Although these new processors have their own instruction sets, it seems they're not going to be available to programmers. And although programs can be compiled to any of the emulated instruction sets, these instruction sets are not designed with translation in mind, and will not be optimal. So, do Transmeta have any plans to release a "nominally native" instruction set for these CPUs ?
Since the technology of this beastie is patented, will Transmeta now be the evil monopoly for the next decade ?
Since the code is compiled at run-time, does this mean that processing will involve the same level of complexity, and hence bugginess, as the average compiler ?
I've been racking my brains for ways to make the battle room look convincing, but to no avail so far.
Well this is Card we're talking about, so maybe a Deus Ex Machina ?
I expect you can do zero-g with people on the end of counterbalanced or servo-controlled[0] arms. Of course, that means finding an actor who can actually DO this stuff....
[0] Effectively a long armed industrial robot with its feedback set up not to hold position, but to just apply one body-weight worth of force upwards at all times. Remove from film digitally, naturally.
One of the reasons that I appreciate Mr Mills work is that it resembles my own system for extracting energy from the collapse of the false vacuum. I'm just off to try the prototype now.
The theory behind his energy source is quite interesting. He proposes that the everyday electronic state of hydrogen is not the lowest energy state, but a metastable state which is long-lived because the transitions to lower states are not radiative. His apparatus allegedly works by using a transition of equal energy in potassium to accept the energy from the hydrogen.
I don't buy it, but what the hell, it's worth trying.
Mr Mills has quite a few Australian patents, but you wouldn't want to suffer the Java 3270 emulation needed to view them. Below are the ones which are WIPO.
Having said that, the website still reads like pure sucker-bait, and the vast majority of the patent still reads as though the author has been sniffing mercury. And yet, if you carefully pick and chose your claims, the patent can still be arranged to make a description of a plausible system.
What it appears that they're doing is turning one of the wires being used for an AC transmission line into something resembling a type of open waveguide, a G-line.
Wow ! That's simply amazing. I'd never even heard of G-Lines before, but what they seem to involve is a surface plasma wave on the wire ! In the light of this, even the stupid patent makes sense, i.e. the wave really is "acoustic" in that it's partially an electron-density wave.
And now, I'm off to piss myself laughing at everyone whose "superior" physics knowledge made them dismiss this idea (myself included). The basic idea has already been tried, it's already known to RF engineers, it already works.
here's a durability issue that i didn't see addressed on their (tech-poor) web site: they say the disks use flourescent dyes. at a guess these will be nice tunable organic dyes. such materials don't tend to live long, especially when illuminated.
are these disks going to last more than a few months?
Since these disks have 112Gbits per layer, each bit is only going to be illuminated a 112,000,000,000th of the time.
I know nothing about this technology, so I have an irresistable urge to lecture about it on/.
How it works is this:
The CD consists of multiple layers which are individually pressed in the usual CD way, then stuck together afterwards. The layers have different fluorescence frequencies, and a single frequency laser stimulates all the fluorecent layers at once. If I read their website right, the pressing process makes the usual pits in the plastic of each layer, and a fluorescent material is then put into the pits. At any rate, the data is encoded by having different thicknesses of fluorescent material for ones and zeros.
Because the fluorescent layers and the intervening glue layers all have identical optical characteristics from the point of view of the laser beam, and because the ones and zeros also look virtually identical to the beam, the medium appears almost completely homogenous, and doesn't scatter or refract the laser, so only the spot you're interested in gets illuminated, and the beam remains parallel right through the medium. This is what allows the highest possible spot densities, and the very large numbers of layers.
The writable version of this probably/will/ need multiple frequency lasers, to cause photochemical changes in each layer separately.
It may be that the ideal approach to the problem is to flood the Patent Office with prior art of every conceivable sort, in a form they can quickly and easily use to test patent applications.
A possible better idea, inspired by the above, is to simply flood the Patent Office. It's relatively cheap to get a patent through the patent office if you don't care about its validity or usefulness, so it would probably be practical for the programming community to churn out a few hundred thousand junk patents per month. Would that be enough to clog up the process ? Would it bring the patent office to their senses ?
Europe has a unified standard "ETS 300 328" making 2.4 to 2.4835 GHz available. Unfortunately in France the Gendarmerie use the bottom of the band, leaving only 2.4465 Ghz upwards available. I think it's explained here.
Amiga enthusiasts could try to pressure developers into improving Linux to a point where it would be an acceptable replacement for the Amiga OS. I can think of a couple of obvious items, like the handling of removable media
How did the Amiga handle removable media ? It's easy to see that the removable media handling in Linux is an absolute idiocy, even by UNIX standards. But what's the right way to do it ?
Slashdot Mirror
Code morphing is compilation, not interpretation. The x86 (or whatever) code is translated into an equivalent program for the Crusoe.
Although these new processors have their own instruction sets, it seems they're not going to be available to programmers. And although programs can be compiled to any of the emulated instruction sets, these instruction sets are not designed with translation in mind, and will not be optimal. So, do Transmeta have any plans to release a "nominally native" instruction set for these CPUs ?
Since the code is compiled at run-time, does this mean that processing will involve the same level of complexity, and hence bugginess, as the average compiler ?
Well this is Card we're talking about, so maybe a Deus Ex Machina ?
I expect you can do zero-g with people on the end of counterbalanced or servo-controlled[0] arms. Of course, that means finding an actor who can actually DO this stuff....
[0] Effectively a long armed industrial robot with its feedback set up not to hold position, but to just apply one body-weight worth of force upwards at all times. Remove from film digitally, naturally.
Not so much "out of focus", more like "blind spot". Weird.
I hope the examples on this page aren't typical - the original jpeg seemed to do a much better job - no "out of focus" and no loss of small details.
One of the reasons that I appreciate Mr Mills work is that it resembles my own system for extracting energy from the collapse of the false vacuum. I'm just off to try the prototype now.
I don't buy it, but what the hell, it's worth trying.
Mr Mills has quite a few Australian patents, but you wouldn't want to suffer the Java 3270 emulation needed to view them. Below are the ones which are WIPO.
HYDROGEN CATALYSIS POWER CELL FOR ENERGY CONVERSION SYSTEMS
INORGANIC HYDROGEN COMPOUNDS, SEPARATION METHODS, AND FUEL APPLICATIONS
LOWER-ENERGY HYDROGEN METHODS AND STRUCTURES
APPARATUS AND METHOD FOR PROVIDING AN ANTIGRAVITATIONAL FORCE
ENERGY/MATTER CONVERSION METHODS AND STRUCTURES
ENERGY/MATTER CONVERSION METHODS AND STRUCTURES
ENERGY/MATTER CONVERSION METHODS AND STRUCTURES
ENERGY/MATTER CONVERSION METHODS AND STRUCTURES
APPARATUS AND METHOD FOR PROVIDING AN ANTIGRAVITATIONAL FORCE
ENERGY/MATTER CONVERSION METHODS AND STRUCTURES
MAGNETIC SUSCEPTIBILITY IMAGING (MSI)
A review of his book with a few useful footnotes is here
Having said that, the website still reads like pure sucker-bait, and the vast majority of the patent still reads as though the author has been sniffing mercury. And yet, if you carefully pick and chose your claims, the patent can still be arranged to make a description of a plausible system.
PS: Mystery explained ? [1] [2]
Wow ! That's simply amazing. I'd never even heard of G-Lines before, but what they seem to involve is a surface plasma wave on the wire ! In the light of this, even the stupid patent makes sense, i.e. the wave really is "acoustic" in that it's partially an electron-density wave.
And now, I'm off to piss myself laughing at everyone whose "superior" physics knowledge made them dismiss this idea (myself included). The basic idea has already been tried, it's already known to RF engineers, it already works.
are these disks going to last more than a few months?
Since these disks have 112Gbits per layer, each bit is only going to be illuminated a 112,000,000,000th of the time.
PS: Nice pun.
How it works is this:
The CD consists of multiple layers which are individually pressed in the usual CD way, then stuck together afterwards. The layers have different fluorescence frequencies, and a single frequency laser stimulates all the fluorecent layers at once. If I read their website right, the pressing process makes the usual pits in the plastic of each layer, and a fluorescent material is then put into the pits. At any rate, the data is encoded by having different thicknesses of fluorescent material for ones and zeros.
Because the fluorescent layers and the intervening glue layers all have identical optical characteristics from the point of view of the laser beam, and because the ones and zeros also look virtually identical to the beam, the medium appears almost completely homogenous, and doesn't scatter or refract the laser, so only the spot you're interested in gets illuminated, and the beam remains parallel right through the medium. This is what allows the highest possible spot densities, and the very large numbers of layers.
The writable version of this probably /will/ need multiple frequency lasers, to cause photochemical changes in each layer separately.
I don't know about you, but I think that would probibly look too cool for words.
What is that glowing thing?
Do not look into the laser with your remaining good eye.....
Er.. I mean on Slashdot, of course....
It may be that the ideal approach to the problem is to flood the Patent Office with prior art of every conceivable sort, in a form they can quickly and easily use to test patent applications.
A possible better idea, inspired by the above, is to simply flood the Patent Office. It's relatively cheap to get a patent through the patent office if you don't care about its validity or usefulness, so it would probably be practical for the programming community to churn out a few hundred thousand junk patents per month.
Would that be enough to clog up the process ? Would it bring the patent office to their senses ?
Europe has a unified standard "ETS 300 328" making 2.4 to 2.4835 GHz available. Unfortunately in France the Gendarmerie use the bottom of the band, leaving only 2.4465 Ghz upwards available. I think it's explained here.
To clear up a possible misinterpretation, I meant "removable media in UNIX is an idiocy, even by UNIX standards".
Possibly. It may be possible to refresh your brain using stem cells. And more stem cells. And yet more.
How did the Amiga handle removable media ? It's easy to see that the removable media handling in Linux is an absolute idiocy, even by UNIX standards. But what's the right way to do it ?
If you need an example of why this just isn't so, you only have to consider GIF/LZW. Paid up that $5000 for your website yet ?
Hmm. Isn't the only thing keeping MS out of the UNIX market the agreement it made with SCO when it sold Xenix ? When SCO dies, what happens then ?
There is no need for 128bit chips, now or ever.
64 bits is enough to address 18 million terabytes.
More than the number of atoms in the earth...
Wrong.
64 bits is barely enough to mmap every atom in one cubic millimetre of material, let alone the earth.
Remember, Moore's law is exponential. 32->64 bits will only take twice as long as 16->32 bits.
Expect 128 bit processors by 2015.