Umm, what? Any decent scientist these days should be able to program. Setting up an SQL database takes next to no time, as long as you use the right tools to do it. pgAdmin comes to mind, or oo.ORG database frontend. Now the problem may be that many scientists and engineers still don't know how to program, but that's another story.
One can mix-n-match: use flat files to store raw data, post-processed stuff, etc, but use a database to keep track of everything. The latter can even be handled by your OS from the get-go. On OS X you could write a spotlight plugin or two for your data files, and as long as the file format allows storing metadata within the files, spotlight will index it. Same goes for Windows Search. You could also use native mechanisms for adding metadata to files - those exist IIRC on both Windows 7 and on OS X.
It's quite, simple really. There is the legal department, in charge of chasing people away. And then there's the marketing department, in charge of pulling people in. And then there's no communication between them. With opposite missions, what do you expect.
Add to the fact that the legal seems to be adept at the chasing away part, while somewhat forgetful of the law they apparently learned many moons ago in, uhhhunh whatwuzitcalled college methinks? To the lawyers: don't party so hard when in college, or you'll have trouble understanding the law later.
Does anyone do performance review on corporate legal teams? As in real reviews where any monetary awards to the company are balanced with lost goodwill and whatnot? There's a lawyer or two waiting to be fired here, methinks.
You could potentially do it using the approach taken by Valgrind. It'd be probably slow as molasses, but it could be done. Just in time re-compiling, that is. You could emulate address space isolation that way.
The list of areas Vinay Deolalikar pulls his ideas from is just amazing, I mean who'd have thought of using statistical physics to solve a deterministic logic problem.
By the end of that summer of 1983, Richard had completed his analysis of the behavior of the router, and much to our surprise and amusement, he presented his answer in the form of a set of partial differential equations. To a physicist this may seem natural, but to a computer designer, treating a set of boolean circuits as a continuous, differentiable system is a bit strange. Feynman's router equations were in terms of variables representing continuous quantities such as "the average number of 1 bits in a message address." I was much more accustomed to seeing analysis in terms of inductive proof and case analysis than taking the derivative of "the number of 1's" with respect to time. Our discrete analysis said we needed seven buffers per chip; Feynman's equations suggested that we only needed five. We decided to play it safe and ignore Feynman.
The decision to ignore Feynman's analysis was made in September, but by next spring we were up against a wall. The chips that we had designed were slightly too big to manufacture and the only way to solve the problem was to cut the number of buffers per chip back to five. Since Feynman's equations claimed we could do this safely, his unconventional methods of analysis started looking better and better to us. We decided to go ahead and make the chips with the smaller number of buffers.
Fortunately, he was right. When we put together the chips the machine worked. The first program run on the machine in April of 1985 was Conway's game of Life.
Look, I know that PI is defined in a certain way. That doesn't mean we know why it comes out just so -- why the particular numerical value is just what it is.
Of course we can show that if we sum the series expansion that comes from the definition, the answer is 3.14... But that doesn't tell us why the first digit isn't say 2. Or 8. There are no further theories to that. It's pretty much an experiment: you take a formula for PI, you compute, and you get a number. So, while we certainly know that if you set things up just so, and derive what the value of PI might be, starting with some axioms and definition being one half the ratio of circle's circumference to radius, and you get the value.
But had we had a real understanding of things, we could take a bunch of say starting digits of PI, and could actually tell that it is such a "special" number -- I'll get to what I mean by special in a second. Right now we can only compare to a list of "special" numbers. If you told all about how PI is derived to someone who never saw a part of the computed value itself, they'd have no way of telling that 3.1415... is PI -- they'd have to go through the motions of calculating it first, and then do the comparison and say "hey, the digits are same".
The way I see that PI is "special" is that it is a condensation of a whole lot of things. We start with a certain metric that works in particular type of a space -- that is, we have to define what distance is, and how to measure it, and that points of the object are equidistant. And this choice -- of metric/space and object -- is embedded in the number PI.
One can then think that real understanding would come if we could, given some number, find some structure in it that would tell us: OK, this number comes from some measurements in a space like so, with such a metric, of an object with infinitely many points equidistant to to one point. Maybe even we could enumerate such numbers, to find interesting spaces and metrics, or even interesting kinds of objects. But we have no such understanding at all, it seems.
We have a better understanding of the structure of integers -- we have, after all, better ways of testing for primality than using sieves (enumerations), we can do factorizations, etc.
But for real numbers, we're pretty much like newborns. Or else I missed some big discoveries, that is.
I'm serious. How would you even start an argument about PI not being a physical constant? It's really just a matter of definition, and in that sense there's no argument.
But we say that physical constants are some things we measure, and other seemingly fundamental things we can measure are not (like PI). PI can be of course measured to a good few digits by manufacturing a sphere or a disk/cylinder, and then measuring the circumference and radius. We then also have mathematical theories that can come up with PI to arbitrary accuracy, but that's just a bonus. We don't know -- maybe we will come up with similarly good theories for other things that can be measured, say the fine-structure constant.
We really don't know how closely the physics of our Universe are coupled with the structure we see in the mathematics. It's kind of philosophical, but we "discover" things in mathematics. So what is this thing that we discover then -- where is it. In our minds only? Or is it really just our minds picking it up, noticing it.
So I can't really say anyone can quite wrap his/her mind around it. I'd go further: anyone claiming to be able to do so is quite a kook. It's like claiming to understand why quantum mechanics or gravitation behaves just like so. We have no clue *why* it behaves just like so, exactly like we have no clue what to make of the value of PI.
We know how to apply all this knowledge, but we know of no "ulterior motive" for it. Certain phenomena can be inferred from other, more fundamental ones -- say Bernoulli effect is just a manifestation of laws of conservation intertwined with laws of dynamics -- so we can say why we see the Bernoulli effect. But we can't say why the quantum mechanical behavior of the atoms that make up the medium is just so -- we know no more basic stuff to explain that. We just observe it to be so, but can claim no further insight.
Same with PI: we have no clue where it came from. I don't claim we need to have such a clue -- but please, don't claim more insight than we actually possess. We are pretty clueless and IMHO that's what's exciting: there's still plenty of stuff to discover.
I'd think noone cares if this stuff is admissible in court. If there were ever trials relevant to those leaked documents, the court would subpoena the originals anyway, right?
You have no fucking clue how wartime intelligence works, do you?
What a fair example of circular logic:
1. It's wartime (supposedly, but let's ignore that for a moment). 2. Superiority is to be maintained. 3. Shit happens, and is embarrassing. 4. But (2), ergo more shit will happen. 5. It gets even more embarrassing, and were it to be uncovered we lose superiority and put our fine soldiers at harm. 6. But (2), ergo we must keep it all covered up. 7. Goto 3
The problem I see is that this is all but a self-propagating fallacy. I'm not a radical pacifist, but maybe, just maybe, we ought not to be wolves and make sure shit doesn't happen, that we won't be embarrassed by all of it? There were surely a few embarrassing moments on the Allied side during WWII, but on the balance I think it was minuscule compared to what's going on today. I'm glad I'm getting almost all of my federal taxes back. It'd feel quite bad to pay for any of that...
I think that it's one of those big things that we just don't quite can wrap our minds around. What does it mean that there's a certain structure and properties to geometries, that topology of things looks just like so and not some other way, etc? Do those things depend at all on the universe we live in? Would, somehow, someone in another universe find, that PI has a different value even though it's not a physical constant as far as we can tell? And maybe we're just mistaken and it is really a physical constant?
I'd say that things would need to be so "different" that if PI had a different value, then also the Jones polynomial for knots would look different -- things would have to be just fundamentally different at a low level -- so it seems to me.
Here we go again with people thinking that the paper paradigm applies to the digital world.
How on Earth do you return digital documents? Do you scrape the oxide layer off the hard drives, put it in a little vial, mark it with volume mount point(s) and put it into an envelope addressed to Pentagon? Oh, yes, I know, you first print out the directory listing (like we used to do with the floppies), tape it to the vial, then scrape, fill the vial and ship.
As for the further documents -- they better watch out, because WikiLeaks may just give up and publish all of the unredacted stuff just to preserve it.
As for WikiLeaks somehow "embarrassing" the U.S. military: waitaminuzel here. Did WikiLeaks compel the military to do all the embarrassing stuff? No? Then well, maybe it was better the taxpayers knew what their money is spent on, huh?
That's spooky. I read The Onion every now and then, but I missed that article. It's not very funny now, though. I give the author props for seeing Bush for who he really was.
I always keep saying that eye-movement control inputs should be a last resort -- for people with things like ALS where you have nothing else left (in ALS you usually have blinks, though, and those are easier to reliably sense).
Eye movements partake in acquisition of visual input. If you have to control things with eye movements, you usually do it at the cost of not seeing what you'd ordinarily see.
As for the interception idea of yours: said screen would have to have a very good refresh rate. For visual exploration, our visual system is pretty much a snapshot taker. When your eyes move from point to point, they do it in a swift motion called a saccade. During a saccade, the image slides on the retina, and the blurred image is suppressed -- we don't get to see blurs several times per second. If you have a tracking fundus imager (it sees what's on the retina in spite of the eye moving around), a third person observer can see what's projected on the retina and see the blurs. A saccade triggers the masking, then once the image stabilizes, a snapshot is taken.
There are simple experiments to demonstrate this effect. You are in a moving train, and you look at the adjacent tracks. Once the train moves fast enough, the tracks are blurry. You then attempt to look forward and backward along the track. As you do, there are points at which the saccadic velocity exactly cancels the track's motion on the retina, and you see the track clearly -- ties and trash and all. Or look around a quickly spinning ceiling fan.
If you could intercept eye muscle signals while keeping the eyeball stable, you could certainly shorten the time it takes to acquire the image -- the saccadic masking is ready to cease and accept a stable image much before the saccade would end.
But you don't really need to cut any nerves. You can let the eyes move, and instead use a tracking projector system that projects stable image on the retina no matter how the eye moves. This IIRC has been done. You can estimate the saccadic motion's end position (saccadic amplitude) in the early 1/3 of the saccade, so you could probably shorten the time it takes from beginning of the saccade to acquisition of the image.
The only thing is: it would be pointless. It takes a good time (150+ ms) to process the image once it's acquired by the visual system's imager front-end, so even if your tracking projector could be omniscient w.r.t. saccadic target, and stabilize a target image on the retina with no latency at all, it wouldn't improve your visual system latency by a whole lot -- it'd be less than a factor of two.
Eye-movement control for the masses is a rather silly idea, I bin it with, say, having answer to the "meaning" of life (everyone feels like knowing that would be "good"). Both things are pretty useless -- the first one because it trades off something our visual system is good at (vision) for a much less accurate version of what our limbs can do better (providing control input). The latter -- well, what the heck would you do with an answer to that? Just think about it.
I completely understand what you're saying, but you list some very specific circumstances. A cal lab nominally is supposed to calibrate, not repair, so they'll simply give up when they use up the adjustment range and they're still out of spec. That's hardly rejecting "due to it being fixed" -- any number of things, including simple passage of time, can make an instrument do this. They won't try to understand why it doesn't work, just that following the cal procedure (usually with lots of in-house clarifications/improvements) the tech gets stuck at a point and that's that.
For older equipment, some cal labs offer calibration bundled with repairs, so that they'll try to fix precisely what you're talking about (when it's still fixable).
Quite often it's very obvious that someone was "working" on an instrument, and you end up undoing most of such "fixes" -- I agree there. This can potentially take many billable hours, so if they are given a go to repair, they can love such jobs.
You have to admit though, that the whole thing is extremely user-friendly even when jailbreaking. No stupid yellow pop-up ActiveX warnings, just tap here, slide there, and off you go. I wonder how much Apple influence was there when the UI was designed for this jailbreak. Compare how nice it looks next to most PC-based cracks/hacks that one can download. I'm half-serious here.
I'd say that a decent scope must have at least the following:
- zero hold-off time (cheap CCD-based scopes like TDS1k series take maybe a hundred sweeps a second, any scope that can do faster than 100k sweeps/s is either in $10k+ range or is analog)
- very good input overload recovery times (100ns) -- there are no production scopes AFAIK that can offer that, you need a used sampling scope/plugin for that, or you need to look at some LTC appnotes;)
- proper antialiasing
Once you "splurge" on those, you may as well have a decent ADC to give you enough dynamic range to offer rudimentary spectrum analysis/receiver functionality.
To get the basics going, you need a working SDR receiver -- you just buy the thing. My coworker recently got one -- the post-processor card with USB output (Cyclone II based), the sampler card with Cyclone III and 16 bit 160 MSPS LTC ADC, and a backplane and power supply. The first two were assembled.
This is quite a decent *receiver*, it has a dynamic range that's pretty much unheard of on oscilloscopes. Would make a half-decent spectrum analyzer, too. Once you add a DC-coupled front end, you may as well have an attenuator with two or three settings (say 10V/div, 1V/div, 0.1V/div), the ADC's resolution is enough to deal with it. Testing it can be done using itself -- the proof is in the pudding, I mean in the stuff you acquire.
Repurposing say a 7A13 plugin for input conditioner can be done without an oscilloscope, a DVM is all you need to get operating points right, and then you watch the signals using the device itself.
Hardly a multi-year R&D project. If you're handy with VHDL, it can be a couple weekends worth of work -- and remember, most of the hardware is already there for you.
Many DAQ cards are notorious for their aliasing problems. You'll be lucky if you find a 2nd order lowpass before the ADC. I've seen cards with 250kHz sampling rate (at 16 bits) that are significantly sensitive (think 30dB down) to stuff at 10MHz. Those are very good -- that is if you want to find cables that have good (low) common mode to differential mode conversion ratios.
What lab will reject an instrument "due to it being fixed"?! Most contemporary instruments are calibrated without opening the covers, you could literally replace everything inside and just make it emulate the original instrument and no one would be any wiser.
Probably the best deal would be to get a digital radio. If you can live with ~150Ms/s (a tad slow, but hey), then a cheap thing to do would be to get a digital radio (SDR) system. Say Mercury SDR. Those things typically have a good, 16 bit 100+Ms/s ADC front end, feeding into an FPGA that can do a lot of processing goodies, with low noise, and you should be able to hook up a Tek 7k plugin as a front-end after a few tweaks (simply to get going). You can get everything for $700. You have open source software, full documentation, and you can put a lot of very interesting signal processing on the FPGA. Keeping sampler's speed limitations in mind, you can otherwise easily match performance of many lower-end spectrum analyzers, and $20k+ scopes.
There are no $2k digital scopes with any decent feature set to speak of, even second-hand ones.
If you're into tweaking analog, then a Tektronix 7k mainframe with proper plugins gives you everything you may need. Heck, you can even get a simple logic analyzer for those. With *analog* zoom, no less.
One way of doing it: laptop running OS-X + a camera with live preview on the laptop. Then use afloat to overlay things. I'm a happy afloat user, and it has helped tremendously with some CAD applications that don't let you overlay graphic files as backgrounds for tracing.
You are talking about 40% duty cycle on the fan. I'm not talking about that. A GPU should not self destruct with ZERO heatsinks attached. It should run very slowly, but it should keep on running.
I say that the chip itself should be able to actively regulate (think control theory) its thermal output to match the attached thermal sinks -- both their thermal resistance, and their thermal mass. It's not a hard problem. Can be done in a few hundred transistors.
Your card's GPU chip(s) obviously don't have this "innovative" technology. I mean, come on, die overtemperature protection is built into $0.10 three pin voltage regulators (think 78L05). Obviously those have simple on-off control, but if you can spare a few hundred transistors, you can have a full blown adaptive PID-like control, with outputs that control clock scaling and computational unit allocation. On a GPU, you can scale thermal output by slowing down the clock, and by idling computational units. Heck, you should be able to scale it down by almost two orders of magnitudey. That should allow a bare GPU to run -- very slowly -- with no heat sink on.
Slashdot Mirror
Umm, what? Any decent scientist these days should be able to program. Setting up an SQL database takes next to no time, as long as you use the right tools to do it. pgAdmin comes to mind, or oo.ORG database frontend. Now the problem may be that many scientists and engineers still don't know how to program, but that's another story.
One can mix-n-match: use flat files to store raw data, post-processed stuff, etc, but use a database to keep track of everything. The latter can even be handled by your OS from the get-go. On OS X you could write a spotlight plugin or two for your data files, and as long as the file format allows storing metadata within the files, spotlight will index it. Same goes for Windows Search. You could also use native mechanisms for adding metadata to files - those exist IIRC on both Windows 7 and on OS X.
I did mean corpus callostomy. It almost completely severs the connection between the hemispheres.
Colostomy is something else, although the smell would be about right, I admit.
It's quite, simple really. There is the legal department, in charge of chasing people away. And then there's the marketing department, in charge of pulling people in. And then there's no communication between them. With opposite missions, what do you expect.
Add to the fact that the legal seems to be adept at the chasing away part, while somewhat forgetful of the law they apparently learned many moons ago in, uhhhunh whatwuzitcalled college methinks? To the lawyers: don't party so hard when in college, or you'll have trouble understanding the law later.
Does anyone do performance review on corporate legal teams? As in real reviews where any monetary awards to the company are balanced with lost goodwill and whatnot? There's a lawyer or two waiting to be fired here, methinks.
You could potentially do it using the approach taken by Valgrind. It'd be probably slow as molasses, but it could be done. Just in time re-compiling, that is. You could emulate address space isolation that way.
The list of areas Vinay Deolalikar pulls his ideas from is just amazing, I mean who'd have thought of using statistical physics to solve a deterministic logic problem.
I think Feynman was close enough.
By the end of that summer of 1983, Richard had completed his analysis of the behavior of the router, and much to our surprise and amusement, he presented his answer in the form of a set of partial differential equations. To a physicist this may seem natural, but to a computer designer, treating a set of boolean circuits as a continuous, differentiable system is a bit strange. Feynman's router equations were in terms of variables representing continuous quantities such as "the average number of 1 bits in a message address." I was much more accustomed to seeing analysis in terms of inductive proof and case analysis than taking the derivative of "the number of 1's" with respect to time. Our discrete analysis said we needed seven buffers per chip; Feynman's equations suggested that we only needed five. We decided to play it safe and ignore Feynman.
The decision to ignore Feynman's analysis was made in September, but by next spring we were up against a wall. The chips that we had designed were slightly too big to manufacture and the only way to solve the problem was to cut the number of buffers per chip back to five. Since Feynman's equations claimed we could do this safely, his unconventional methods of analysis started looking better and better to us. We decided to go ahead and make the chips with the smaller number of buffers.
Fortunately, he was right. When we put together the chips the machine worked. The first program run on the machine in April of 1985 was Conway's game of Life.
Look, I know that PI is defined in a certain way. That doesn't mean we know why it comes out just so -- why the particular numerical value is just what it is.
Of course we can show that if we sum the series expansion that comes from the definition, the answer is 3.14... But that doesn't tell us why the first digit isn't say 2. Or 8. There are no further theories to that. It's pretty much an experiment: you take a formula for PI, you compute, and you get a number. So, while we certainly know that if you set things up just so, and derive what the value of PI might be, starting with some axioms and definition being one half the ratio of circle's circumference to radius, and you get the value.
But had we had a real understanding of things, we could take a bunch of say starting digits of PI, and could actually tell that it is such a "special" number -- I'll get to what I mean by special in a second. Right now we can only compare to a list of "special" numbers. If you told all about how PI is derived to someone who never saw a part of the computed value itself, they'd have no way of telling that 3.1415... is PI -- they'd have to go through the motions of calculating it first, and then do the comparison and say "hey, the digits are same".
The way I see that PI is "special" is that it is a condensation of a whole lot of things. We start with a certain metric that works in particular type of a space -- that is, we have to define what distance is, and how to measure it, and that points of the object are equidistant. And this choice -- of metric/space and object -- is embedded in the number PI.
One can then think that real understanding would come if we could, given some number, find some structure in it that would tell us: OK, this number comes from some measurements in a space like so, with such a metric, of an object with infinitely many points equidistant to to one point. Maybe even we could enumerate such numbers, to find interesting spaces and metrics, or even interesting kinds of objects. But we have no such understanding at all, it seems.
We have a better understanding of the structure of integers -- we have, after all, better ways of testing for primality than using sieves (enumerations), we can do factorizations, etc.
But for real numbers, we're pretty much like newborns. Or else I missed some big discoveries, that is.
I'm serious. How would you even start an argument about PI not being a physical constant? It's really just a matter of definition, and in that sense there's no argument.
But we say that physical constants are some things we measure, and other seemingly fundamental things we can measure are not (like PI). PI can be of course measured to a good few digits by manufacturing a sphere or a disk/cylinder, and then measuring the circumference and radius. We then also have mathematical theories that can come up with PI to arbitrary accuracy, but that's just a bonus. We don't know -- maybe we will come up with similarly good theories for other things that can be measured, say the fine-structure constant.
We really don't know how closely the physics of our Universe are coupled with the structure we see in the mathematics. It's kind of philosophical, but we "discover" things in mathematics. So what is this thing that we discover then -- where is it. In our minds only? Or is it really just our minds picking it up, noticing it.
So I can't really say anyone can quite wrap his/her mind around it. I'd go further: anyone claiming to be able to do so is quite a kook. It's like claiming to understand why quantum mechanics or gravitation behaves just like so. We have no clue *why* it behaves just like so, exactly like we have no clue what to make of the value of PI.
We know how to apply all this knowledge, but we know of no "ulterior motive" for it. Certain phenomena can be inferred from other, more fundamental ones -- say Bernoulli effect is just a manifestation of laws of conservation intertwined with laws of dynamics -- so we can say why we see the Bernoulli effect. But we can't say why the quantum mechanical behavior of the atoms that make up the medium is just so -- we know no more basic stuff to explain that. We just observe it to be so, but can claim no further insight.
Same with PI: we have no clue where it came from. I don't claim we need to have such a clue -- but please, don't claim more insight than we actually possess. We are pretty clueless and IMHO that's what's exciting: there's still plenty of stuff to discover.
I'd think noone cares if this stuff is admissible in court. If there were ever trials relevant to those leaked documents, the court would subpoena the originals anyway, right?
You have no fucking clue how wartime intelligence works, do you?
What a fair example of circular logic:
1. It's wartime (supposedly, but let's ignore that for a moment).
2. Superiority is to be maintained.
3. Shit happens, and is embarrassing.
4. But (2), ergo more shit will happen.
5. It gets even more embarrassing, and were it to be uncovered we lose superiority and put our fine soldiers at harm.
6. But (2), ergo we must keep it all covered up.
7. Goto 3
The problem I see is that this is all but a self-propagating fallacy. I'm not a radical pacifist, but maybe, just maybe, we ought not to be wolves and make sure shit doesn't happen, that we won't be embarrassed by all of it? There were surely a few embarrassing moments on the Allied side during WWII, but on the balance I think it was minuscule compared to what's going on today. I'm glad I'm getting almost all of my federal taxes back. It'd feel quite bad to pay for any of that...
I think that it's one of those big things that we just don't quite can wrap our minds around. What does it mean that there's a certain structure and properties to geometries, that topology of things looks just like so and not some other way, etc? Do those things depend at all on the universe we live in? Would, somehow, someone in another universe find, that PI has a different value even though it's not a physical constant as far as we can tell? And maybe we're just mistaken and it is really a physical constant?
I'd say that things would need to be so "different" that if PI had a different value, then also the Jones polynomial for knots would look different -- things would have to be just fundamentally different at a low level -- so it seems to me.
Here we go again with people thinking that the paper paradigm applies to the digital world.
How on Earth do you return digital documents? Do you scrape the oxide layer off the hard drives, put it in a little vial, mark it with volume mount point(s) and put it into an envelope addressed to Pentagon? Oh, yes, I know, you first print out the directory listing (like we used to do with the floppies), tape it to the vial, then scrape, fill the vial and ship.
As for the further documents -- they better watch out, because WikiLeaks may just give up and publish all of the unredacted stuff just to preserve it.
As for WikiLeaks somehow "embarrassing" the U.S. military: waitaminuzel here. Did WikiLeaks compel the military to do all the embarrassing stuff? No? Then well, maybe it was better the taxpayers knew what their money is spent on, huh?
That's spooky. I read The Onion every now and then, but I missed that article. It's not very funny now, though. I give the author props for seeing Bush for who he really was.
I always keep saying that eye-movement control inputs should be a last resort -- for people with things like ALS where you have nothing else left (in ALS you usually have blinks, though, and those are easier to reliably sense).
Eye movements partake in acquisition of visual input. If you have to control things with eye movements, you usually do it at the cost of not seeing what you'd ordinarily see.
As for the interception idea of yours: said screen would have to have a very good refresh rate. For visual exploration, our visual system is pretty much a snapshot taker. When your eyes move from point to point, they do it in a swift motion called a saccade. During a saccade, the image slides on the retina, and the blurred image is suppressed -- we don't get to see blurs several times per second. If you have a tracking fundus imager (it sees what's on the retina in spite of the eye moving around), a third person observer can see what's projected on the retina and see the blurs. A saccade triggers the masking, then once the image stabilizes, a snapshot is taken.
There are simple experiments to demonstrate this effect. You are in a moving train, and you look at the adjacent tracks. Once the train moves fast enough, the tracks are blurry. You then attempt to look forward and backward along the track. As you do, there are points at which the saccadic velocity exactly cancels the track's motion on the retina, and you see the track clearly -- ties and trash and all. Or look around a quickly spinning ceiling fan.
If you could intercept eye muscle signals while keeping the eyeball stable, you could certainly shorten the time it takes to acquire the image -- the saccadic masking is ready to cease and accept a stable image much before the saccade would end.
But you don't really need to cut any nerves. You can let the eyes move, and instead use a tracking projector system that projects stable image on the retina no matter how the eye moves. This IIRC has been done. You can estimate the saccadic motion's end position (saccadic amplitude) in the early 1/3 of the saccade, so you could probably shorten the time it takes from beginning of the saccade to acquisition of the image.
The only thing is: it would be pointless. It takes a good time (150+ ms) to process the image once it's acquired by the visual system's imager front-end, so even if your tracking projector could be omniscient w.r.t. saccadic target, and stabilize a target image on the retina with no latency at all, it wouldn't improve your visual system latency by a whole lot -- it'd be less than a factor of two.
Eye-movement control for the masses is a rather silly idea, I bin it with, say, having answer to the "meaning" of life (everyone feels like knowing that would be "good"). Both things are pretty useless -- the first one because it trades off something our visual system is good at (vision) for a much less accurate version of what our limbs can do better (providing control input). The latter -- well, what the heck would you do with an answer to that? Just think about it.
I completely understand what you're saying, but you list some very specific circumstances. A cal lab nominally is supposed to calibrate, not repair, so they'll simply give up when they use up the adjustment range and they're still out of spec. That's hardly rejecting "due to it being fixed" -- any number of things, including simple passage of time, can make an instrument do this. They won't try to understand why it doesn't work, just that following the cal procedure (usually with lots of in-house clarifications/improvements) the tech gets stuck at a point and that's that.
For older equipment, some cal labs offer calibration bundled with repairs, so that they'll try to fix precisely what you're talking about (when it's still fixable).
Quite often it's very obvious that someone was "working" on an instrument, and you end up undoing most of such "fixes" -- I agree there. This can potentially take many billable hours, so if they are given a go to repair, they can love such jobs.
You have to admit though, that the whole thing is extremely user-friendly even when jailbreaking. No stupid yellow pop-up ActiveX warnings, just tap here, slide there, and off you go. I wonder how much Apple influence was there when the UI was designed for this jailbreak. Compare how nice it looks next to most PC-based cracks/hacks that one can download. I'm half-serious here.
I'd say that a decent scope must have at least the following:
- zero hold-off time (cheap CCD-based scopes like TDS1k series take maybe a hundred sweeps a second, any scope that can do faster than 100k sweeps/s is either in $10k+ range or is analog)
- very good input overload recovery times (100ns) -- there are no production scopes AFAIK that can offer that, you need a used sampling scope/plugin for that, or you need to look at some LTC appnotes ;)
- proper antialiasing
Once you "splurge" on those, you may as well have a decent ADC to give you enough dynamic range to offer rudimentary spectrum analysis/receiver functionality.
To get the basics going, you need a working SDR receiver -- you just buy the thing. My coworker recently got one -- the post-processor card with USB output (Cyclone II based), the sampler card with Cyclone III and 16 bit 160 MSPS LTC ADC, and a backplane and power supply. The first two were assembled.
This is quite a decent *receiver*, it has a dynamic range that's pretty much unheard of on oscilloscopes. Would make a half-decent spectrum analyzer, too. Once you add a DC-coupled front end, you may as well have an attenuator with two or three settings (say 10V/div, 1V/div, 0.1V/div), the ADC's resolution is enough to deal with it. Testing it can be done using itself -- the proof is in the pudding, I mean in the stuff you acquire.
Repurposing say a 7A13 plugin for input conditioner can be done without an oscilloscope, a DVM is all you need to get operating points right, and then you watch the signals using the device itself.
Hardly a multi-year R&D project. If you're handy with VHDL, it can be a couple weekends worth of work -- and remember, most of the hardware is already there for you.
Many DAQ cards are notorious for their aliasing problems. You'll be lucky if you find a 2nd order lowpass before the ADC. I've seen cards with 250kHz sampling rate (at 16 bits) that are significantly sensitive (think 30dB down) to stuff at 10MHz. Those are very good -- that is if you want to find cables that have good (low) common mode to differential mode conversion ratios.
What lab will reject an instrument "due to it being fixed"?! Most contemporary instruments are calibrated without opening the covers, you could literally replace everything inside and just make it emulate the original instrument and no one would be any wiser.
Probably the best deal would be to get a digital radio. If you can live with ~150Ms/s (a tad slow, but hey), then a cheap thing to do would be to get a digital radio (SDR) system. Say Mercury SDR. Those things typically have a good, 16 bit 100+Ms/s ADC front end, feeding into an FPGA that can do a lot of processing goodies, with low noise, and you should be able to hook up a Tek 7k plugin as a front-end after a few tweaks (simply to get going). You can get everything for $700. You have open source software, full documentation, and you can put a lot of very interesting signal processing on the FPGA. Keeping sampler's speed limitations in mind, you can otherwise easily match performance of many lower-end spectrum analyzers, and $20k+ scopes.
There are no $2k digital scopes with any decent feature set to speak of, even second-hand ones.
If you're into tweaking analog, then a Tektronix 7k mainframe with proper plugins gives you everything you may need. Heck, you can even get a simple logic analyzer for those. With *analog* zoom, no less.
One way of doing it: laptop running OS-X + a camera with live preview on the laptop. Then use afloat to overlay things. I'm a happy afloat user, and it has helped tremendously with some CAD applications that don't let you overlay graphic files as backgrounds for tracing.
You are talking about 40% duty cycle on the fan. I'm not talking about that. A GPU should not self destruct with ZERO heatsinks attached. It should run very slowly, but it should keep on running.
I say that the chip itself should be able to actively regulate (think control theory) its thermal output to match the attached thermal sinks -- both their thermal resistance, and their thermal mass. It's not a hard problem. Can be done in a few hundred transistors.
Your card's GPU chip(s) obviously don't have this "innovative" technology. I mean, come on, die overtemperature protection is built into $0.10 three pin voltage regulators (think 78L05). Obviously those have simple on-off control, but if you can spare a few hundred transistors, you can have a full blown adaptive PID-like control, with outputs that control clock scaling and computational unit allocation. On a GPU, you can scale thermal output by slowing down the clock, and by idling computational units. Heck, you should be able to scale it down by almost two orders of magnitudey. That should allow a bare GPU to run -- very slowly -- with no heat sink on.
I agree. Thermal engineering bites even big names in the ass. Repeatedly. MacBook Air, anyone?