melting ice is currently in glaciers and ice caps on land and would flow TO the oceans when it melted.
If we take two air masses with the same number of molecules in each, a dry air sample and a “wet” air sample, then because the water molecule is light compared to N2 and O2, the wet air sample will have less weight.
Russ, you talk with such an authority. The problem is that you are incorrect. Humid air at a given pressure is lighter compared to dry air at the same pressure. You can read it in any meteorology book.
Re #245: [That was at 1 in the morning after a very long day!]
I thought as much :-) I have days like that too.
[…Blue gene CPUs possess a special FPU grafted onto the CPU core via the APU interface(FPU2) that is specialized to handle math and SIMD to accelerate througput.]
I don’t know how special it is. I had the impression it was pretty much a stock design from the PPC family, but that sort of thing isn’t really my area of expertise. To get back at least to first-order off-topicness, though, what I do know (’cause I’ve written & profiled a lot of code) is that both suffer from memory bandwidth issues. Most of the time, you can’t pull in data from memory fast enough to keep even a 1 GHz CPU busy.
The processor speed thing is like cars, in a way. Sure, you may have a turbocharged 440 Hemi (or whatever, I’m not up on the lingo) under your hood, and maybe it has a top end of 200 mph or so, but just how often do you get a chance to go that fast?
It depends on the workload. If you write programs that use large datasets (like you do.) then yes the proccessor is DRAM BW limited. If the dataset is small enough to fit in cache or it is a typical branching application, then the cache architectures of current CPUs effectively can and do hide the latency of main memory and the acute scarcity of memory BW. Modern cache’s get a 90 to 95% hit ratio and they are striving for more. Just can’t cache 1GB yet though! :)
The PowerPC440 used in blue gene has a relatively slow FSB to main memory (Espcially compared with today’s most modern proccessors.) It ain’t bad at all, it is 128bit 5.5GB/sec bus. IBM covers this up using a tiered cache architecture that culimunates in a really good 4MB L3 edram cache (30 cycle latency to l2/l1 cache, 22GB/s.) However with datasets larger than 4MB the BW becomes constrained by the 5.5GB main memory bus to that node’s DRAM bank. Here is a good block diagram of the CPUs per node.
http://www.research.ibm.com/journal/rd/492/gara5.gif
Modern consumer processors have DRAM memory buses that are also 128 bits wide with 10-12 GB/Sec throughput which helps but does not solve bw issues. Progress is frustratingly slow on this front I agree. BTW that night was hell, my 15 month old son had a middle ear infection at the time. Whew! :)