Before I go on with the actual reply, I just though I would trow in the Transmeta Reality Check. Here it comes :
The Transmeta Reality Check : Any processor can run any other processor's machine code.
Any processor (ie. not just the Crusoe). Case in point: one can run SNES rips of cardrige on any pc no problem. All is that needed is an interpretor or a JIT. Now, JIT's are not that hard to understand : think of them as second generation interpretor that don't suffert from a x10 speed hit. Quickly put, they work as on-the-fly optimizing compilers. The catch is they are even harder to write. So please people, stop expecting that Transmeta will write dozens of them in the upcomming year, it's not happening.
We've all seen the reviews of the P4's lackluster performance, until apps are recompiled... well, Transmeta CPUs, in theory, doesn't suffer from that problem!
It doesn't suffer that problem because it suffers a greater `problem` : JITing is not exactly free (performance-wise). Otherwise you could think of Intel writing a JIT to convert from P3 code to P4 code on the fly and distribute it along with the chip.
Remember, the rationale behind the Crusoe is to save transitors by delagating to a software module the task of breaking-down the ridiculously complex x86 machine code.
And of course less transitors means less power used, less heat produced, less pricey chip, no need for a blow fan, and some silicon-space left available to embeed a pci bus chipset.
All very good things for a sub-sub-portable processor.
Yes, one could argue that the extra transistors could be used to boost the speed of the chip, ultimatly compensate for the cost of the JITing and pull ahead of the speed race. I wonder weither some back-lab at Transmeta isn't trying to do just this.
Here in Quebec a distinction is made between person-to-person contract and user-agreement.
A large number of clauses are forbiden from the later. For example, if someone breaks into my box and cream my dsl-connexion, I don't have to pay for the over-limit megabytes of transfert. The law forbid to discard extra-contractual reponsability unless the contract is person-to-person.
The principle is fairly ovious. Pretty much the same deal as for spam mail: during the discussion leading to the sale, the company has so little work to do in comparison with the user (they amortize the legal composition fee over all sales), it just doesn't serve the invible-hand properly.
It seems so natural, I'm very puzzled it has not spread.
The hypothied components of again are metabolism and programed cell-suicide.
Aging appears corrolated with the speed of your metabolism. Body fuel (atp) is very reactive, just as any good fuel should be. It seems that carring it around create regular damage that has constantly has to be repaired. However, every so often an important piece of dna gets broken which reders that particular cell les efficient, but also all it upcomming child-cells.
It is no surprise in that case that that all 100+ year elders describe they diet as 'frugal'.
The second component relates to a protein that repairs the buffering tips of the dna string after each copy. That particular protein is only present in the 'immortal' cells, sperms, ovulas and associated gametes. In other cell, copying strips some length of the buffers each time until it eats important dna code.
Lowering the body temparature stops problem one and, since a doubt cells can split in such cold conditions, would also stop problem two: effective body suspention.
As far as consciousness goes, if your brain can't process, you can't be concious. It would fell similar to passing out on boze: you would be terribly confused about where you are, but with a reason.
The court below expressly sought to try to adapt existing copyright provisions to the new realities of Internet technology. In doing so, the court ignored the counsel of the Sony Court (464 U.S. at 431-432) and this Court in Diamond Multimedia that extending copyright protections in response to new technologies should be left to Congress.
This is a very reasonable statement from the court people. It reconize that democratic poolings where designed to steer and bend laws into a citizen-serving state, even through times of technological and sociological change.
This kind of statement raise my confidence in the court system. Its redundancy makes it a robust and trustable decision maker.
The problem is that in doing the Right Thing, the court is forwarding the problem to the very democatic pools which brough us the DMCA and other similarly loved relatives. The court is forwarding the problem to the module with the worst design.
- all the imperative feature you needed, including those unsafe operations like unchecked array reads.
- all the tricks of functionnal programming (like higher order function, ie function composition)
- The type system is the most useful I've ever seen by far. It also catches more bugs. Think of the last 20 ClassCastException crashes, ocaml's type system would have got then at compile time. As a mater of fact, most code I write work the first time they get through the type checker. You should give it a try, it is quite a felling.
- The type system is completly unclutering since types are infered for you. Writing in ocaml feels as light as scripting (typicaly untyped languages).
- Ocaml is very fast (C-fast) Check out those banchmark results They include Python:18.906secs, Perl:12.797secs, C: 4.855secs, Ocaml:4.803secs.
Ocaml is ready for prime time. The only things it misses is some big buck company to market it.
-
Re:Why open source is nice, part LXXVIII
on
Mattel Spyware
·
· Score: 1
I see two origins for our problem those days:
"Not knowing the law is not an escuse from breaching it" and the fact it takes at least three years of Law studies in college to learn it is quite an amazing contradiction of our society. The only reason why it work is that the law is keep in-sync with what people expect from it.
Now, people expect for now expect big compagnies that do unreasonable use of their computer, of their bandwidth, and of click-through licence to be punished. But this is not happening.
We have to get the law back in sync with what people expect.
I can only do an inspection of the car I buy to the extent of my competance, which is not much. I trust the courts the beat on GM if later the car I bougth is found to be defect or plainly dangerous. That why I can buy a car in peace without becoming a skilled macnical engineer first.
-
This is just like Take-home exams
on
Laptop Exams?
·
· Score: 2
Think about it. This exams format is in many way just like a take-home exams. Hell, I did very valid take homes in Complexity Theory class and Numerical Analysis.
I see the time frame is more restricted in this case, and also the bandwidth between the student is limited. I don't see how this make up for a wholy different kind of tests though.
oh yeah, and all this shit about how laptops split the poor from the rich is irrelevant to this story. This discussion happed when universities started first considering those kind of policies.
Sorry, I do assume something here: The only admissible attack to NP-complete problem and undecidable problem is estimations. This is also what happens when so start coding for one of those problems, you have come up with some estimation heuristic.
Any other approaches won't realy work for NP-complete problem (not until the arival of the quantum computer at least); even more so for undecidable problems. This is what brain excel at: cooking up heuristics. This faculty even has a name of its own: Intuition. Moreover, if you spend enough time developping your intuition of a subject, you will get very good at focusing your thinking time on the good solution threads.
I would say this is what realy can out of the Kasparov vs Deep blue chess match. It gave us an idea of how much computation time we save ourselves by being very good heuristic cooks.
Almost everything that a compiler need to do to optimize code structure un undecidable. Human brains kick-ass at undecideble problems. Almost anyting that pertain to instruction scheduling is NP complete. Human brains kick-ass at NP-complete problems
Once a programmer has infered some property about his program, he needs a highi-level language to express his solution to the compiler. ML's typing system is the best thing I know that acheives that.
Once a program has figure a fast instruction ordering, he need a low-level language to express his solution in. We all know what kind of language does that.
But I don't know any languges that combines both.
As for todays' question : if JITs looks cool, and one time compiles looks boring, think of compiling n-times with a profile session in between and you will get much better result. (all the sugar, no r.time JIT overhead)
yep, totaly. And the fact that they are not talking about that upsets me quite a bunch.
Either: - they are actualy dumb and totaly blinded by their wish of being x86 compatible or : (maybe more likely) - their are not saying anything to not divert the attension from the x86-compatible hype. In which case their marketers are just as evil/more as any other.
Intel and AMD can make fast bloated chips. powerARM makes a fast low power consumption chip Transmeta makes a fast low power consumption chip + they have enough good compiler people to write a state-of-the-art fast JITs for it.
A figure the ARM people should get into compilers. In fact, all those hardware people should get into compilers. It is a wonderfull way to get over the backward compatibility problem
Well... if you throw a crusoe chip at your fuel cell, you will be able to make the battery that much smaller. If the idea of having your portable run for 3 days doesn't apeal to you, try thinking of running a 14" ldc pad-like computer for a day of a AA-sized battery. Sounds very good to me. Very lightweight too.
Anybody has an idea of how flexible the translation phase is? Say, could it be reprogramed to interpret PPC machine code. Do you think it could do java bytecode? Maybe Transmeta can re-microcode it to do it easily. I'm not sure which answer what answer I want to get. Having it too flexible basicaly means it is a general purpose processor, and as such it would really be any different than using Java's Hotspot.
I remember some newspaper editing a photo of Clinton to make him stand closer to the woman walking by him. They were blamed big time by the industry. In particular by those that infered Clinton was having an affair from the picture (This was before it was reveal he actually had an affair..) They claimed they did it only to make the picture fit in the frame and promised to never do it again.
Slashdot Mirror
Programming is a Craft: functional work where esthetics have a place.
Map and level design are Technical Arts: esthetic work constrained by the techniques. Sculpting come to mind as another example.
So there, you may now go in peace.
-
Has anyone made an xkoules computer player yet? That always wins?
-
The longing search for a non-zero-sum game eh? That *must* be why people are playing so much solitaire!
-
Somebody has to tell them to stop before they try to go to 1/3 of an atom. Otherwise there won't be anyone left alive to talk about it.
-
Mommy! Mommy! Can I have a glow-in-the-dark-brain for chrismass too?! oh please! oh please! oh please!
-
Before I go on with the actual reply, I just though I would trow in the Transmeta Reality Check. Here it comes :
The Transmeta Reality Check : Any processor can run any other processor's machine code.
Any processor (ie. not just the Crusoe). Case in point: one can run SNES rips of cardrige on any pc no problem. All is that needed is an interpretor or a JIT. Now, JIT's are not that hard to understand : think of them as second generation interpretor that don't suffert from a x10 speed hit. Quickly put, they work as on-the-fly optimizing compilers. The catch is they are even harder to write. So please people, stop expecting that Transmeta will write dozens of them in the upcomming year, it's not happening.
We've all seen the reviews of the P4's lackluster performance, until apps are recompiled... well, Transmeta CPUs, in theory, doesn't suffer from that problem!
It doesn't suffer that problem because it suffers a greater `problem` : JITing is not exactly free (performance-wise). Otherwise you could think of Intel writing a JIT to convert from P3 code to P4 code on the fly and distribute it along with the chip.
Remember, the rationale behind the Crusoe is to save transitors by delagating to a software module the task of breaking-down the ridiculously complex x86 machine code.
And of course less transitors means less power used, less heat produced, less pricey chip, no need for a blow fan, and some silicon-space left available to embeed a pci bus chipset. All very good things for a sub-sub-portable processor.
Yes, one could argue that the extra transistors could be used to boost the speed of the chip, ultimatly compensate for the cost of the JITing and pull ahead of the speed race. I wonder weither some back-lab at Transmeta isn't trying to do just this.
-
Here in Quebec a distinction is made between person-to-person contract and user-agreement.
A large number of clauses are forbiden from the later. For example, if someone breaks into my box and cream my dsl-connexion, I don't have to pay for the over-limit megabytes of transfert. The law forbid to discard extra-contractual reponsability unless the contract is person-to-person.
The principle is fairly ovious. Pretty much the same deal as for spam mail: during the discussion leading to the sale, the company has so little work to do in comparison with the user (they amortize the legal composition fee over all sales), it just doesn't serve the invible-hand properly.
It seems so natural, I'm very puzzled it has not spread.
-
The hypothied components of again are metabolism and programed cell-suicide.
Aging appears corrolated with the speed of your metabolism. Body fuel (atp) is very reactive, just as any good fuel should be. It seems that carring it around create regular damage that has constantly has to be repaired. However, every so often an important piece of dna gets broken which reders that particular cell les efficient, but also all it upcomming child-cells.
It is no surprise in that case that that all 100+ year elders describe they diet as 'frugal'.
The second component relates to a protein that repairs the buffering tips of the dna string after each copy. That particular protein is only present in the 'immortal' cells, sperms, ovulas and associated gametes. In other cell, copying strips some length of the buffers each time until it eats important dna code.
Lowering the body temparature stops problem one and, since a doubt cells can split in such cold conditions, would also stop problem two: effective body suspention.
As far as consciousness goes, if your brain can't process, you can't be concious. It would fell similar to passing out on boze: you would be terribly confused about where you are, but with a reason.
-
The court below expressly sought to try to adapt existing copyright provisions to the new realities of Internet technology. In doing so, the court ignored the counsel of the Sony Court (464 U.S. at 431-432) and this Court in Diamond Multimedia that extending copyright protections in response to new technologies should be left to Congress.
This is a very reasonable statement from the court people. It reconize that democratic poolings where designed to steer and bend laws into a citizen-serving state, even through times of technological and sociological change.
This kind of statement raise my confidence in the court system. Its redundancy makes it a robust and trustable decision maker.
The problem is that in doing the Right Thing, the court is forwarding the problem to the very democatic pools which brough us the DMCA and other similarly loved relatives. The court is forwarding the problem to the module with the worst design.
-
- all the imperative feature you needed, including those unsafe operations like unchecked array reads.
- all the tricks of functionnal programming (like higher order function, ie function composition)
- The type system is the most useful I've ever seen by far. It also catches more bugs. Think of the last 20 ClassCastException crashes, ocaml's type system would have got then at compile time. As a mater of fact, most code I write work the first time they get through the type checker. You should give it a try, it is quite a felling.
- The type system is completly unclutering since types are infered for you. Writing in ocaml feels as light as scripting (typicaly untyped languages).
- Ocaml is very fast (C-fast) Check out those banchmark results They include Python:18.906secs, Perl:12.797secs, C: 4.855secs, Ocaml:4.803secs.
Ocaml is ready for prime time. The only things it misses is some big buck company to market it.
-
I see two origins for our problem those days:
"Not knowing the law is not an escuse from breaching it" and the fact it takes at least three years of Law studies in college to learn it is quite an amazing contradiction of our society. The only reason why it work is that the law is keep in-sync with what people expect from it.
Now, people expect for now expect big compagnies that do unreasonable use of their computer, of their bandwidth, and of click-through licence to be punished. But this is not happening.
We have to get the law back in sync with what people expect.
I can only do an inspection of the car I buy to the extent of my competance, which is not much. I trust the courts the beat on GM if later the car I bougth is found to be defect or plainly dangerous. That why I can buy a car in peace without becoming a skilled macnical engineer first.
-
I see the time frame is more restricted in this case, and also the bandwidth between the student is limited. I don't see how this make up for a wholy different kind of tests though.
oh yeah, and all this shit about how laptops split the poor from the rich is irrelevant to this story. This discussion happed when universities started first considering those kind of policies.
-
The only admissible attack to NP-complete problem and undecidable problem is estimations.
This is also what happens when so start coding for one of those problems, you have come up with some estimation heuristic.
Any other approaches won't realy work for NP-complete problem (not until the arival of the quantum computer at least); even more so for undecidable problems. This is what brain excel at: cooking up heuristics. This faculty even has a name of its own: Intuition. Moreover, if you spend enough time developping your intuition of a subject, you will get very good at focusing your thinking time on the good solution threads.
I would say this is what realy can out of the Kasparov vs Deep blue chess match. It gave us an idea of how much computation time we save ourselves by being very good heuristic cooks.
-
Once a programmer has infered some property about his program, he needs a highi-level language to express his solution to the compiler. ML's typing system is the best thing I know that acheives that.
Once a program has figure a fast instruction ordering, he need a low-level language to express his solution in. We all know what kind of language does that.
But I don't know any languges that combines both.
As for todays' question : if JITs looks cool, and one time compiles looks boring, think of compiling n-times with a profile session in between and you will get much better result. (all the sugar, no r.time JIT overhead)
-
Either :
- they are actualy dumb and totaly blinded by their wish of being x86 compatible
or : (maybe more likely)
- their are not saying anything to not divert the attension from the x86-compatible hype.
In which case their marketers are just as evil/more as any other.
-
powerARM makes a fast low power consumption chip
Transmeta makes a fast low power consumption chip + they have enough good compiler people to write a state-of-the-art fast JITs for it.
A figure the ARM people should get into compilers. In fact, all those hardware people should get into compilers. It is a wonderfull way to get over the backward compatibility problem
-
Well... if you throw a crusoe chip at your fuel cell, you will be able to make the battery that much smaller. If the idea of having your portable run for 3 days doesn't apeal to you, try thinking of running a 14" ldc pad-like computer for a day of a AA-sized battery. Sounds very good to me. Very lightweight too.
-
Anybody has an idea of how flexible the translation phase is? Say, could it be reprogramed to interpret PPC machine code. Do you think it could do java bytecode? Maybe Transmeta can re-microcode it to do it easily. I'm not sure which answer what answer I want to get. Having it too flexible basicaly means it is a general purpose processor, and as such it would really be any different than using Java's Hotspot.
-
They claimed they did it only to make the picture fit in the frame and promised to never do it again.
To bad I can't remeber the details.
-