I asked this question of a friend once who works in the transmission buisness. The simple answer is 2 fold: 1) because you don't have lots of space to insulate with free air, you either need fantastic insulators (gassious sodium being popular), or to run at a lower voltage(and simply use good insulators). 2) It's hard to quickly evolve the power system. By quicky you're talking about changing more than about once every 4 or 5 decades.
When they do bury power underground, (the example he gave was the main supply into Northampton that he had recently been working on the project to upgrade their underground power distrubution) you end up with cables that are less like fex, and more like foot wide girders of steel you bury underground - it'd take some amasing conduit to service that(as you may imagine you'd service cat 5 cable/fibre in conduit)!
Unless you need* a Red Hat like install. Than Fedora is the most legal way I know of getting an up to date one.
* e.g. my new raid controller only has official support under red hat enterprise - so far running with fedora everything has been ok. As a rule I find that Fedora has the best driver support because it is the testing platform for redhat...
You know there's a word for someone who looks at a new technology and sees that it doesn't have direct application to his life and therefore talks it down.
Seriously this is research, they are pushing the limits as far in one direction as they possibly can with the assumption that if you research at the extreme then you'll learn things that can be applied to more mundane situations. What next? IBM issue a press release about new transistors based on nanotubes that go 1000X faster and you complain that because there won't be a processor available based on them available any time soon that they are wasting their time?
Profit = benefit for (at least one) human. Please explain to me how this is a bad thing. Explain how (military use aside) any expoitation of space would be a bad thing. If we polute the moon by dumping tons of co2 or mining waste into it, how is it going to be any less habitable? How do we destroy the environment of the moon to make it any less habitable for mankind?
Now if we are talking about venus - which is a lush paradise filled with dense forest and buxom warrior women - then I'd be all against mining that and instead booking the first out of control spacecraft there! Oh wait a second...
Or should we just look at the moon and all objects in the sky and never use them?
Re: Ferengi... Ok... Glad to see people know where the library is these days...
Seems like we're hitting some problems on assumptions here. You seem to be assuming I'm talking about a tradition von Neumon like CPU architecture. I'm not. I'm assuming that context switching and communication can be free. Which they can, depending on your solution architecture. Modern OSes though don't have free context switiching and communication though, so you're right, choosing the correct degree of serialisation vs parallel is a difficult decision.
I'm saying that conceptually the problem could be as parallel as you want it to be, right down to one thread per sfx. one per client etc. With _current_ _common_ cpu architectures i agree this would be insane. If however we are talking about possible future architectures, or past architectres such as the inmos transputer, (occam etc) then I believe a finer grained parallism such as i propose is much more appropriate. I propose this partly because in my last job I was designing sound processing hardware that did process each stream in parrallel because that was the simplest thing to do in hardware. So it is possible, and it is sometimes desirable if it is appropriate to your solution architecture.
i.e. I;m trying to say that the problem is inherently parrallel if you want it to be. If your architecture forces you to serialise stuff, then that's fine too, but just because on current architectures you think about a problem as being particularly serial, doesn't mean that it always and forever has to be. for example, if you knew that context switches and communication channels did not have a performance penalty (no mutexes) then would your design decisions be the same?
Did you read mine? Scene management is one thing that is hard to parralise, fine. but i quote "Well, then show me how to insert a parallel architecture in your typical 3D game" Which I did. each of the items you listed as threads i suggested how thay could be further parallelised - that was why I mentioned them. So I assert that your statement of "Multithreading can work only in very high levels" is false as I have suggested how it could be further parrallelised.
I agree there are some segments that cannot be split up in themselves, but there is no reason why you cannot be having the serial scene management thread running while all the other parallel things are running. As my understanding of modern 3d games goes, scene management is a relatively small part of the CPU time. Even if there are 100 things that cannot themselves be parralised within their own algorithm, they can often be run in parrallel with the correct architecture.
Well I was taking a bit of liberty there... I've lived in the south east for a while myself, and in 2 of those places the locals knew the local accents (Harlow & Kings Langley). Currently in Cambridge where it's a little more of a mix, but there is still the difference in the accents between say Ely & Kings Hedges.
"But you cannot insert multithreading into the graphics subsystem" Are you serious? Are we even talking about the same thing? You know the whole point of a GPU is to parallise everything right? You know each pixel can be rendered in parallel? You know each sprite can usually be drawn to polygons in parallel? You know most of the physical simulation can be done in parallel? You know that with the correct architecture all of the above can happen in parralel to each other?
Even the sound thread can be parralised. You've an input queue of sound events that need playing, each sound effect can be a separate thread within that. Networking thread? Well I need to communicate many things, let me have one thread talking to each other player perhaps, and then within that a thread to do each of the things such as communicating my character's current position/action, communicating other character's position, triggering game events, handling player-play communication etc.
Now I agree this requires communication between threads which current architectues/languages don't do well. But that doesn't mean that the problem isn;t parallel, merely our expression and solution of the problem that lacks parallelism.
I'd also add the argument, that in a higher level language, the code to mundane functionc is likley to be quite optimal. e.g. I'd bet that the standard C compiler makes a more optimal for loop than I would unless I really thought about it. i.e. in a high level language, the compiler is probably using optimising tricks at every single iunstruction, in a low level language I'm only hand optimising the bits that I believe will matter.
i.e. 10% of my code in a low level language may be at the absolute optimum, but the other 90% of the code i didn't think mattered is vastly sub-optimal.
With this in mind it bothers me that cross language support and profiling tools aren't better.
i.e. why isn't there a decent high level language* like python that lets me embed C, when the profiling tells me i need it, or assembler** when I really need it.
*C++ or system C don't count!:-p **Yes know i can embed assembler in C, but not C (or similar low level language) in most other languages.
Regardless of if you are a performance critical app or not, the philosophy of code in as high a level language as possible and then profile your code/redesign from scratch many time learning what you did wrong is still valid. Anything that makes code easier to maintain is _always_ a good thing. And in most embedded environments, the code simply has to go fast enough - if the performance metric is reached, you can stop worrying! i.e. if your java prototype that made full use of objects, structures and other gubbins of a high level language , and it happened to run fast enough, who cares that it is sub-optimal?
Now I'd certainly agree that in most embedded environemtns any garbage collection is unviable because they are not suited to a real time system. I'd also conceed that if you know in advance 90% of your code is going to be performace critical then it might be better to just keep everything in a low level language like C. I'd similarly conceed that in some embedded environments the processor is the main cost (in $ or in power) so minimum cpu cycles is worth the development time.
As long as you know you're making that tradeoff...
While on holiday in spain a decade back we once met a man (from Dover Strangely enough) who could tell where we came from down to about a 2 mile radius thanks to our accents. The strange thing is this didn't really surprise us because the accent changes enough across our town that we can do it, so the fact that someone else can is not a surprise.
This is also not uncommon in the uk for accents to be this local. I have never lived in a town where the locals could not do this trick (localise a local to within a mile or so based upon his accent).
So when it seems that all american's think the english have either a cockney accent or an RPS (or whatever it's called) accent, this is somewhat annoying.
I sense sarcasm! I'll have you know there are lots of well adjusted, sexy men with stable relationships who hang out on Livejournal! Just like Slashdot...
The BPI does not set the law though. This holds as much weight as Bill Gates saying Linux encourages Piracy.
By that I mean, he may be right, he may be wrong, an opinion stated loud enough frequently enough by someone sounding official will start to carry weight.
a tube 11km kept as a vaccum? Sure you do this in your big particle accelerators, but there are precious few maglev trains 11km long, never mind running them vertically inside a particle accelerator style environment!
Maybe this is a case of scaling up current technology, and maybe apollo was just a case of scaling up the v2 rocket, I'm still somewhat sceptical though since we've never built a building 1km tall, never mind 11!
Remeber once a law is passed a government can choose how closely it enforces it. You cannot.
You have to follow the law, it doesn't have to uphold it (if it should or not is another issue) A law that can be applied at will can be used to cause trouble for the people you don't like.
"Free speech has this proviso, but this is a world away from what we're talking about here."
Right, so you agree with the point I'm trying to make then that speech shouldn't be free if it causes others harm. The problem is where to draw the line, should I not be allowed to talk openly about how to build a home-made bomb because it could help any terrorists? Should i not be able to criticise the government because a revolution will cause harm/death to a lot of people? It's all about where you draw the line and that is a matter of opinion, not facts or rights. How much more important is one man's right to speach over anothers life (or a probablility of anothers death).
Re: Antagonising, I'm trying to stick to the generic real world case, as opposed to the actual principles. Along the same lines that in principle I don't want to have a military...
In principle you're totally right that I should be able to antagonise anyone on a verbal level without a physical consequence. But this is the real world, and there does come a time when a physical response to a verbal event is arguably justified. FWIW I'm with the kid on this issue, the school is way out of line, I'm just trying to make the point that speech and what you say has a responsibility to the rest of society. What I was trying to say was an action that merely antagonises is probably best worth avoiding. Taking the MLK case, "a black woman drinking out of a white's fountain" may have just antagonising at the time, it may have been many things, but it started events that changed society for the better. I don't think I ever said all antagonising was bad (see the last line in my post) just that it should be done with care.
You say that,but A friend who is an instrutor had problems when he was mugged recently. His first instinct was to fight back using his techniques. He then remembered the advice to not do so and let himself get kicked in.
Good job, because even though the police caught the mugger, he got let off because of the injuries he sustained from my friend initially defending himself counted as provocation. Heaven knows what would have happened if he'd have properly disabled the mugger!
Then there was the other friend (BIG guy, into swords as a hobby, claims he has an 6' broadsword for a toy - never seen it except in photos though). Who got attacked by 3 guys and he ended up getting stabbed during trying to hold them off. At that point he went nuts and let loose his self defense/fighting training and ended up hospitalising two of them. The only reason he got off without criminal charges was because there was cctv footage of him being stabbed before he properly attacked them.
Slashdot Mirror
Sounds a lot like the ATM concpets of Virtual Paths...
Old jokes are always the oldest!
t e=990417
http://www.dragon-tails.com/comics/archive.php?da
I asked this question of a friend once who works in the transmission buisness.
The simple answer is 2 fold:
1) because you don't have lots of space to insulate with free air, you either need fantastic insulators (gassious sodium being popular), or to run at a lower voltage(and simply use good insulators).
2) It's hard to quickly evolve the power system. By quicky you're talking about changing more than about once every 4 or 5 decades.
When they do bury power underground, (the example he gave was the main supply into Northampton that he had recently been working on the project to upgrade their underground power distrubution) you end up with cables that are less like fex, and more like foot wide girders of steel you bury underground - it'd take some amasing conduit to service that(as you may imagine you'd service cat 5 cable/fibre in conduit)!
Unless you need* a Red Hat like install. Than Fedora is the most legal way I know of getting an up to date one.
* e.g. my new raid controller only has official support under red hat enterprise - so far running with fedora everything has been ok. As a rule I find that Fedora has the best driver support because it is the testing platform for redhat...
You know there's a word for someone who looks at a new technology and sees that it doesn't have direct application to his life and therefore talks it down.
....
Seriously this is research, they are pushing the limits as far in one direction as they possibly can with the assumption that if you research at the extreme then you'll learn things that can be applied to more mundane situations.
What next? IBM issue a press release about new transistors based on nanotubes that go 1000X faster and you complain that because there won't be a processor available based on them available any time soon that they are wasting their time?
Watching Karma burn in 5, 4, 3, 2
Profit = benefit for (at least one) human. Please explain to me how this is a bad thing. Explain how (military use aside) any expoitation of space would be a bad thing.
If we polute the moon by dumping tons of co2 or mining waste into it, how is it going to be any less habitable? How do we destroy the environment of the moon to make it any less habitable for mankind?
Now if we are talking about venus - which is a lush paradise filled with dense forest and buxom warrior women - then I'd be all against mining that and instead booking the first out of control spacecraft there! Oh wait a second...
Or should we just look at the moon and all objects in the sky and never use them?
Re: Ferengi... Ok... Glad to see people know where the library is these days...
There's the cold war joke about the 2 soviet tank commanders sat in paris, drinking champagne asking who won the air war.
The war is won by men on the ground - not that other assets don't help them of course...
Not quite RPS, but I was close:o n
http://en.wikipedia.org/wiki/Received_Pronunciati
I'm sure not all American's think that, just merely from watching american TV it seems like that at times.
But if I did that I'd assume 90 % of the americans are white but with every social group having no less and no more than one black man.
Seems like we're hitting some problems on assumptions here. You seem to be assuming I'm talking about a tradition von Neumon like CPU architecture. I'm not. I'm assuming that context switching and communication can be free. Which they can, depending on your solution architecture. Modern OSes though don't have free context switiching and communication though, so you're right, choosing the correct degree of serialisation vs parallel is a difficult decision.
I'm saying that conceptually the problem could be as parallel as you want it to be, right down to one thread per sfx. one per client etc.
With _current_ _common_ cpu architectures i agree this would be insane.
If however we are talking about possible future architectures, or past architectres such as the inmos transputer, (occam etc) then I believe a finer grained parallism such as i propose is much more appropriate. I propose this partly because in my last job I was designing sound processing hardware that did process each stream in parrallel because that was the simplest thing to do in hardware. So it is possible, and it is sometimes desirable if it is appropriate to your solution architecture.
i.e. I;m trying to say that the problem is inherently parrallel if you want it to be. If your architecture forces you to serialise stuff, then that's fine too, but just because on current architectures you think about a problem as being particularly serial, doesn't mean that it always and forever has to be.
for example, if you knew that context switches and communication channels did not have a performance penalty (no mutexes) then would your design decisions be the same?
Did you read mine?
Scene management is one thing that is hard to parralise, fine. but i quote "Well, then show me how to insert a parallel architecture in your typical 3D game"
Which I did. each of the items you listed as threads i suggested how thay could be further parallelised - that was why I mentioned them. So I assert that your statement of "Multithreading can work only in very high levels" is false as I have suggested how it could be further parrallelised.
I agree there are some segments that cannot be split up in themselves, but there is no reason why you cannot be having the serial scene management thread running while all the other parallel things are running. As my understanding of modern 3d games goes, scene management is a relatively small part of the CPU time.
Even if there are 100 things that cannot themselves be parralised within their own algorithm, they can often be run in parrallel with the correct architecture.
Well I was taking a bit of liberty there...
I've lived in the south east for a while myself, and in 2 of those places the locals knew the local accents (Harlow & Kings Langley).
Currently in Cambridge where it's a little more of a mix, but there is still the difference in the accents between say Ely & Kings Hedges.
You mean like the chef programming language?g uage
http://en.wikipedia.org/wiki/Chef_programming_lan
"But you cannot insert multithreading into the graphics subsystem"
Are you serious? Are we even talking about the same thing?
You know the whole point of a GPU is to parallise everything right?
You know each pixel can be rendered in parallel?
You know each sprite can usually be drawn to polygons in parallel?
You know most of the physical simulation can be done in parallel?
You know that with the correct architecture all of the above can happen in parralel to each other?
Even the sound thread can be parralised. You've an input queue of sound events that need playing, each sound effect can be a separate thread within that.
Networking thread? Well I need to communicate many things, let me have one thread talking to each other player perhaps, and then within that a thread to do each of the things such as communicating my character's current position/action, communicating other character's position, triggering game events, handling player-play communication etc.
Now I agree this requires communication between threads which current architectues/languages don't do well. But that doesn't mean that the problem isn;t parallel, merely our expression and solution of the problem that lacks parallelism.
I'd also add the argument, that in a higher level language, the code to mundane functionc is likley to be quite optimal.
:-p
e.g. I'd bet that the standard C compiler makes a more optimal for loop than I would unless I really thought about it.
i.e. in a high level language, the compiler is probably using optimising tricks at every single iunstruction, in a low level language I'm only hand optimising the bits that I believe will matter.
i.e. 10% of my code in a low level language may be at the absolute optimum, but the other 90% of the code i didn't think mattered is vastly sub-optimal.
With this in mind it bothers me that cross language support and profiling tools aren't better.
i.e. why isn't there a decent high level language* like python that lets me embed C, when the profiling tells me i need it, or assembler** when I really need it.
*C++ or system C don't count!
**Yes know i can embed assembler in C, but not C (or similar low level language) in most other languages.
I disagree (but then this is a philosphy thing).
Regardless of if you are a performance critical app or not, the philosophy of code in as high a level language as possible and then profile your code/redesign from scratch many time learning what you did wrong is still valid. Anything that makes code easier to maintain is _always_ a good thing. And in most embedded environments, the code simply has to go fast enough - if the performance metric is reached, you can stop worrying! i.e. if your java prototype that made full use of objects, structures and other gubbins of a high level language , and it happened to run fast enough, who cares that it is sub-optimal?
Now I'd certainly agree that in most embedded environemtns any garbage collection is unviable because they are not suited to a real time system. I'd also conceed that if you know in advance 90% of your code is going to be performace critical then it might be better to just keep everything in a low level language like C.
I'd similarly conceed that in some embedded environments the processor is the main cost (in $ or in power) so minimum cpu cycles is worth the development time.
As long as you know you're making that tradeoff...
While on holiday in spain a decade back we once met a man (from Dover Strangely enough) who could tell where we came from down to about a 2 mile radius thanks to our accents.
The strange thing is this didn't really surprise us because the accent changes enough across our town that we can do it, so the fact that someone else can is not a surprise.
This is also not uncommon in the uk for accents to be this local. I have never lived in a town where the locals could not do this trick (localise a local to within a mile or so based upon his accent).
So when it seems that all american's think the english have either a cockney accent or an RPS (or whatever it's called) accent, this is somewhat annoying.
I know what you mean, these days for technical news I prefer:
:-)
http://technocrat.net/
But not enough commentry yet - hence the reason I stay around slashdot.
off the main topic I know, but what the hey! it's ontopic for a reply
I sense sarcasm!
I'll have you know there are lots of well adjusted, sexy men with stable relationships who hang out on Livejournal!
Just like Slashdot...
The BPI does not set the law though.
This holds as much weight as Bill Gates saying Linux encourages Piracy.
By that I mean, he may be right, he may be wrong, an opinion stated loud enough frequently enough by someone sounding official will start to carry weight.
Attempt to use BFG to get into orbit - mostly successful, damn politics though.
http://www.astronautix.com/articles/abroject.htm
a tube 11km kept as a vaccum?
Sure you do this in your big particle accelerators, but there are precious few maglev trains 11km long, never mind running them vertically inside a particle accelerator style environment!
Maybe this is a case of scaling up current technology, and maybe apollo was just a case of scaling up the v2 rocket, I'm still somewhat sceptical though since we've never built a building 1km tall, never mind 11!
Remeber once a law is passed a government can choose how closely it enforces it. You cannot.
You have to follow the law, it doesn't have to uphold it (if it should or not is another issue) A law that can be applied at will can be used to cause trouble for the people you don't like.
Add to that any camera, or indeed, the good old pen and pencil.
Sorry, can't chaulk on cave walls, you could breach copyright!
Just watch then stick a "digital" in there somewhere to avoid this...
"Free speech has this proviso, but this is a world away from what we're talking about here."
Right, so you agree with the point I'm trying to make then that speech shouldn't be free if it causes others harm. The problem is where to draw the line, should I not be allowed to talk openly about how to build a home-made bomb because it could help any terrorists? Should i not be able to criticise the government because a revolution will cause harm/death to a lot of people?
It's all about where you draw the line and that is a matter of opinion, not facts or rights. How much more important is one man's right to speach over anothers life (or a probablility of anothers death).
Re: Antagonising, I'm trying to stick to the generic real world case, as opposed to the actual principles. Along the same lines that in principle I don't want to have a military...
In principle you're totally right that I should be able to antagonise anyone on a verbal level without a physical consequence. But this is the real world, and there does come a time when a physical response to a verbal event is arguably justified.
FWIW I'm with the kid on this issue, the school is way out of line, I'm just trying to make the point that speech and what you say has a responsibility to the rest of society. What I was trying to say was an action that merely antagonises is probably best worth avoiding. Taking the MLK case, "a black woman drinking out of a white's fountain" may have just antagonising at the time, it may have been many things, but it started events that changed society for the better. I don't think I ever said all antagonising was bad (see the last line in my post) just that it should be done with care.
You say that,but A friend who is an instrutor had problems when he was mugged recently. His first instinct was to fight back using his techniques. He then remembered the advice to not do so and let himself get kicked in.
Good job, because even though the police caught the mugger, he got let off because of the injuries he sustained from my friend initially defending himself counted as provocation. Heaven knows what would have happened if he'd have properly disabled the mugger!
Then there was the other friend (BIG guy, into swords as a hobby, claims he has an 6' broadsword for a toy - never seen it except in photos though). Who got attacked by 3 guys and he ended up getting stabbed during trying to hold them off. At that point he went nuts and let loose his self defense/fighting training and ended up hospitalising two of them. The only reason he got off without criminal charges was because there was cctv footage of him being stabbed before he properly attacked them.