It doesn't have to be very fast at all if the software can break up the instructions and send them to multiple processors. This way, it can be both cheaper AND faster than a state-of-the-art Intel processor (although it may be just cheaper per processor.
As far as translating instructions, yes, almost all CPUs translate their "native" instruction set into the microcode which actually controls the CPU. Very few CPUs have the abilility to change that translation process, or change the microcode, on the fly.
Uh, dude! If the microcode is reprogramable on the fly, it's better than firmware that can be upgrading. Although yes, it will take a software upgrade to emulate a new Intel instruction set.
It's gonna be a lot cheaper per CPU, but you'll need a bunch of them to do the same work -- no more single-processor systems! Remember, in theory, a large group of 6502's can emulate a PIII faster than a PIII... if you can manage to write the software to coordinate the breaking up and reordering of instructions. Finally, a true RISC machine -- why have multiple instruction pipelines on a single processor, when you can have multiple processors with single pipelines?
Correct me if I'm wrong, but wouldn't this only be useful if you had multiple processors? This would be, however, one one of emulating a Pentium faster than the Pentium itself -- breaking up the instructions to operate on multiple processors in parallel. Instead of multiple pipelines on a single processor, why not simply a bunch of cheap processors, with a control processor that reorders instructions and assigns each instruction or group of instructions to a specific processor. Speculative branching? Have a CPU follow each branch! Sounds simply in theory, but extremely difficult to make work in practice.
1) Asynchronous execution of instructions -- instead of requiring an integral number of clock cycles, each instruction takes as long as it takes, then passes control to the next instruction.
2) FPGA-like microcode that is changable on the fly. This wouldn't really be useful unless you had a bunch of these puppies running in parallel, then you could microcode one to do DCT's one to emulate an X86, one to execute Java byte codes, etc.
Sounds like speculative execution of partially-translated instructions to me, which requires a mechanism for committing the results if the instructions are valid, and backing out if they are not. Very useful if you are doing Java Just-In-Time compiler-like translation of instructions to your native code -- which is what I've been suggesting that Transmeta is doing all along.
The question remains, of course, as to which processors instruction sets they are translating... I suspect x86 to begin with, but doing both x86 and PowerPC would be nice. And of course, translating the instructions is only half the battle; you also have to provide a compatible run-time environment (which is where I suspect Linus comes in.)
Too much of a price to pay?
on
Dear Mr. Straw
·
· Score: 2
I understand law enforcement's predicament when it comes to crypto, but it's no different than any other civil-rights vs. law-enforcement issue. Basically the crypto-issue reduces down to Search. Sure having cops rabndomly raid someone's home will prevent crime, but is it to much of a price to pay? Sure key-escrow/recovery will allow the cops to evesdrop on you and the criminals, but is it too much of a price to pay?
The problem is that any mechanism that allows the cops to intercept your communications and use them against you as part of a valid law enforcement effort may also be used by non-police. What's to stop me from alleging that a business communication from one of my competitors contains evidence of illegal activity, then bribing the law enforcement officials to disclose the contents to me, thus gaining a competitive advantage? Would Nixon have needed to send burgulars into the Watergate Hotel if he could simply tell the NSA to retreive the DNC's keys from key escrow, and intercept all their communications?
The problem here is that the massive potential for abuse here far outweighs the legitimate law enforcement needs. Besides which, if your want to see what is in an email you can always simply get a warrant and record the keystrokes as they are typed in, negating the need for anyone disclosing the key.
This is absolutely the wrong solution to the problem of unauthorized vistors. The right solution? Having few enough students so that staff can recognize them all on sight. Obviously with 1200 students, this isn't possible. In the high school I attended, it was possible to know each of the 150 students.
And of course, as somebody pointed out, most high school shootings were committed by members of the school who would have had a badge anyway. And what's to keep a student from simply keeping the badge when they get expelled? "What're you going to do if I don't turn in my badge -- expell me?"
This system fails several tests of reasonableness. It not only doesn't solve the problem in the least intrusive way, it also doesn't completely solve the problem.
The best Java IDE anyone could find was written in Java. That's a very scary thing to think about. Sun should dump the tools onto someone who cares (and who'll write the IDE in ANSI C).
If you can't write a decent IDE in Java, then perhaps Java isn't worth using as a language. A corollary to "Write once, run anywhere" should be the ability to develop on any platform that will run Java -- something that you CAN NOT do in ANSI C!
I taught a course in Java, and encouraged all my students to download Java Workshop because it's free for educational purposes. Unfortunately, it's extremely slow and a little buggy on Windows, not up to par with even Microsoft's "Visual" IDE, which after several iterations is turning into a decent product.
Java, and especially Java Beans, NEEDS a decent, freely available IDE. If Sun is going to push one, I like to see them push one more refined than Workshop.
Agreed, if you are starting at zero (or near-zero) than Linux is an attractive client solution. However, one of the main things that the hours of frustration in learning Win95 has taught most casual computer users is that learning a new OS is a maddening, frustrating, hair-pulling experience. So of course, they assume that having to learn Linux will be a simular experience. This, coupled with most people's natural resistance to change, means that forcing them to switch will not be a popular choice!
Also, Linux doesn't work exactly like Windows until it runs every shrinkwrapped software and game they've bought to run on Windows -- and due to undocumented APIs, etc. it will be quite awhile before this happens.
That being said, if you're training people on their first computer system, go ahead, teach 'em Linux. My perception is that Linux user ease of use and application available is on the verge of becoming equal to that of Windows; given Linux's more rapid rate of improvement, it should soon exceed that of Windows. Unfortunately, a slightly better solution frequently isn't enough to overcome human resistance to change.
This confirms what I've been telling friends for a while now: Linux will beat NT handily in the server market, and this will be obvious to everyone very, i.e. in 1999 or 2000 at the latest. Microsoft has been hemoraging credibility right and left over the past year, and their one-year-and-growing tardiness in delivering NT5.0, er Win2000 is going to hurt them. Unfortunately, M$ will continue to dominate the desktop client market for quite some time, due to momentum and the difficulty and expense of retraining users to use a new client platform. (No flames please, this is just my opinion.) I suspect that this market will decrease in significance as networks of "information appliances" take on more and more of the functionality of the traditional standalone desktop computer. (The PC will survive in the form of a gateway between your home network and the internet, but currently Linux is a much better gateway platform than is Windows.) Microsoft's biggest mistake was to try to position "Windows" as the solution to all possible computing needs. It is clearly architecturally and structurally deficient to scale from palmtops to supercomputers. The customers would have been better served if M$ had concectrated on client side software (the only area where a GUI is an advantage) instead of trying to be all things to all people, and failing to completely implement any of it's promises in the process.
This sounds like a smart choice by Sony -- to try to detour some open source developers into developing games for the PSX, meaning more games will be available when it ships. I'd love to do some 3D game development on a Visual Workstation running Linux myself!
You must be a student. Only in school is an author's output judged by weight, not by content. Asimov is honored not merely because he was prolific, but because he despite the volume of his work, he still had something to say in each one of them. If you're going purely by volume, you must think that 10,000 monkeys with keyboards can produce "better" prose than Shakespeare... or Katz, for that matter
It seems to me that if Compaq signed an NDA for it's discussions with Be, than ANY mention of it's talks with Be to Micro$oft would be a violation of that NDA, regardless of whether or not M$ killed the deal. So are we now going to see Be suing Compaq?
By my understanding, ALL the MP3s on MP3.com are legal, authorized cuts... so what's their rationale for killing the ad? Could it be that the established record companies can't stand even a little honest competition?
Visual Studio is good, but it's debugger sucks, since M$ is more concerned with preventing you from using the debugger to reverse-engineer M$ code than in helping you actually find and fix bugs...
The really ridiculous thing is, playboy thinks it owns the word "playmate". Doesn't almost every child in the world have a playmate? What if Excite places the banner for a daycare center tied to that keyword? Would playboy still argue that their infringing on their perverted use of the word?
Slashdot Mirror
As far as translating instructions, yes, almost all CPUs translate their "native" instruction set into the microcode which actually controls the CPU. Very few CPUs have the abilility to change that translation process, or change the microcode, on the fly.
Uh, dude! If the microcode is reprogramable on the fly, it's better than firmware that can be upgrading. Although yes, it will take a software upgrade to emulate a new Intel instruction set.
It's gonna be a lot cheaper per CPU, but you'll need a bunch of them to do the same work -- no more single-processor systems! Remember, in theory, a large group of 6502's can emulate a PIII faster than a PIII... if you can manage to write the software to coordinate the breaking up and reordering of instructions. Finally, a true RISC machine -- why have multiple instruction pipelines on a single processor, when you can have multiple processors with single pipelines?
Correct me if I'm wrong, but wouldn't this only be useful if you had multiple processors? This would be, however, one one of emulating a Pentium faster than the Pentium itself -- breaking up the instructions to operate on multiple processors in parallel. Instead of multiple pipelines on a single processor, why not simply a bunch of cheap processors, with a control processor that reorders instructions and assigns each instruction or group of instructions to a specific processor. Speculative branching? Have a CPU follow each branch! Sounds simply in theory, but extremely difficult to make work in practice.
2) FPGA-like microcode that is changable on the fly. This wouldn't really be useful unless you had a bunch of these puppies running in parallel, then you could microcode one to do DCT's one to emulate an X86, one to execute Java byte codes, etc.
The question remains, of course, as to which processors instruction sets they are translating... I suspect x86 to begin with, but doing both x86 and PowerPC would be nice. And of course, translating the instructions is only half the battle; you also have to provide a compatible run-time environment (which is where I suspect Linus comes in.)
The problem is that any mechanism that allows the cops to intercept your communications and use them against you as part of a valid law enforcement effort may also be used by non-police. What's to stop me from alleging that a business communication from one of my competitors contains evidence of illegal activity, then bribing the law enforcement officials to disclose the contents to me, thus gaining a competitive advantage? Would Nixon have needed to send burgulars into the Watergate Hotel if he could simply tell the NSA to retreive the DNC's keys from key escrow, and intercept all their communications?
The problem here is that the massive potential for abuse here far outweighs the legitimate law enforcement needs. Besides which, if your want to see what is in an email you can always simply get a warrant and record the keystrokes as they are typed in, negating the need for anyone disclosing the key.
And of course, as somebody pointed out, most high school shootings were committed by members of the school who would have had a badge anyway. And what's to keep a student from simply keeping the badge when they get expelled? "What're you going to do if I don't turn in my badge -- expell me?"
This system fails several tests of reasonableness. It not only doesn't solve the problem in the least intrusive way, it also doesn't completely solve the problem.
I forget my ID badge for work all the time, and I'm a responsible adult. How can we expect a higher standard out of a bunch of teenagers?
If you can't write a decent IDE in Java, then perhaps Java isn't worth using as a language. A corollary to "Write once, run anywhere" should be the ability to develop on any platform that will run Java -- something that you CAN NOT do in ANSI C!
Java, and especially Java Beans, NEEDS a decent, freely available IDE. If Sun is going to push one, I like to see them push one more refined than Workshop.
Also, Linux doesn't work exactly like Windows until it runs every shrinkwrapped software and game they've bought to run on Windows -- and due to undocumented APIs, etc. it will be quite awhile before this happens.
That being said, if you're training people on their first computer system, go ahead, teach 'em Linux. My perception is that Linux user ease of use and application available is on the verge of becoming equal to that of Windows; given Linux's more rapid rate of improvement, it should soon exceed that of Windows. Unfortunately, a slightly better solution frequently isn't enough to overcome human resistance to change.
This confirms what I've been telling friends for a while now: Linux will beat NT handily in the server market, and this will be obvious to everyone very, i.e. in 1999 or 2000 at the latest. Microsoft has been hemoraging credibility right and left over the past year, and their one-year-and-growing tardiness in delivering NT5.0, er Win2000 is going to hurt them. Unfortunately, M$ will continue to dominate the desktop client market for quite some time, due to momentum and the difficulty and expense of retraining users to use a new client platform. (No flames please, this is just my opinion.) I suspect that this market will decrease in significance as networks of "information appliances" take on more and more of the functionality of the traditional standalone desktop computer. (The PC will survive in the form of a gateway between your home network and the internet, but currently Linux is a much better gateway platform than is Windows.) Microsoft's biggest mistake was to try to position "Windows" as the solution to all possible computing needs. It is clearly architecturally and structurally deficient to scale from palmtops to supercomputers. The customers would have been better served if M$ had concectrated on client side software (the only area where a GUI is an advantage) instead of trying to be all things to all people, and failing to completely implement any of it's promises in the process.
This sounds like a smart choice by Sony -- to try to detour some open source developers into developing games for the PSX, meaning more games will be available when it ships. I'd love to do some 3D game development on a Visual Workstation running Linux myself!
http://www.techserver.com/story/0,1643,26978-43529 -322106-0,00.html
The republicans are having a field day with this...
You must be a student. Only in school is an author's output judged by weight, not by content. Asimov is honored not merely because he was prolific, but because he despite the volume of his work, he still had something to say in each one of them. If you're going purely by volume, you must think that 10,000 monkeys with keyboards can produce "better" prose than Shakespeare... or Katz, for that matter
If your main focus is 3D gaming, then [the PIII] is going to be worth the investment...
Uh, so I should really pay $700 extra for a chip that does what should have been done in the 3D accelerator board in the first place?
It seems to me that if Compaq signed an NDA for it's discussions with Be, than ANY mention of it's talks with Be to Micro$oft would be a violation of that NDA, regardless of whether or not M$ killed the deal. So are we now going to see Be suing Compaq?
By my understanding, ALL the MP3s on MP3.com are legal, authorized cuts... so what's their rationale for killing the ad? Could it be that the established record companies can't stand even a little honest competition?
Voodoo3 should be out by then. With a 3500, overclocked Celeron, and a fragmaster joystick, this might actually be fun...
So, my friend who always lines his hat with tinfoil to block the beams being sent at him by the government isn't as crazy as we thought?
Visual Studio is good, but it's debugger sucks, since M$ is more concerned with preventing you from using the debugger to reverse-engineer M$ code than in helping you actually find and fix bugs...
The really ridiculous thing is, playboy thinks it owns the word "playmate". Doesn't almost every child in the world have a playmate? What if Excite places the banner for a daycare center tied to that keyword? Would playboy still argue that their infringing on their perverted use of the word?
People don't want servers. People wants to play the latest and coolest computer games. Conclusion: People wants Windows!
People don't want to work. People want to feel good. Conclusion: People want to smoke crack!
Anybody know what bandwith reservation protocol they are using? Do they support IPNG?