I've talked to Mark Tilden, and seen his robots. They cannot learning anything more complex than resonance. That is definately a neat trick that many other systems have a hard time doing, but it is their only trick.
It can make some fairly simple yet effective walkers, that have no concept of what to do when they get there, or even have any concept of getting there. They just go, they don't do anything else. They are toys.
What we need is a vote per candidate. Acceptable/Not Acceptable. Under this system, the best candidate (not just who the public happens to sway with for fear of 'throwing their vote away') gets elected in a more-than-two way race. And, it's also possible to throw out an election if people get up and go to the polls and every candidate gets more Not Acceptable than Acceptable votes. But, I don't think that should bar them from any future election. The system worked once, it can work again and continue to give them no-confidence results.
It's allready getting hard to do hardware hacking with anything much more than an 8 bit processor. I think you can still get a 68000 in a 64pin DIP, but everything else is heading to QFP, TSOP, and worst of all BGA. To work with these you either have to solder wires spaced.5 mm apart or less, or somehow solder an entire array of contacts simultaneously while they sit between the board and the chip.
It's because of things like this that I have very few designs on building any sort of major computing device in the near future.
Allready processors are too fast for the rest of the system. This has been alleviated for the last decade by an increasingly complicated system of caches and chipsets. At worst you'll go throgh 3 levels of processor cache, main memory, disk cache and finally disk, for a total of 6 levels of memory. This could go on indefinately but will have decreasing returns, unless the architecture of the computer can catch up to be generally faster. SGI/Cray has done this well.
2. Megahertz == Marketing
Ever since the P2, it's been terribly obvious that Intel just develops to satisfy what the majority clueless consumer wants- a higher megahertz number. The P2 made it blatant by being inferior to the older P's when run at equal megahertz. The only benefit was that it would run at higher megahertz.
Efficiency
No x86 has been really efficent- in many ways. More gates, more watts, more space, more heat. The unfortunate predominance of x86 is leading to space robots being designed with pentiums because Intel can push through to get the chips certified. When multiprocessing becomes a necessity as clock speeds dead end, who will be able to afford the power and large case for cooling that 8-64 P[3-5]'s will need? It's absurd.
AMD claims 12 Watts, which sounds OK until compared to competition. Motorola PPC 750 (G3) at 400 MHz draws 5.8 to 8 Watts. PPC 7400 (G4) at 400 MHz draws 5.0 to 11.5 Watts and architechturally outstrips the others.
If the CPU were the only thing to run (I know, this will exagerate the differences) a battery could last up to twice as long if either the G3 or G4 spent most of its time chilling in the low range of usage. User driven tasks like word processing allow such idling to take advantage of the PPC doze and nap power conservation modes, which is appearantly how Apple can claim 6 hour battery life on the iBooks. Other tasks would certainly run it down faster by preventing either the processor or the disk from idling.
I'm CMU CS of 2001. The number I heard for my year was 17 or 18 out of 138 a ratio of 7:1. Perhaps a few have left, I can offhand name about 10 of them.
The number I've heard for the incoming freshman class (2003) is 50 to 70 female/male. This has been verified by sightings of orientation presentations where the frosh were in the audience and appearance of the crowd to be frequently split boy-girl-boy-girl-etc.
I know some CS females who feel they got favored for admission because they are female, but performance in classes regularly shows that the women aren't predominantly winding up on the bottom. I don't think there has been the proper analysis to compare the whole distributions, that will now be possible with the incoming large sample. It will be interesting to see if the bulk of the male curve winds up higher than the bulk of the female curve or any other such dead giveaway.
It will soon be impossible to really make as much money as has been made before off of recorded music sales. Look for recording artists to start selling music by the track for cents apiece as they only thing they can possibly sell is convenience.
Bands of the future will make money more through live performances and less through record sales. Money will still be made from recordings. A million people paying $.10 a track on a few tracks will still make you money, but only the U2's of the world could count that as primary income for very long.
Unfortunately the current credit system isn't prepared for the myriad of $.02 transactions that will come with such business models. Music serving companies will have to set up their own credit lines and customer accounts payable montly to accrue enough dollars to make it worth the transaction. There's a long way to go still for online money development. Read "Earth" by David Brin for an appealing model of usage (implementation is left as an exercise to the reader).
For those of us who are striking it rich hacking away all day at what the marketroids want often have beloved pet projects that we come home to every night and hack on some more, and when I come home to hack the night away I want to do it in all the style and comfort that my soul-stealing job can buy me.
Or I could be the Scrooge and put that $3500 into stocks so that I can have more $3500's later.
I wasn't able to find it on the site, does anyone know what the memory usage of the little kernel is? How much of that 8MB RAM/1-4MB flash do I get to play with?
Re: MMU/Supervisor mode in 68k- The 68030 introduced an MMU (as did a certian other chip with a conspicuosly distinguishing 3 in it) and before that some MacII's with 68020's had an external MMU. An MMU can prevent memory thrashing, but the Supervisor mode present in all 68k chips only controls who gets executed at interrupt time and system traps. Supervisor mode doesn't do much for safety aside from providing some organization to buttress the OS on.
And the current state of game AI is why they should all have an API that I can program my own AI to because I can do it better (or at least what I want).
If you've played Civ2 perhaps you can sympathise with the brain dead behaviour of "Automate settler" which irrigates evrything including the forests you were counting on for shield production.
AI API's would also make way for AI contests. I'd love to see a few more of those.
Gripe #2 on current state of game AI: cheating! Game programmers know that at some point the human learns and the computer has stayed the same so to up the difficulty the computer has to cheat. Civ again: building units becomes cheaper for the AI than the human, and the subjects ruled become more agreeable and easier to please requiring less expenditures freeing up effort for smashing the human.
As said in the counter-point, rule based AI is a dead end. With the exception of if the computer can generate new rules and learn. Learning must be the next step for game AI and I wish (as per gripe #1) that I could write some for some of the games I like.
Any analogy between this effort and the beneficial parts of parallel processing will most likely be wrong. The processing is uncoordinated so every compute node (person) must go through the same computations. Thousands of times more computing power with all parts geting about the same distance towrds the solution. The only benefit of this system is fault-tolerance, the stupid nodes get voted down.
The end result will be that the effective chess ability of the group will be the average (or slightly higher if they pay attention to the experts) of the group.
Personally, I believe that benchmark. Unfortunately eveyone knows you can get el-cheapo PC for $1000 to do almost as much. Heck, upgrade to the good el-cheapo for $2000 and yer pretty well loaded. But Apple won't stand for such comparisons, and will insist on being compared only to quality vendors such as Dell, Compaq, IBM and Sun. If it were my small business to run, I'd buy a quality machine over el-cheapo, and I'd buy the Mac because it comes headache-free.
It's a pity about that DEC, I saw it up close in person and there were some big ceramic PGA chips that would have made lovely jewelry if they hadn't been smashed like last year's coffe mug. (I gave my last girlfriend a beautiful 486dx33 cut open to expose the silicon)
Ah well, destruction was fitting for such an old computer. I know some people who spend much more time than reasonable trying to get those old junkers to keep working and be useful.
Slashdot Mirror
BEAM is useless.
I've talked to Mark Tilden, and seen his robots. They cannot learning anything more complex than resonance. That is definately a neat trick that many other systems have a hard time doing, but it is their only trick.
It can make some fairly simple yet effective walkers, that have no concept of what to do when they get there, or even have any concept of getting there. They just go, they don't do anything else. They are toys.
It runs vxWorks. (Hence the real-time processing, being an RTOS and all.)
CNN is having a poll, vote for favorite streaming media system ( MS, Real, QuickTime, Other ). When I looked QT and Real were in the lead. Speak your mind at http://cnn.com/2000/TECH/computing/01/28/stream.sh ootout.idg/index.html
What we need is a vote per candidate. Acceptable/Not Acceptable. Under this system, the best candidate (not just who the public happens to sway with for fear of 'throwing their vote away') gets elected in a more-than-two way race. And, it's also possible to throw out an election if people get up and go to the polls and every candidate gets more Not Acceptable than Acceptable votes. But, I don't think that should bar them from any future election. The system worked once, it can work again and continue to give them no-confidence results.
(ie in your bedroom.... hardware geeks with me?)
.5 mm apart or less, or somehow solder an entire array of contacts simultaneously while they sit between the board and the chip.
It's allready getting hard to do hardware hacking with anything much more than an 8 bit processor. I think you can still get a 68000 in a 64pin DIP, but everything else is heading to QFP, TSOP, and worst of all BGA. To work with these you either have to solder wires spaced
It's because of things like this that I have very few designs on building any sort of major computing device in the near future.
1. Megahertz is a dead end
Allready processors are too fast for the rest of the system. This has been alleviated for the last decade by an increasingly complicated system of caches and chipsets. At worst you'll go throgh 3 levels of processor cache, main memory, disk cache and finally disk, for a total of 6 levels of memory. This could go on indefinately but will have decreasing returns, unless the architecture of the computer can catch up to be generally faster. SGI/Cray has done this well.
2. Megahertz == Marketing
Ever since the P2, it's been terribly obvious that Intel just develops to satisfy what the majority clueless consumer wants- a higher megahertz number. The P2 made it blatant by being inferior to the older P's when run at equal megahertz. The only benefit was that it would run at higher megahertz.
Efficiency
No x86 has been really efficent- in many ways. More gates, more watts, more space, more heat. The unfortunate predominance of x86 is leading to space robots being designed with pentiums because Intel can push through to get the chips certified. When multiprocessing becomes a necessity as clock speeds dead end, who will be able to afford the power and large case for cooling that 8-64 P[3-5]'s will need? It's absurd.
If the CPU were the only thing to run (I know, this will exagerate the differences) a battery could last up to twice as long if either the G3 or G4 spent most of its time chilling in the low range of usage. User driven tasks like word processing allow such idling to take advantage of the PPC doze and nap power conservation modes, which is appearantly how Apple can claim 6 hour battery life on the iBooks. Other tasks would certainly run it down faster by preventing either the processor or the disk from idling.
I'm CMU CS of 2001. The number I heard for my year was 17 or 18 out of 138 a ratio of 7:1. Perhaps a few have left, I can offhand name about 10 of them.
The number I've heard for the incoming freshman class (2003) is 50 to 70 female/male. This has been verified by sightings of orientation presentations where the frosh were in the audience and appearance of the crowd to be frequently split boy-girl-boy-girl-etc.
I know some CS females who feel they got favored for admission because they are female, but performance in classes regularly shows that the women aren't predominantly winding up on the bottom. I don't think there has been the proper analysis to compare the whole distributions, that will now be possible with the incoming large sample. It will be interesting to see if the bulk of the male curve winds up higher than the bulk of the female curve or any other such dead giveaway.
It will soon be impossible to really make as much money as has been made before off of recorded music sales. Look for recording artists to start selling music by the track for cents apiece as they only thing they can possibly sell is convenience.
Bands of the future will make money more through live performances and less through record sales. Money will still be made from recordings. A million people paying $.10 a track on a few tracks will still make you money, but only the U2's of the world could count that as primary income for very long.
Unfortunately the current credit system isn't prepared for the myriad of $.02 transactions that will come with such business models. Music serving companies will have to set up their own credit lines and customer accounts payable montly to accrue enough dollars to make it worth the transaction. There's a long way to go still for online money development. Read "Earth" by David Brin for an appealing model of usage (implementation is left as an exercise to the reader).
For those of us who are striking it rich hacking away all day at what the marketroids want often have beloved pet projects that we come home to every night and hack on some more, and when I come home to hack the night away I want to do it in all the style and comfort that my soul-stealing job can buy me.
Or I could be the Scrooge and put that $3500 into stocks so that I can have more $3500's later.
Beware geeks with money.
I wasn't able to find it on the site, does anyone know what the memory usage of the little kernel is? How much of that 8MB RAM/1-4MB flash do I get to play with?
Re: MMU/Supervisor mode in 68k-
The 68030 introduced an MMU (as did a certian other chip with a conspicuosly distinguishing 3 in it) and before that some MacII's with 68020's had an external MMU. An MMU can prevent memory thrashing, but the Supervisor mode present in all 68k chips only controls who gets executed at interrupt time and system traps. Supervisor mode doesn't do much for safety aside from providing some organization to buttress the OS on.
And the current state of game AI is why they should all have an API that I can program my own AI to because I can do it better (or at least what I want).
If you've played Civ2 perhaps you can sympathise with the brain dead behaviour of "Automate settler" which irrigates evrything including the forests you were counting on for shield production.
AI API's would also make way for AI contests. I'd love to see a few more of those.
Gripe #2 on current state of game AI: cheating! Game programmers know that at some point the human learns and the computer has stayed the same so to up the difficulty the computer has to cheat. Civ again: building units becomes cheaper for the AI than the human, and the subjects ruled become more agreeable and easier to please requiring less expenditures freeing up effort for smashing the human.
As said in the counter-point, rule based AI is a dead end. With the exception of if the computer can generate new rules and learn. Learning must be the next step for game AI and I wish (as per gripe #1) that I could write some for some of the games I like.
Any analogy between this effort and the beneficial parts of parallel processing will most likely be wrong. The processing is uncoordinated so every compute node (person) must go through the same computations. Thousands of times more computing power with all parts geting about the same distance towrds the solution. The only benefit of this system is fault-tolerance, the stupid nodes get voted down.
The end result will be that the effective chess ability of the group will be the average (or slightly higher if they pay attention to the experts) of the group.
Maybe it only happens here but the Geek support structure is alive and well at this college.
I guess I had it easy, the geek population at my HS was probably well over 10%, a minority big enough not to be intimidated.
Apple says a G3 400 beats a Dell P2 450 measured by connections per second over 10bT.
See the official hype from Apple.
Personally, I believe that benchmark. Unfortunately eveyone knows you can get el-cheapo PC for $1000 to do almost as much. Heck, upgrade to the good el-cheapo for $2000 and yer pretty well loaded. But Apple won't stand for such comparisons, and will insist on being compared only to quality vendors such as Dell, Compaq, IBM and Sun. If it were my small business to run, I'd buy a quality machine over el-cheapo, and I'd buy the Mac because it comes headache-free.
It's a pity about that DEC, I saw it up close in person and there were some big ceramic PGA chips that would have made lovely jewelry if they hadn't been smashed like last year's coffe mug. (I gave my last girlfriend a beautiful 486dx33 cut open to expose the silicon)
Ah well, destruction was fitting for such an old computer. I know some people who spend much more time than reasonable trying to get those old junkers to keep working and be useful.