Indeed, surely the expectation of privacy in a public space approaches zero as technology increases? Why should it be any other way? The whole AT&T meets the NSA is just a consequence of a space that most people thought was private (relaying messages) turning out to be public. The rich & important in society have always treated messaging as a public space, otherwise we wouldn't have developed crypto systems.
But in this case the video being sent is from cameras mounted *in the street*. If I walk out my front door I can watch what you are doing there anyway, so why expect that it is private? Besides there could be other interesting applications for this that we don't find until we try it. One odd aspect is why transmit the video as a TV signal? 400 cameras, 400 URLs and a constant live stream. That would be interesting. Wondering what's going on in town - have a fly around and see. The hack that ties it into the OS polygon data for UK cities and Google Maps would be pretty awesome.
If it's derived from any of the Abrahamic religions then it probably says that moral (ie religious) law is above the law of man. And unless it's a *really* new offshoot I doubt they have a specific policy on computer crime....
I'm not so sure that hydro is tapped out. They've started to talk about a tidal barge across the Severn Estuary again. If this got past the conservationists it would supply 25% of the current UK power demand. Not a single city, but the whole country. Of course it will never get built until there is £50B lying around that the government can claim, and all of the local residents / environmentalists have died. The irony is that not building it is contributing to the destruction of those habitats anyway.
So all of the Americans are bitchin about how expensive the console is.... umm, weak dollar. According to Google US$ 599 = £321.85, which is almost exactly what I payed for my PS2 when it launched. Sounds ok to me. Even if Sony try and pull the stupid $1=£1 conversion trick when they launch in the UK I'll just buy a US unit and ship it.
You've hit the nail on the head - if it's too big to go in a pocket then it may as well be big enough to be useful. I don't think that problem is going to change until they actually get folding screens to market. Then something can be big enough to be useful, but fold into a portable shape.
ScuttleMonkey is running an article that describes how hip2b2 is automatically generating articles to get better slashdot coverage. The technology is based on bayesian generation and requires that articles are not read in depth, but merely scanned. Today, articles are not read. Slashdot readers only skim them. This makes random generation suitable for getting articles posted. Now I will mention diamonds.
Congratulations, you have very nearly passed the Turing Test.
However, the authors have so far only done the maths to verify that the concept could work. Building such a device would undoubtedly pose a significant challenge.
Come on, just admit that you're talking shit.
Re:I looked for the Churchill speech
on
On The BBC 2.0
·
· Score: 1
Good tip, that's a much better result. I guess google has made me lazy...
Re:I looked for the Churchill speech
on
On The BBC 2.0
·
· Score: 1
But a search for doctor who only found *nine* items. Which is a bit shit really.
It seems unlikely. The synthetic benchmark that they describe is four particular FP instructions, repeated several million times. Firstly no compiler would unroll a loop that far. Secondly, the four instructions would never occur in practice because you would need some memory accesses at some point to load/store data. Between the loop control, and the memory accesses, the FP instructions are broken up enough that they don't trigger the failure.
You're mixing up a few pieces of technology here. Processors with their own dedicated memory has been invented many times by different people. Modern loosely coupled clusters fit this bill, but further back there was the transputer systems in which each processor had memory on board. Systems like this are more difficult to program than single image systems (even with a CSP derivative as the language) but they produce higher performance.
The other thing that you are describing is multiway branch prediction. A processor like the Pentium guesses which way a branch goes and despatching instructions down that path to the pipeline. When it is wrong there is a hit as the pipeline stalls and all of those cycles are lost. In multiway branching both outcomes of the branch are despatched to the pipeline. The cost is that half the instructions being executed will be thrown away. If you go 2 branches deep then it is 75%. The advantage is the latency is minimised as the pipeline is always full.
The last thing is processor-in-RAM, or smart memory. In this system a miniture processor is embedded on the DRAM die. The small processor is capable of computing striding patterns in arrays. As the program executes on the main processor the smaller processor predicts which memory locations are going to be accessing and presending the data to the host processor, reducing latency.
Good luck on your class. Architecture is one of the more interesting courses in a CS degree.
It sounds like an education is expensive in the states.
Over here in the UK there are two types of textbooks; those that are specific to some course, becoming useless after you've finished it, and those that are more general and retain use as a reference after you graduate. The latter kind are quite expensive (£60-£100), but the money is generally worth it for the good ones. For the former I think there is only one set that does the rounds. Each year the outgoing year sells them to the incoming year on bulletin boards, either on the department website or on any of the paper based ones around the building.
The combination of both means that books aren't too expensive over here.
- An example of an ad for gambling. Instructional material, and play money sites don't count.
Just search for poker. I get three ads across the top for real-money sites, and a chunk down the side.
But I don't understand why this is a big deal. Google don't have a policy on banning adverts based on what they sell. If you look at the Screened Products section in the adwords faq you'll find the phrase 'Specifically, we currently monitor the following in ad text and keywords:', not 'We do not allow...' In order words, they reserve the right to reject advertising of certain products, but they don't say what they are. They give a list of categories (including gambling and porn) that they *may* reject if they feel like it.
Going off topic a bit, why should Google have a problem with gambling? It's perfectly legally in most of the countries that they serve with search, the WTO has decided that it should come under free trade, and other than some dead words from an obsolete religion, there is no reason that it should be restricted.
The GP is completely wrong. The advantage of a console for a games developer is one standard configuration. They rely on this to squeeze as much out of the platform as necessary, which is pretty much your second point. And as others have pointed out they may use less energy but that is just a side effect.
Decreasing the feature size will increase the yield, and so the cost per chip will drop. This is the primary reason for the reduction. Your first point isn't technically wrong, just very unlikely. The behavioural specification for the chip will be the same as the unreduced one - so there will be no new bugs unless there is a side-effect from the change in size. Chip companies tend to hold off on reducing feature sizes until they've worked the kinks (ie side-effects) out of the fabrication process.
This reduction is not like going from a PIII -> PIV, the design of the chip is the same. It's just the unerlying fabrication technology that differs.
Dude, you are a god amongst men. Your sig is probably the best thing that I have read in years. The number of miles that you will save my left hand in an average year is just flabbergasting.
That would make sense, although from what I know of hyperthreading the virtual core doesn't have as high a priority as the real core. So the despatcher issues as many instructions from the main thread as possible, but the second thread is more of a guest with instructions being issued when the real thread isn't using all the resources. I'm not 100% sure about that though.
The difficulty in keeping the cores full is because most program don't expose that much parallelism. Adding cores doesn't magically fix that - but some programs do. Chosing the chip design is just an engineering tradeoff - which is the most common case, and then optimise for that. So Intel/AMD went down the superscalar route for as far as they could, but they got diminishing returns after a while.
Multicore designs are optimising for a different kind of code - but they suck at running the programs that do expose a lot of implicit parallelism. Think game inner loops that have been partially unrolled by a compiler and lots of registers have been used so that the processor can see that there are no data dependencies. A multicore chip can't take advantage of this kind of code - the parallelism is too fine grained. It is pretty good at code with lots of coarse grain parallalism that has been separated into lots of threads. But when it is executing inner loops half of the EU's are on another core, unused.
AMDs idea (I guess) is that you can switch which way the chip operates, so in 'single-thread' mode the EUs on both cores can be fed by a common despatcher which means that implicitly parallel code can be accelerated. In the other mode two threads (without much implicit parallelism) can be executed so you still get an increase in throughput.
So it isn't easier for the despatcher to keep the EUs fed on both cores, in fact as you say, it is much harder. I don't know how they're going to get around the complexity issue (despatch difficulty isn't linear in the number of cores), or the communication costs between the cores. The gain is that you can still run multi-threaded code fast by switching modes, whereas in a pure superscalar design you are chosing one sweetspot in the design over another.
The effect is the same but you gain more flexibility. If you only add more EUs to a chip then they can get starved because it's hard to to the despatch to keep them full. If you have multicore, but only one main thread then at least half of your EUs are getting unused. The cute answer, from an engineering point of view is to allow both, and then switch between them. Then if you have long single-threaded sections of code with lots of implicit parallelism (ie games) you can load up the EUs, or if you have lots of threads without lots of implicit parallelism then you can still load them up - but segrated into cores.
The article had no details at all, one of the hard questions is who decides what mode the chip is in? Does it dynamically switch modes depending on the current load (very tricky, and will require a lot of logic between the cores)? Or is it a software enabled mode switch - tough on the thread scheduler in the kernel.
VR sickness is induced differently. That occurs because there is real motion (of your head), but there is a timelag between this motion and the change in picture. Your brain can't handle the lag as your ears and eyes don't match up, and so you feel sick.
The alternative that they are referring to in the article is a motion pod. In a pod you get thrown around alot, which will make you feel sick anyway, but you probably also have a lag between the feeling and motion. Every year we get a bunch of students that write a game on the motion pod as their project. Most are pretty good, but you should see the colour the guys turn when they are debugging them.
The effect that the article is talking about is different. The analogy that the speaker at APGV used last year is when you are sitting on a train. If the train is stationary but another train moves in the opposite direction it feels (physically) as if you are moving. This is a perceptual illusion, and the project uses these types of illusion to make you feel as if you are moving, when in fact you are not.
So what you are calling VR sickness (no motion) isn't actually true of VR, but may be true in this case (if it is installed in a cabinet rather than a headset).
I think the point that you are missing is that the dissemenation of knowledge is not to other inventors. The dissemination occurs because company A makes some product, company B figures they can make it cheaper, and the patent serves as dissemenation of the knowledge from A to B. The landgrab that you refer to, is making sure that company A gets paid for their original idea. Inventors don't generally read patents, they study problems and play with ideas.
I think that you're using the term post-doc in a misleading way. Are you talking about people working as a *post-doc* i.e. a first research position within academia, or do you merely mean someone who has completed a PhD but is working in industry?
Slashdot Mirror
Indeed, surely the expectation of privacy in a public space approaches zero as technology increases? Why should it be any other way? The whole AT&T meets the NSA is just a consequence of a space that most people thought was private (relaying messages) turning out to be public. The rich & important in society have always treated messaging as a public space, otherwise we wouldn't have developed crypto systems.
But in this case the video being sent is from cameras mounted *in the street*. If I walk out my front door I can watch what you are doing there anyway, so why expect that it is private? Besides there could be other interesting applications for this that we don't find until we try it. One odd aspect is why transmit the video as a TV signal? 400 cameras, 400 URLs and a constant live stream. That would be interesting. Wondering what's going on in town - have a fly around and see. The hack that ties it into the OS polygon data for UK cities and Google Maps would be pretty awesome.
If it's derived from any of the Abrahamic religions then it probably says that moral (ie religious) law is above the law of man. And unless it's a *really* new offshoot I doubt they have a specific policy on computer crime....
I'm not so sure that hydro is tapped out. They've started to talk about a tidal barge across the Severn Estuary again. If this got past the conservationists it would supply 25% of the current UK power demand. Not a single city, but the whole country. Of course it will never get built until there is £50B lying around that the government can claim, and all of the local residents / environmentalists have died. The irony is that not building it is contributing to the destruction of those habitats anyway.
So all of the Americans are bitchin about how expensive the console is .... umm, weak dollar. According to Google US$ 599 = £321.85, which is almost exactly what I payed for my PS2 when it launched. Sounds ok to me. Even if Sony try and pull the stupid $1=£1 conversion trick when they launch in the UK I'll just buy a US unit and ship it.
You've hit the nail on the head - if it's too big to go in a pocket then it may as well be big enough to be useful. I don't think that problem is going to change until they actually get folding screens to market. Then something can be big enough to be useful, but fold into a portable shape.
Good tip, that's a much better result. I guess google has made me lazy...
But a search for doctor who only found *nine* items. Which is a bit shit really.
It seems unlikely. The synthetic benchmark that they describe is four particular FP instructions, repeated several million times. Firstly no compiler would unroll a loop that far. Secondly, the four instructions would never occur in practice because you would need some memory accesses at some point to load/store data. Between the loop control, and the memory accesses, the FP instructions are broken up enough that they don't trigger the failure.
You're mixing up a few pieces of technology here. Processors with their own dedicated memory has been invented many times by different people. Modern loosely coupled clusters fit this bill, but further back there was the transputer systems in which each processor had memory on board. Systems like this are more difficult to program than single image systems (even with a CSP derivative as the language) but they produce higher performance.
The other thing that you are describing is multiway branch prediction. A processor like the Pentium guesses which way a branch goes and despatching instructions down that path to the pipeline. When it is wrong there is a hit as the pipeline stalls and all of those cycles are lost. In multiway branching both outcomes of the branch are despatched to the pipeline. The cost is that half the instructions being executed will be thrown away. If you go 2 branches deep then it is 75%. The advantage is the latency is minimised as the pipeline is always full.
The last thing is processor-in-RAM, or smart memory. In this system a miniture processor is embedded on the DRAM die. The small processor is capable of computing striding patterns in arrays. As the program executes on the main processor the smaller processor predicts which memory locations are going to be accessing and presending the data to the host processor, reducing latency.
Good luck on your class. Architecture is one of the more interesting courses in a CS degree.
It sounds like an education is expensive in the states.
Over here in the UK there are two types of textbooks; those that are specific to some course, becoming useless after you've finished it, and those that are more general and retain use as a reference after you graduate. The latter kind are quite expensive (£60-£100), but the money is generally worth it for the good ones. For the former I think there is only one set that does the rounds. Each year the outgoing year sells them to the incoming year on bulletin boards, either on the department website or on any of the paper based ones around the building.
The combination of both means that books aren't too expensive over here.
That's quite a nice link. I think what the OP is looking for is this section covering scripting.
I really don't understand your last sentence in the slightest.
It's called humour. You might want to remove the stick from your arse and reread his entire post. Trust me, it'll make more sense.
One of the most useful sites for finding 3d engines has always been here. There are hundreds, and they can be rearranged into chosen categories.
Your comment is troll / insightful / flamebait. Pick any two...
- An example of an ad for gambling. Instructional material, and play money sites don't count.
Just search for poker. I get three ads across the top for real-money sites, and a chunk down the side.
But I don't understand why this is a big deal. Google don't have a policy on banning adverts based on what they sell. If you look at the Screened Products section in the adwords faq you'll find the phrase 'Specifically, we currently monitor the following in ad text and keywords:', not 'We do not allow...' In order words, they reserve the right to reject advertising of certain products, but they don't say what they are. They give a list of categories (including gambling and porn) that they *may* reject if they feel like it.
Going off topic a bit, why should Google have a problem with gambling? It's perfectly legally in most of the countries that they serve with search, the WTO has decided that it should come under free trade, and other than some dead words from an obsolete religion, there is no reason that it should be restricted.
The GP is completely wrong. The advantage of a console for a games developer is one standard configuration. They rely on this to squeeze as much out of the platform as necessary, which is pretty much your second point. And as others have pointed out they may use less energy but that is just a side effect.
Decreasing the feature size will increase the yield, and so the cost per chip will drop. This is the primary reason for the reduction. Your first point isn't technically wrong, just very unlikely. The behavioural specification for the chip will be the same as the unreduced one - so there will be no new bugs unless there is a side-effect from the change in size. Chip companies tend to hold off on reducing feature sizes until they've worked the kinks (ie side-effects) out of the fabrication process.
This reduction is not like going from a PIII -> PIV, the design of the chip is the same. It's just the unerlying fabrication technology that differs.
Dude, you are a god amongst men. Your sig is probably the best thing that I have read in years. The number of miles that you will save my left hand in an average year is just flabbergasting.
That would make sense, although from what I know of hyperthreading the virtual core doesn't have as high a priority as the real core. So the despatcher issues as many instructions from the main thread as possible, but the second thread is more of a guest with instructions being issued when the real thread isn't using all the resources. I'm not 100% sure about that though.
The difficulty in keeping the cores full is because most program don't expose that much parallelism. Adding cores doesn't magically fix that - but some programs do. Chosing the chip design is just an engineering tradeoff - which is the most common case, and then optimise for that. So Intel/AMD went down the superscalar route for as far as they could, but they got diminishing returns after a while.
Multicore designs are optimising for a different kind of code - but they suck at running the programs that do expose a lot of implicit parallelism. Think game inner loops that have been partially unrolled by a compiler and lots of registers have been used so that the processor can see that there are no data dependencies. A multicore chip can't take advantage of this kind of code - the parallelism is too fine grained. It is pretty good at code with lots of coarse grain parallalism that has been separated into lots of threads. But when it is executing inner loops half of the EU's are on another core, unused.
AMDs idea (I guess) is that you can switch which way the chip operates, so in 'single-thread' mode the EUs on both cores can be fed by a common despatcher which means that implicitly parallel code can be accelerated. In the other mode two threads (without much implicit parallelism) can be executed so you still get an increase in throughput.
So it isn't easier for the despatcher to keep the EUs fed on both cores, in fact as you say, it is much harder. I don't know how they're going to get around the complexity issue (despatch difficulty isn't linear in the number of cores), or the communication costs between the cores. The gain is that you can still run multi-threaded code fast by switching modes, whereas in a pure superscalar design you are chosing one sweetspot in the design over another.
The effect is the same but you gain more flexibility. If you only add more EUs to a chip then they can get starved because it's hard to to the despatch to keep them full. If you have multicore, but only one main thread then at least half of your EUs are getting unused. The cute answer, from an engineering point of view is to allow both, and then switch between them. Then if you have long single-threaded sections of code with lots of implicit parallelism (ie games) you can load up the EUs, or if you have lots of threads without lots of implicit parallelism then you can still load them up - but segrated into cores.
The article had no details at all, one of the hard questions is who decides what mode the chip is in? Does it dynamically switch modes depending on the current load (very tricky, and will require a lot of logic between the cores)? Or is it a software enabled mode switch - tough on the thread scheduler in the kernel.
VR sickness is induced differently. That occurs because there is real motion (of your head), but there is a timelag between this motion and the change in picture. Your brain can't handle the lag as your ears and eyes don't match up, and so you feel sick.
The alternative that they are referring to in the article is a motion pod. In a pod you get thrown around alot, which will make you feel sick anyway, but you probably also have a lag between the feeling and motion. Every year we get a bunch of students that write a game on the motion pod as their project. Most are pretty good, but you should see the colour the guys turn when they are debugging them.
The effect that the article is talking about is different. The analogy that the speaker at APGV used last year is when you are sitting on a train. If the train is stationary but another train moves in the opposite direction it feels (physically) as if you are moving. This is a perceptual illusion, and the project uses these types of illusion to make you feel as if you are moving, when in fact you are not.
So what you are calling VR sickness (no motion) isn't actually true of VR, but may be true in this case (if it is installed in a cabinet rather than a headset).
I think the point that you are missing is that the dissemenation of knowledge is not to other inventors. The dissemination occurs because company A makes some product, company B figures they can make it cheaper, and the patent serves as dissemenation of the knowledge from A to B. The landgrab that you refer to, is making sure that company A gets paid for their original idea. Inventors don't generally read patents, they study problems and play with ideas.
I think that you're using the term post-doc in a misleading way. Are you talking about people working as a *post-doc* i.e. a first research position within academia, or do you merely mean someone who has completed a PhD but is working in industry?