The article cites use of a "high-k dielectric" and a "depleted substrate".
In english, this means using a different material for insulating layers and tweaking the doping of the substrate. A refinement, but hardly a breakthrough.
A couple of points about this puzzle me:
Doesn't Si-on-I make the substrate less relevant?
You could still call the channel material a substrate, and doping it might still do something, but it sounds like they're working with a bulk-silicon technique here. I'd thought that everyone and their dog was moving to silicon-on-insulator for capacitance reasons.
I suppose if you left the substrate undoped (depleted of carriers) it would act more like an insulator, but I question why you wouldn't just use Si-on-I.
Weren't we trying to _reduce_ the k of dielectrics?
The higher the k - dielectric constant - of a material, the higher the capacitance of a thin layer of the material between electrodes. A higher-k gate insulator, for instance, would cause your chip to run _slower_ due to increased gate capacitance. This is why we've had things like foamed dielectrics invented (bubbles of gas or vacuum in a high-k material reduces the k value).
Perhaps there are other effects of using a high-k material that offset this. If this is actually the case, please enlighten me.
In summary, this sounds like a suspiciously marginal improvement. I'm curious as to what they're actually trying to do with these process adjustments.
That doesn't make a whole lot of sense to me! The Delta II site has the following info on it's capabilities:
...
The near earth orbit of 162Km isn't so much more than the X-prize 100Km
The difference is that the Delta doesn't just send things up to 162 km - it puts them in orbit.
To just send something up to 162 km, you'd just fire it straight up at a moderate speed. I'll fall right back down to earth again after reaching the target height.
To get something into orbit, you have to fire it mostly *sideways*, so that when it falls back to earth, it misses the planet (most basic explanation). The payload has to be moving a *heck* of a lot faster to do this. That's the difference between a sub-orbital and orbital trajectory, even if they reach the same altitude.
In my opinion, the U.S. should ban cloning an entire human for whatever purpose, as this could be used for some very evil things.
And so could normal child-rearing.
If you clone a human and bring the cloned baby to term, you have... a human baby, like any other.
Why not sidestep most of the debate arguments, and just rewrite parenthood laws to define parents as people who directly caused a child to come into existence? This will cover cloning and any other technologies that come up that could cause humans to be born in any but the old-fashioned way. It would declare clones human ("duh"), and would ensure that responsibility for these humans would be placed somewhere.
This doesn't even have to touch the abortion issue (the question of where in the line between zygote and baby a child becomes a human under the law). That can be left for the courts to fight out.
It's one thing to launch an 11m rocket some 5000ft, it's quite another to build a functioning spacecraft!
At the very minimum it would have to carry a ton of payload; most probably quite a bit more. To get an idea of the kinds of equipment involved, this link [af.mil] on the Delta II provides a good overview of the kind of sheer power and equipment needed to put even a relatively small 5 ton payload into space.
It turns out that it's a lot easier than you think to build an X-Prize-winning rocket.
The Delta rockets and other commercial launch vehicles need to get an object into _orbit_. This takes about 30 MJ/kg (the binding energy for LEO), or about 8 km/sec delta-V.
Satisfying the X prize only requires sending a payload up to 100 km. It doesn't have to stay there. This only takes about 1 MJ/kg (1.0e5 metres times about 10 m/sec^2). This corresponds to a delta-V of about 1.4 km/sec. This is much, much easier to achieve.
The reason why this is *much* easier (or more accurately, why anything higher than 2-3 km/sec is *really* hard), is that when the delta-V of your rocket is larger than your exhaust velocity, the amount of fuel needed to give that delta-V to a fixed amount of payload starts growing exponentially (it's roughly linear below this threshold). Typical specific impulses for rocket fuels are in the 2000-3000 N*s/kg range, corresponding to exhaust velocities of 2-3 km/sec. So, anything below about 2 km/sec can be accomplished with relative ease, while anything above about 4 km/sec requires a rocket that's mostly fuel (and probably multi-stage, unless you have extremely strong and light materials).
In summary, building a rocket that can lift a payload into a sub-orbital trajectory that tops out at 100 km is certainly within reach of a small group's resources.
[The real problem will be finding someone willing to pay for it. You can't lift payloads into orbit with this rocket, and cost of developing the rocket will probably be more than you'd get from the X-Prize.]
I'm sure we'll have hundreds of suggestions for nice hardware in short order, so I'm going to post a few gift suggestions that can't run Linux but still have geek appeal:
A good soldering iron, a stack of vector board cards, and a bucket of BJTs, resistors, caps, and LEDs.
Every once in a while I pull out a soldering iron and rediscover the fun of building widgetry from the ground up. Project books giving an introduction to electronics and a set of simple but neat building block circuits are still kicking around, and would be a useful addition to the pile as well.
A very large pile of Meccano or Construx.
I'll dig out my own pile of each someday. Geek appeal comes from trying to build things that most people would never think of (a working mechanical clock out of Construx was my biggest accomplishment with that medium).
Decent origami paper and a couple of books on the subject.
This falls under the "intricate hobbies" category, and so has a good chance of being welcome. I know I'm not the only geek with folded paper critters gracing his cube (a dragon, a Pierson's Puppeteer, and a Federation starship - yes, it can be done!).
Gift certificates for the nearest bookstore that has a decent computer reference section.
At $50-$100 Cdn apiece, one reference book costs as much as a large stack of sci-fi books. Help with getting new ones is always welcome, and I'm sure I'm not the only geek who likes documentation on the nifty tools I'm thinking about using (or am already using, for that matter).
There's no need to stick with hardware that will be obsolete in six months:).
Caveat with most of these - make sure your recipient is interested in them first. Yes, it ruins the surprise, but it's better than getting a bucketful of transistors when the sight of copper and lead make you cringe.
However, in stodgy old binary, the levels are typically something like 0 Volts (i.e. "off") and 5 Volts (or 3.5 Volts). A "typical" ternary system would add a negative voltage, like -5V (or -3.5V), since that's easier to detect reliably than an intermediate positive voltage value.
This set of articles uses current-mode devices with "forwards", "backwards", and "off" as the current levels. Similar idea, and it makes some aspects of the design simpler.
This is just a poster that glows -- it's a static picture that glows using a low amount of electricity. Unless you're running Windows, and all you need to display is the same bsod, you'll need a more "dynamic" display:).
If you can print conducting traces, you could set up a grid pattern of traces around pixels that would let you selectively activate pixels, much as you do in a passive-matrix LED. At any given time, one horizontal line (say) would be ground, and the rest would be at Vdd. Vertical lines would be driven or not driven depending on whether you want pixels in the active line on or off. If these printed pixels really are OLEDs - diodes - then you won't have to worry about the other horizontal traces shorting across the vertical lines.
I'm sure there are a number of ways of printing conducting traces with ink. Even a high-resistance trace could be electroplated after printing with thicker metal.
The only question is whether a) the type of OLEDs printed with this technology are really diodes, passing current only in one direction, and 2) whether instantaneous current can be high enough to give an acceptable _average_ current (and brightness) per row over the whole scanning cycle. A row turned on one thousandth of the time needs to be a thousand times as bright when it's on.
Other methods of addressing pixels in a display are of course possible. This is just one of the easiest (not necessarily best).
It sounds like a second-hand description of "quantum dot" technology. This is where you create a potential well in a conducting material and confine an electron within the well. Because the well is small, you get only certain energy levels permitted for the electron, just as in an atom. By changing the properties of the well, you change the properties of this "fake atom".
Just out of curiousity, do you any more info on this, or on applications for it? I buy that you can do this, but it seems hard to control this sort of potential except using nuclei?
It's actually quite easy to control the resulting energy configuration. The "allowed" energy levels depend on the size of the well (controlled when you etch it) and the electric potential between the inside and the outside of the well (which you can get "for free" by making the well on a semiconductor wafer and doping the inside and outside differently, or which you can fine-tune by having an electrode next to the well).
Applications include quantum computing (if you put multiple dots on a chip close enough together to interact with each other), and building semiconductor lasers with any frequency you like (even tunable frequency). More applications will undoubtedly arise; we've only just started to play with these things.
The more I read about modern physics, the more it seems our current models are flawed. I recently read an article in 'wired' about programmable materials made from 'atoms' which do not contain a nucleus. Simply lots of electrons forced into atom-like patterns.
I really wonder if we might not be better of throwing the physics textbooks out of the window and starting over again.
Bear in mind that "Wired" is not known for its contributors' understanding of science:).
It sounds like a second-hand description of "quantum dot" technology. This is where you create a potential well in a conducting material and confine an electron within the well. Because the well is small, you get only certain energy levels permitted for the electron, just as in an atom. By changing the properties of the well, you change the properties of this "fake atom".
There are many examples of materials where electrons aren't bound to individual atoms. Metals are a great example of this.
All of this is perfectly consistent with the models of how electrons and atoms behave (look up "Schrodinger's Equation" in a first-year physics text for a description of the model used for this).
Summary: Most perceived flaws are the result of bad or oversimplified explanations:).
"On a statistical basis, that would be a 1 in 400 probability of happening as a result of chance. "
That doesn't seem like a big deal to me. That sounds more like a problem in the experiement. I don't think anyone should be jumping for joy at this discovery until they duplicate it in another test.
This *is* a duplicate experiment - or close to it. Check the previous Slashdot article on the subject. This project is measuring a value that was measured by three previous experiments. Two of the previous experiments gave a very wide range for results, and the other one gave a narrow range for the results consistent with this experiment's results.
What the experiment shows is that the plan-vanilla Standard Model doesn't perfectly match reality. This is a surprise to nobody.
The results give a tantalizing look at one region of this breakdown, but proclaiming "a new form of energy or matter" is a bit premature at this point. What this will actually do is help confirm, refute, or fine-tune a few of the new models that are replacement candidates for the Standard Model.
The two operand Intel architecture does not allow the fused multiply add, so that the latency of such an operation is the latency of a multiply plus the latency of an add (and the destination register has to be one of the operands, although the other operand can be in memory, saving you a load). There are plenty of practical algorithms which benefit greatly from the fused multiply-add, for example polynomial evaluations, matrix multiplications, etc, a feature pioneered by IBM in the RS6000 series and that Intel is using in Inanium.
And people who claim that you can do loop unrolling to hide the latencies should check their math: with only 8 registers, there is no way to hide the latencies of a multiply plus an add on a P4, while it is almost trivial on a G4
Actually, it turns out that you can still mask the loop latency with a limited register set.
First, you can use "software pipelining" to mask quite a bit of the loop latency without having to unroll (it's a clever reshuffling of the loop instructions; for brevity, I won't describe it here). This requires one extra FP register over the straightforward implementation of an x86 dot-product loop (four instead of three, because I can no longer re-use scratch registers between steps).
Second, branch prediction will to a limited extent perform unrolling for you. While the architectural register file has only 8 registers, there are many more internal registers on the chip. Register renaming allows the processor to run several iterations of the loop in parallel without having to worry about namespace conflicts (though true dependencies remain intact). This works as long as the total number of iterations being unrolled fits within the scheduler's window (usually 8-16 instructions; I don't know how big the P4's window is).
In summary, for something as straightforward as a dot product, it's certainly possible to write x86 code that will avoid the penalty of having separate add and multiply instructions.
[You'll really be bound by the memory subsystem for both chips, but that's moot point for this discussion.]
the high end ppc desktops are topping out around 900MHz, while the p4's are hitting 2GHz. there has to be another explanation besides the complaint that jobs is ignorantly sitting on his thumbs.
Two factors come into play here.
The first is that, if I remember correctly, PPC and x86 chips use a different clocking scheme. This means that clock rates between them aren't even directly comparable (what a "clock" is depends on the clocking scheme).
The second is that it's perfectly possible that the PPC architecture is limited to lower clock rates than the x86 architecte. Signal propagation through gates takes time. If one architecture expects signals to propagate through logic three gates deep per clock, and another architecture expects signals to propagate through logic five gates deep, then of *course* one will have a faster maximum clock rate than the other. They would hopefully still be doing the same amount of work per unit of real time.
You should already be familiar with this from the Athlon/P4 spin war. A 0.18 micron Athlon core simply cannot be clocked as quickly as a 0.18 micron P4 core - no matter what you do. Does this make the Athlon automatically a poor performer? No, because it can do more per clock. Does this make the Athlon automatically kill the P4, because it "can do more per clock"? No, because the P4 can be clocked faster. Only real benchmarks will tell.
A point against Apple is that Apple has been allergic to publishing SPECmarks for its processors for the past couple of years (the only PPC-ish benchmarks are IBM's benchmarks of the Power series of chips, which forked after the G3 IIRC). This removes a very consistent (if somewhat flawed) means of comparison.
First of all, if you're one semester from graduation - finish your year. The piece of paper will still be worth something, especially when the economy rebounds.
As for finding the fun again... Take a break. Explore hobbies other than coding. Let your coding skills sit quietly in the back of your mind, and some time later, you'll feel the itch again - the need to code a little widget that's Really Cool. It mainly sounds like you're getting burned out to me.
OTOH, coding may or may not be what you really want to do. If your primary goal was to awe the world with your m4d sk1llz, you may simply not have noticed that you weren't having fun doing it. That will reveal itself during your sabbatical. If coding ever was fun for you, the desire to code will come back.
While there is no doubt that there is lot of cruft in the x86, you have to give Intel credit for getting way more performance out of it than anyone thought they wood. I remember back in the early 90s everyone kept talking about how RISC was going to kick Intel's ass for these very reasons: they would never be able to overcome the limitations of having to support backward compatibility. Yet, they are still standing, and RISC's advantages are very small in real terms.
You should probably doublecheck your sources, as they seem to have misinformed you on a couple of points.
Firstly, the past several generations _are_ RISC chips, with a wrapper around them that translates x86 instructions. This is why Intel chips have more decode stages in the pipeline than any clean architecture would (and why they were so eager to use a trace cache in the Itanium - among other things, it lets them skip the decode stages for instruction batches the processor has seen recently).
Secondly, there is a *huge* performance difference in practice between RISC and CISC architectures, for the simple reason that you can't pipeline CISC processors. You have instructions that do wildly varying amounts of work, taking wildly varying amounts of time to do it, sometimes without the total execution time being known (like the "loop" and "rep [foo]" instructions). Pipelining requires an instruction set with instructions that take roughly the same amount of time and that share many steps in common between instructions. RISC neatly provides all of this.
You can partially pipeline a CISC machine by only pipelining some types of instruction - heck, even a RISC machine will need to special-case things like divide operations - but pipelining is far, far more effective with a RISC architecture.
This was one more nail in the coffin of CISC cores (there are serious hardware and compiler complexity problems too).
If all the science was open, then everyone could have an understanding of all the risks and work together to prevent anything terrible happening.
First of all, science already is open. While individual implementations of a technology may be proprietary, the basic research underpinning it all is public (browse the journals section of a university's library some time).
Secondly, you are assuming that people a) are able to keep up with and understand all facets of science, b) are willing to do so, and c) are able to tell that someone's growing anthrax in their basement in time to do something about it.
a) can't even be done by scientists. This is why experts exist. Even within a fairly bounded discipline there are far too many papers and references for one person to keep up with - so people specialize in a niche that interests them and keep current with directly related topics.
Joe Average would have a much harder time of it.
b) Given that keeping up with even a niche in science in enough detail to truly understand it is pretty much a full-time job, and that most people already have full-time jobs and don't read scientific papers for leisure, I doubt that most people would be willing to keep abreast of all of science.
c) There are a wide range of diabolical terrorist plots that look surprisingly innocent right until the end. Concealment is easy. Detection is hard.
So, the populace at large will have a hard time policing itself. You could delegate the problem to a team of experts... which gives you something that looks a lot like the existing police force plus the various special agencies. In other words, we're already implementing what's probably the most pragmatic approximation to this ideal.
It is unrealistic to prevent information to be hidden in our modern would, instead we need to control how it can be used and by who.
The problem is that in most cases such control is also not feasible to implement in practice.
That leaves us with deterrents as a disincentive, and damage control plans for the inevitable few who are not deterred. Clever and nasty terrorist attacks will continue to happen, with a wide variety of technologies (basic and advanced). The best that IMO can be done is to attempt to minimize them and deal effectively with them when they do occur. Others, of course, will have widely varying opinions.
First of all, how are you going to build a self-replicating machine? The obstacles are so large as to be practically insurmountable.
It's easy to demonstrate that it's possible and to put an upper bound on the complexity of a replicater by looking for existing examples. Bacteria are self-replicating machines capable of synthesizing a wide variety of things, and while they're quite complex, understanding them is far from being an insurmountable challenge. Ditto understanding enough to design our own similar machines from scratch.
When it comes to commercial apps though like Windoze, rather than make something extraordinarily efficient that runs on the newest machines, they say "well the hardware takes care of efficiency, let's just make something with a lot of bells and whistles." What you end up with is grossly large applications that sloth along on extremely powerful machines that have the capability to be so much more. This is yet another reason to use Linux.
And of course, here on Linux, we never make the same mistake. We're all just chugging along with fvwm as our window manager and pico as our editor and all of our apps have a footprint of less than 4 megs in total...
And I have a bridge in Brooklyn to sell you.
Bloat and feature creep happen on all platforms. It's just easier to escape under Linux, because you aren't locked into a single toolset. Calling it a Windows-only problem is a gross misnomer, however.
The real problem with sluggishness under Windows is actually device probing (during boot) and hard drive seek time (when launching anything, due to the many configuration files it checks). Swapping isn't a concern if you use your system wisely, and applications are usually quite responsive (YMMV). Branding Windows application programmers as lazy when your system bogs down is grossly oversimplifying.
Doesn't that make a double exposure on the x-ray film? Would not the two illuminating point sources make a stereo image? Then you would be talking about a very tiny and detailed three dimensional x-ray images of flies.
You only get a stereoscopic image if you store each (single-source) image on a different piece of film. Two light sources for the same piece of film just gives you a double exposure. The next time you're walking in the evening, stand between two street lights and look at your shadow for an example of this; you're not going to get a 3D picture of yourself from those shadows very easily.
Cave painting, on the other hand, lasts at least tens of thousands of years, so if you REALLY want to preserve your history, I suggest you find a cave and paint in it with some yaks blood.
Or silkscreen using oxide pigments on to fiberglass cloth, and fire it to diffuse the oxides into the silica.
This will be as durable as any other form of quartz as far as fire, cold, water, and chemical attack are concerned, and would be reasonably resistant to physical wear if it was treated with respect.
A raging inferno would still melt the glass. A hot fire would cause the pigments on adjacent pages in a glass-cloth book to blend into each other, too. You can reduce this problem by using corundum fibers (aluminum oxide) and oxides that don't diffuse very quickly. This would take sustained forge-fire to destroy (corundum melts at over 2000 degrees centigrade, and is harder *and* more resistant to chemical attack than quartz).
I've been meaning to test this with a blowtorch, a patch of fiberglass fabric, and some rust powder for a while now. They're all about 30 feet from me; I just haven't bothered yet.
Problems are drawing/writing resolution, lack of a really nice range of pigment colours, and (for corundum) producing the cloth (corundum is a lot harder to spin into fibers than glass; I'm told that it doesn't go through the same "mushy" stage glass does).
If real-time strategy games required much more in the way of "strategy", we'd need drastic changes to the game interface.
Even with grouping and quick unit selection, I'm hard-pressed to manage more than two or three groups, and I'm in serious trouble if I have to deal with more than one part of the map at a time.
Multiple windows would help for this, but you'd still have the player having to divide their attention in real-time. You could give groups of units enough AI smarts to implement strategies you give them autonomously ("General PFault, take your troops to the ambush point and wait for my signal; then support my troops"), but then it becomes more of a computer-vs-computer game instead of a human-vs-computer/human game.
It's an interesting problem, and the easy solutions don't work very well. It'll be interesting to see what, if anything, finally emerges.
Slashdot Mirror
In english, this means using a different material for insulating layers and tweaking the doping of the substrate. A refinement, but hardly a breakthrough.
A couple of points about this puzzle me:
You could still call the channel material a substrate, and doping it might still do something, but it sounds like they're working with a bulk-silicon technique here. I'd thought that everyone and their dog was moving to silicon-on-insulator for capacitance reasons.
I suppose if you left the substrate undoped (depleted of carriers) it would act more like an insulator, but I question why you wouldn't just use Si-on-I.
The higher the k - dielectric constant - of a material, the higher the capacitance of a thin layer of the material between electrodes. A higher-k gate insulator, for instance, would cause your chip to run _slower_ due to increased gate capacitance. This is why we've had things like foamed dielectrics invented (bubbles of gas or vacuum in a high-k material reduces the k value).
Perhaps there are other effects of using a high-k material that offset this. If this is actually the case, please enlighten me.
In summary, this sounds like a suspiciously marginal improvement. I'm curious as to what they're actually trying to do with these process adjustments.
That doesn't make a whole lot of sense to me! The Delta II site has the following info on it's capabilities:
...
The near earth orbit of 162Km isn't so much more than the X-prize 100Km
The difference is that the Delta doesn't just send things up to 162 km - it puts them in orbit.
To just send something up to 162 km, you'd just fire it straight up at a moderate speed. I'll fall right back down to earth again after reaching the target height.
To get something into orbit, you have to fire it mostly *sideways*, so that when it falls back to earth, it misses the planet (most basic explanation). The payload has to be moving a *heck* of a lot faster to do this. That's the difference between a sub-orbital and orbital trajectory, even if they reach the same altitude.
In my opinion, the U.S. should ban cloning an entire human for whatever purpose, as this could be used for some very evil things.
And so could normal child-rearing.
If you clone a human and bring the cloned baby to term, you have... a human baby, like any other.
Why not sidestep most of the debate arguments, and just rewrite parenthood laws to define parents as people who directly caused a child to come into existence? This will cover cloning and any other technologies that come up that could cause humans to be born in any but the old-fashioned way. It would declare clones human ("duh"), and would ensure that responsibility for these humans would be placed somewhere.
This doesn't even have to touch the abortion issue (the question of where in the line between zygote and baby a child becomes a human under the law). That can be left for the courts to fight out.
It's one thing to launch an 11m rocket some 5000ft, it's quite another to build a functioning spacecraft!
At the very minimum it would have to carry a ton of payload; most probably quite a bit more. To get an idea of the kinds of equipment involved, this link [af.mil] on the Delta II provides a good overview of the kind of sheer power and equipment needed to put even a relatively small 5 ton payload into space.
It turns out that it's a lot easier than you think to build an X-Prize-winning rocket.
The Delta rockets and other commercial launch vehicles need to get an object into _orbit_. This takes about 30 MJ/kg (the binding energy for LEO), or about 8 km/sec delta-V.
Satisfying the X prize only requires sending a payload up to 100 km. It doesn't have to stay there. This only takes about 1 MJ/kg (1.0e5 metres times about 10 m/sec^2). This corresponds to a delta-V of about 1.4 km/sec. This is much, much easier to achieve.
The reason why this is *much* easier (or more accurately, why anything higher than 2-3 km/sec is *really* hard), is that when the delta-V of your rocket is larger than your exhaust velocity, the amount of fuel needed to give that delta-V to a fixed amount of payload starts growing exponentially (it's roughly linear below this threshold). Typical specific impulses for rocket fuels are in the 2000-3000 N*s/kg range, corresponding to exhaust velocities of 2-3 km/sec. So, anything below about 2 km/sec can be accomplished with relative ease, while anything above about 4 km/sec requires a rocket that's mostly fuel (and probably multi-stage, unless you have extremely strong and light materials).
In summary, building a rocket that can lift a payload into a sub-orbital trajectory that tops out at 100 km is certainly within reach of a small group's resources.
[The real problem will be finding someone willing to pay for it. You can't lift payloads into orbit with this rocket, and cost of developing the rocket will probably be more than you'd get from the X-Prize.]
Where did you get the model for a pierson's puppeteer? Or did you invent it yourself?
Came up with it myself when I was bored. Click on "user info" and check my previous post for a link.
And BTW, I'd love to know how to fold your Pierson's Puppeteer!
One free web account later:
http://www.angelfire.com/d20/roll_d3_for_this/
Enjoy.
Every once in a while I pull out a soldering iron and rediscover the fun of building widgetry from the ground up. Project books giving an introduction to electronics and a set of simple but neat building block circuits are still kicking around, and would be a useful addition to the pile as well.
I'll dig out my own pile of each someday. Geek appeal comes from trying to build things that most people would never think of (a working mechanical clock out of Construx was my biggest accomplishment with that medium).
This falls under the "intricate hobbies" category, and so has a good chance of being welcome. I know I'm not the only geek with folded paper critters gracing his cube (a dragon, a Pierson's Puppeteer, and a Federation starship - yes, it can be done!).
At $50-$100 Cdn apiece, one reference book costs as much as a large stack of sci-fi books. Help with getting new ones is always welcome, and I'm sure I'm not the only geek who likes documentation on the nifty tools I'm thinking about using (or am already using, for that matter).
There's no need to stick with hardware that will be obsolete in six months
Caveat with most of these - make sure your recipient is interested in them first. Yes, it ruins the surprise, but it's better than getting a bucketful of transistors when the sight of copper and lead make you cringe.
However, in stodgy old binary, the levels are typically something like 0 Volts (i.e. "off") and 5 Volts (or 3.5 Volts). A "typical" ternary system would add a negative voltage, like -5V (or -3.5V), since that's easier to detect reliably than an intermediate positive voltage value.
This set of articles uses current-mode devices with "forwards", "backwards", and "off" as the current levels. Similar idea, and it makes some aspects of the design simpler.
Proofread, and still goofed. I need sleep :).
This is just a poster that glows -- it's a static picture that glows using a low amount of electricity. Unless you're running Windows, and all you need to display is the same bsod, you'll need a more "dynamic" display :).
If you can print conducting traces, you could set up a grid pattern of traces around pixels that would let you selectively activate pixels, much as you do in a passive-matrix LED. At any given time, one horizontal line (say) would be ground, and the rest would be at Vdd. Vertical lines would be driven or not driven depending on whether you want pixels in the active line on or off. If these printed pixels really are OLEDs - diodes - then you won't have to worry about the other horizontal traces shorting across the vertical lines.
I'm sure there are a number of ways of printing conducting traces with ink. Even a high-resistance trace could be electroplated after printing with thicker metal.
The only question is whether a) the type of OLEDs printed with this technology are really diodes, passing current only in one direction, and 2) whether instantaneous current can be high enough to give an acceptable _average_ current (and brightness) per row over the whole scanning cycle. A row turned on one thousandth of the time needs to be a thousand times as bright when it's on.
Other methods of addressing pixels in a display are of course possible. This is just one of the easiest (not necessarily best).
It sounds like a second-hand description of "quantum dot" technology. This is where you create a potential well in a conducting material and confine an electron within the well. Because the well is small, you get only certain energy levels permitted for the electron, just as in an atom. By changing the properties of the well, you change the properties of this "fake atom".
/ 1097corcoran.html
t .html
Just out of curiousity, do you any more info on this, or on applications for it? I buy that you can do this, but it seems hard to control this sort of potential except using nuclei?
It's actually quite easy to control the resulting energy configuration. The "allowed" energy levels depend on the size of the well (controlled when you etch it) and the electric potential between the inside and the outside of the well (which you can get "for free" by making the well on a semiconductor wafer and doping the inside and outside differently, or which you can fine-tune by having an electrode next to the well).
A decent introduction into quantum dots is here:
http://www.sciam.com/specialissues/1097solidstate
Scientific American has a few other articles on quantum dots, which you can find through their search page.
A collection of more in-depth articles is here:
http://www.mitre.org/research/nanotech/quantum_do
Applications include quantum computing (if you put multiple dots on a chip close enough together to interact with each other), and building semiconductor lasers with any frequency you like (even tunable frequency). More applications will undoubtedly arise; we've only just started to play with these things.
The more I read about modern physics, the more it seems our current models are flawed. I recently read an article in 'wired' about programmable materials made from 'atoms' which do not contain a nucleus. Simply lots of electrons forced into atom-like patterns.
:).
:).
I really wonder if we might not be better of throwing the physics textbooks out of the window and starting over again.
Bear in mind that "Wired" is not known for its contributors' understanding of science
It sounds like a second-hand description of "quantum dot" technology. This is where you create a potential well in a conducting material and confine an electron within the well. Because the well is small, you get only certain energy levels permitted for the electron, just as in an atom. By changing the properties of the well, you change the properties of this "fake atom".
There are many examples of materials where electrons aren't bound to individual atoms. Metals are a great example of this.
All of this is perfectly consistent with the models of how electrons and atoms behave (look up "Schrodinger's Equation" in a first-year physics text for a description of the model used for this).
Summary: Most perceived flaws are the result of bad or oversimplified explanations
"On a statistical basis, that would be a 1 in 400 probability of happening as a result of chance. "
That doesn't seem like a big deal to me. That sounds more like a problem in the experiement. I don't think anyone should be jumping for joy at this discovery until they duplicate it in another test.
This *is* a duplicate experiment - or close to it. Check the previous Slashdot article on the subject. This project is measuring a value that was measured by three previous experiments. Two of the previous experiments gave a very wide range for results, and the other one gave a narrow range for the results consistent with this experiment's results.
This was posted a few days ago, along with links to much better articles:
1 3
http://slashdot.org/article.pl?sid=01/11/08/22212
What the experiment shows is that the plan-vanilla Standard Model doesn't perfectly match reality. This is a surprise to nobody.
The results give a tantalizing look at one region of this breakdown, but proclaiming "a new form of energy or matter" is a bit premature at this point. What this will actually do is help confirm, refute, or fine-tune a few of the new models that are replacement candidates for the Standard Model.
The two operand Intel architecture does not allow the fused multiply add, so that the latency of such an operation is the latency of a multiply plus the latency of an add (and the destination register has to be one of the operands, although the other operand can be in memory, saving you a load). There are plenty of practical algorithms which benefit greatly from the fused multiply-add, for example polynomial evaluations, matrix multiplications, etc, a feature pioneered by IBM in the RS6000 series and that Intel is using in Inanium.
And people who claim that you can do loop unrolling to hide the latencies should check their math: with only 8 registers, there is no way to hide the latencies of a multiply plus an add on a P4, while it is almost trivial on a G4
Actually, it turns out that you can still mask the loop latency with a limited register set.
First, you can use "software pipelining" to mask quite a bit of the loop latency without having to unroll (it's a clever reshuffling of the loop instructions; for brevity, I won't describe it here). This requires one extra FP register over the straightforward implementation of an x86 dot-product loop (four instead of three, because I can no longer re-use scratch registers between steps).
Second, branch prediction will to a limited extent perform unrolling for you. While the architectural register file has only 8 registers, there are many more internal registers on the chip. Register renaming allows the processor to run several iterations of the loop in parallel without having to worry about namespace conflicts (though true dependencies remain intact). This works as long as the total number of iterations being unrolled fits within the scheduler's window (usually 8-16 instructions; I don't know how big the P4's window is).
In summary, for something as straightforward as a dot product, it's certainly possible to write x86 code that will avoid the penalty of having separate add and multiply instructions.
[You'll really be bound by the memory subsystem for both chips, but that's moot point for this discussion.]
the high end ppc desktops are topping out around 900MHz, while the p4's are hitting 2GHz. there has to be another explanation besides the complaint that jobs is ignorantly sitting on his thumbs.
Two factors come into play here.
The first is that, if I remember correctly, PPC and x86 chips use a different clocking scheme. This means that clock rates between them aren't even directly comparable (what a "clock" is depends on the clocking scheme).
The second is that it's perfectly possible that the PPC architecture is limited to lower clock rates than the x86 architecte. Signal propagation through gates takes time. If one architecture expects signals to propagate through logic three gates deep per clock, and another architecture expects signals to propagate through logic five gates deep, then of *course* one will have a faster maximum clock rate than the other. They would hopefully still be doing the same amount of work per unit of real time.
You should already be familiar with this from the Athlon/P4 spin war. A 0.18 micron Athlon core simply cannot be clocked as quickly as a 0.18 micron P4 core - no matter what you do. Does this make the Athlon automatically a poor performer? No, because it can do more per clock. Does this make the Athlon automatically kill the P4, because it "can do more per clock"? No, because the P4 can be clocked faster. Only real benchmarks will tell.
A point against Apple is that Apple has been allergic to publishing SPECmarks for its processors for the past couple of years (the only PPC-ish benchmarks are IBM's benchmarks of the Power series of chips, which forked after the G3 IIRC). This removes a very consistent (if somewhat flawed) means of comparison.
First of all, if you're one semester from graduation - finish your year. The piece of paper will still be worth something, especially when the economy rebounds.
:). Good luck.
As for finding the fun again... Take a break. Explore hobbies other than coding. Let your coding skills sit quietly in the back of your mind, and some time later, you'll feel the itch again - the need to code a little widget that's Really Cool. It mainly sounds like you're getting burned out to me.
OTOH, coding may or may not be what you really want to do. If your primary goal was to awe the world with your m4d sk1llz, you may simply not have noticed that you weren't having fun doing it. That will reveal itself during your sabbatical. If coding ever was fun for you, the desire to code will come back.
YMMV
While there is no doubt that there is lot of cruft in the x86, you have to give Intel credit for getting way more performance out of it than anyone thought they wood. I remember back in the early 90s everyone kept talking about how RISC was going to kick Intel's ass for these very reasons: they would never be able to overcome the limitations of having to support backward compatibility. Yet, they are still standing, and RISC's advantages are very small in real terms.
You should probably doublecheck your sources, as they seem to have misinformed you on a couple of points.
Firstly, the past several generations _are_ RISC chips, with a wrapper around them that translates x86 instructions. This is why Intel chips have more decode stages in the pipeline than any clean architecture would (and why they were so eager to use a trace cache in the Itanium - among other things, it lets them skip the decode stages for instruction batches the processor has seen recently).
Secondly, there is a *huge* performance difference in practice between RISC and CISC architectures, for the simple reason that you can't pipeline CISC processors. You have instructions that do wildly varying amounts of work, taking wildly varying amounts of time to do it, sometimes without the total execution time being known (like the "loop" and "rep [foo]" instructions). Pipelining requires an instruction set with instructions that take roughly the same amount of time and that share many steps in common between instructions. RISC neatly provides all of this.
You can partially pipeline a CISC machine by only pipelining some types of instruction - heck, even a RISC machine will need to special-case things like divide operations - but pipelining is far, far more effective with a RISC architecture.
This was one more nail in the coffin of CISC cores (there are serious hardware and compiler complexity problems too).
If all the science was open, then everyone could have an understanding of all the risks and work together to prevent anything terrible happening.
First of all, science already is open. While individual implementations of a technology may be proprietary, the basic research underpinning it all is public (browse the journals section of a university's library some time).
Secondly, you are assuming that people a) are able to keep up with and understand all facets of science, b) are willing to do so, and c) are able to tell that someone's growing anthrax in their basement in time to do something about it.
a) can't even be done by scientists. This is why experts exist. Even within a fairly bounded discipline there are far too many papers and references for one person to keep up with - so people specialize in a niche that interests them and keep current with directly related topics.
Joe Average would have a much harder time of it.
b) Given that keeping up with even a niche in science in enough detail to truly understand it is pretty much a full-time job, and that most people already have full-time jobs and don't read scientific papers for leisure, I doubt that most people would be willing to keep abreast of all of science.
c) There are a wide range of diabolical terrorist plots that look surprisingly innocent right until the end. Concealment is easy. Detection is hard.
So, the populace at large will have a hard time policing itself. You could delegate the problem to a team of experts... which gives you something that looks a lot like the existing police force plus the various special agencies. In other words, we're already implementing what's probably the most pragmatic approximation to this ideal.
It is unrealistic to prevent information to be hidden in our modern would, instead we need to control how it can be used and by who.
The problem is that in most cases such control is also not feasible to implement in practice.
That leaves us with deterrents as a disincentive, and damage control plans for the inevitable few who are not deterred. Clever and nasty terrorist attacks will continue to happen, with a wide variety of technologies (basic and advanced). The best that IMO can be done is to attempt to minimize them and deal effectively with them when they do occur. Others, of course, will have widely varying opinions.
First of all, how are you going to build a self-replicating machine? The obstacles are so large as to be practically insurmountable.
It's easy to demonstrate that it's possible and to put an upper bound on the complexity of a replicater by looking for existing examples. Bacteria are self-replicating machines capable of synthesizing a wide variety of things, and while they're quite complex, understanding them is far from being an insurmountable challenge. Ditto understanding enough to design our own similar machines from scratch.
When it comes to commercial apps though like Windoze, rather than make something extraordinarily efficient that runs on the newest machines, they say "well the hardware takes care of efficiency, let's just make something with a lot of bells and whistles." What you end up with is grossly large applications that sloth along on extremely powerful machines that have the capability to be so much more. This is yet another reason to use Linux.
And of course, here on Linux, we never make the same mistake. We're all just chugging along with fvwm as our window manager and pico as our editor and all of our apps have a footprint of less than 4 megs in total...
And I have a bridge in Brooklyn to sell you.
Bloat and feature creep happen on all platforms. It's just easier to escape under Linux, because you aren't locked into a single toolset. Calling it a Windows-only problem is a gross misnomer, however.
The real problem with sluggishness under Windows is actually device probing (during boot) and hard drive seek time (when launching anything, due to the many configuration files it checks). Swapping isn't a concern if you use your system wisely, and applications are usually quite responsive (YMMV). Branding Windows application programmers as lazy when your system bogs down is grossly oversimplifying.
Doesn't that make a double exposure on the x-ray film? Would not the two illuminating point sources make a stereo image? Then you would be talking about a very tiny and detailed three dimensional x-ray images of flies.
You only get a stereoscopic image if you store each (single-source) image on a different piece of film. Two light sources for the same piece of film just gives you a double exposure. The next time you're walking in the evening, stand between two street lights and look at your shadow for an example of this; you're not going to get a 3D picture of yourself from those shadows very easily.
Cave painting, on the other hand, lasts at least tens of thousands of years, so if you REALLY want to preserve your history, I suggest you find a cave and paint in it with some yaks blood.
Or silkscreen using oxide pigments on to fiberglass cloth, and fire it to diffuse the oxides into the silica.
This will be as durable as any other form of quartz as far as fire, cold, water, and chemical attack are concerned, and would be reasonably resistant to physical wear if it was treated with respect.
A raging inferno would still melt the glass. A hot fire would cause the pigments on adjacent pages in a glass-cloth book to blend into each other, too. You can reduce this problem by using corundum fibers (aluminum oxide) and oxides that don't diffuse very quickly. This would take sustained forge-fire to destroy (corundum melts at over 2000 degrees centigrade, and is harder *and* more resistant to chemical attack than quartz).
I've been meaning to test this with a blowtorch, a patch of fiberglass fabric, and some rust powder for a while now. They're all about 30 feet from me; I just haven't bothered yet.
Problems are drawing/writing resolution, lack of a really nice range of pigment colours, and (for corundum) producing the cloth (corundum is a lot harder to spin into fibers than glass; I'm told that it doesn't go through the same "mushy" stage glass does).
We all know the solution is wireless electricity distribution.... :) Tesla didn't finish his work in this arena though
;)
Great idea. Um... why are my CDs sparking?
If real-time strategy games required much more in the way of "strategy", we'd need drastic changes to the game interface.
Even with grouping and quick unit selection, I'm hard-pressed to manage more than two or three groups, and I'm in serious trouble if I have to deal with more than one part of the map at a time.
Multiple windows would help for this, but you'd still have the player having to divide their attention in real-time. You could give groups of units enough AI smarts to implement strategies you give them autonomously ("General PFault, take your troops to the ambush point and wait for my signal; then support my troops"), but then it becomes more of a computer-vs-computer game instead of a human-vs-computer/human game.
It's an interesting problem, and the easy solutions don't work very well. It'll be interesting to see what, if anything, finally emerges.