Is what enabled the truly amazing compression ratio's in the first place... wavelet versus dft/dct is a lesser issue IMO.
There is no straightforward way to combine wavelets with motion compensation, thats the real problem. Although theres a lot of work been done and being done on it, wavelets are not a new theory.
Why? You would rather Slashdot would be hypocritical and ignore the fact that the majority of its readers do in fact pirate software? (including mp3's&movies)
That crypto algorythm of yours could turn out to have helped terrorists plan a bombing raid unchecked, costing the lives of masses of people. Better video compression is not inherently bad, it does facilitate pirating... but then so does wideband.
Translation costs are usually minimal, speeding them up in hardware in parallel is not efficient compared to doing it in software. So yes software is better than hardware, not faster... but better.
Strongarm gives more work per Watt for sure, and theres plenty of other high-end embedded chips which do likewise... but they cant run Windows:) Its all about legacy support, not about Linux. Which can easily be ported after all, rather ironic that Linux works for a company who's main focus is making chips for Microsofts desktop OS.
With the only feedback from their watermark extraction being essentially non real-time and not allowing for experimentation all this is is a PR stunt.
In the real world I assume we will have access to their actual hardware, sure it will still be a black box (security through obscurity, they should hope they manage to keep it obscure a little better than the DVD consortium though) but it will allow realtime feedback to trial methods of removing the watermark. If they allow you to upload a sample and give you near realtime feedback on wether the watermark could still be detected it would be a powerfull way to test their algorithms (of course they would be sensitive to DOS attacks that way). What they have is a pathetically weak one, which prooves next to nothing about how well they will do in the wild.
which are incompatible with the GPL. So you can tack on the GPL (sublicense) without the original license biting the GPL. And since all source in a GPL'd work has to be GPL you are implicitly sublicensing the original source the moment you distribute it.
AFAICS they all can be sublicensed under GPL, which you do the moment you use it in a GPL'd work.
If thats true then the original statement stands, GPL is only compatible with itself. I find licenses which allow sublicensing under GPL a rather trivial counter-example, not really worthy of mention.
"I'm surprised that nobody ever considers the possibility of parallel development tracks."
Everyone knows the story about the invention of the phone... they (the legislators) considered it alright, and then decided to forget it.
Patents only earn money for the top dogs and lawyers, forgetting the exceptions which confirm the rule, and they like it that way... and they pay the campaign dollars to keep it that way.
They could have rootkit'd your box already, checking the integrity of files now is too late. My box was hacked within the first week of installing Linux (that was fun...). I just formatted and reinstalled the whole mess and this time turned off most all of the services, since I had no idea what the weak link was and what the hell they had done to it in the mean time.
Which is a "compiler for a safe C-like language". One thing which they pointed really caught my eye though.
"we couldn't help but notice that 12 out of the 16 programs written in unsafe languages failed to meet the basic correctness criteria whereas only 4 out of the remaining 21 safe language programs were rejected."
None of my claims was shot down:) I never said the trademark cost money.
Was simply way too much FUD being spread on this thread. Noone took the trouble to read the USB 2.0 spec and the adopters agreement... simple facts were that patent licensing for USB 2.0 is free and it isnt for Firewire, and speed mixing does not lower the overall throughput.
"You're not going to convince anyone that USB 2.0 isn't a joke compared to 1394."
I wasnt trying to, I was stating mere fact and pointing out FUD. Frankly the eagerness with which people try make clearly false assumptions to proove USB 2.0's inferiority only makes me doubt the validity of the arguments they do seem to back up a little better.
I havent got the background to make technical assesment's on either, but I can read unlike some people here apperently.
You can leave all your USB 1.x hubs and devices like they are, you can use the free port on your shiny new USB 2.x host for 1 USB 2.0 device... but if you want more than 1 yes you are going to have to shell out for a new hub yeah.
There is an IEEE-1394 patent pool for which you have to pay per system, whereas the USB 2.0 adopter's agreement costs you nothing apart from a license to technique's used in your implementation of the standard.
As long as I am pointing it out in other threads might as well remind people here:)
They assume the high speed bus will kick down for low/fast speeds (those are USB 1.x speeds). Instead they use translators, all communications between high speed devices occur at high speed.
Yes I have seen articles linked below which said as much... but their respective writers were apparently a bit too much firewire zealots to point out the obvious alternative which the USB forum has chosen. The USB hub's and controller's have translator's, all communications between high speed devices occur at high speed.
You are quite correct in that it has been reported, but those reporters were quite wrong. Pity that they couldnt be content with just arguing Firewire's strongpoint but had to assume weaknesses on USB 2.0's without proper facts to back the assumption up... if they had at least pointed out the obvious alternative I would have had some respect for them, but as it is those articles were just partisan propoganda.
They assumed the same scheme of low/fast speed bus sharing would be used with low-fast/high speed bus sharing... or in other words they were assuming Intel and the rest of the USB forum are morons.
If there were morons it was them for making such stupid assumptions. Intel uses a translator to translate low-fast to high speed communication and vice versa.
As for complexity in hubs... gates are cheap, I am willing to bet you could loose the cost of the extra gates (manufacturing costs) in the IEEE-1394 patent pool licensing fee's for a single system. As for cables, fully USB 1.x compliant cable's are USB 2.0 compliant too. That said, I have no doubt what percentage of cables is fully compliant:)
And I dont think 50 cents on a 10$ mouse is ridiculously low for instance so there was a real need for USB 1.x. And from there its only a small step to seeing the market for USB 2.0, backwards compatibility and being able to use the old connectors and cabling are both very good reasons... and you save yourself a couple of cents, given how low margins tend to be in the computer biz its not unimportant.
First develop it for ages putting it in the public eye, and then all of a sudden reveal its patent encumbered. Nice...
Thats not what I wanted to talk about though, Linux the trademark is under control of Linus. So if the community was to shun Linus, they'd have to rename their fork... and anyone who thinks the name is unimportant they are terribly naive.
Slashdot Mirror
Is what enabled the truly amazing compression ratio's in the first place... wavelet versus dft/dct is a lesser issue IMO.
There is no straightforward way to combine wavelets with motion compensation, thats the real problem. Although theres a lot of work been done and being done on it, wavelets are not a new theory.
Why? You would rather Slashdot would be hypocritical and ignore the fact that the majority of its readers do in fact pirate software? (including mp3's&movies)
That crypto algorythm of yours could turn out to have helped terrorists plan a bombing raid unchecked, costing the lives of masses of people. Better video compression is not inherently bad, it does facilitate pirating... but then so does wideband.
Actually I think its amazing how well MPEG manages to keep up with progress given the amount of delay inherent in the standard process.
Translation costs are usually minimal, speeding them up in hardware in parallel is not efficient compared to doing it in software. So yes software is better than hardware, not faster... but better.
Nuff Said
Strongarm gives more work per Watt for sure, and theres plenty of other high-end embedded chips which do likewise... but they cant run Windows :) Its all about legacy support, not about Linux. Which can easily be ported after all, rather ironic that Linux works for a company who's main focus is making chips for Microsofts desktop OS.
IBM has been doing research pretty much in parallel with Transmeta. They even designed a VLIW processor with a PowerPC translation front end...
Will anyone be able to download code compiled on a PC based host to a normal L600 for testing, or will special development kits be necessary?
To what extent will Java be supported?
Will patches have to be certified individually?
But how the hell can they call that open?
With the only feedback from their watermark extraction being essentially non real-time and not allowing for experimentation all this is is a PR stunt.
In the real world I assume we will have access to their actual hardware, sure it will still be a black box (security through obscurity, they should hope they manage to keep it obscure a little better than the DVD consortium though) but it will allow realtime feedback to trial methods of removing the watermark. If they allow you to upload a sample and give you near realtime feedback on wether the watermark could still be detected it would be a powerfull way to test their algorithms (of course they would be sensitive to DOS attacks that way). What they have is a pathetically weak one, which prooves next to nothing about how well they will do in the wild.
which are incompatible with the GPL. So you can tack on the GPL (sublicense) without the original license biting the GPL. And since all source in a GPL'd work has to be GPL you are implicitly sublicensing the original source the moment you distribute it.
AFAICS they all can be sublicensed under GPL, which you do the moment you use it in a GPL'd work.
If thats true then the original statement stands, GPL is only compatible with itself. I find licenses which allow sublicensing under GPL a rather trivial counter-example, not really worthy of mention.
"I'm surprised that nobody ever considers the possibility of parallel development tracks."
Everyone knows the story about the invention of the phone... they (the legislators) considered it alright, and then decided to forget it.
Patents only earn money for the top dogs and lawyers, forgetting the exceptions which confirm the rule, and they like it that way... and they pay the campaign dollars to keep it that way.
They could have rootkit'd your box already, checking the integrity of files now is too late. My box was hacked within the first week of installing Linux (that was fun...). I just formatted and reinstalled the whole mess and this time turned off most all of the services, since I had no idea what the weak link was and what the hell they had done to it in the mean time.
The DSP unit is basically a 2 way SIMD integer MAC unit. Nothing as sofisticated (some might say bloated...) as Altivec.
Which is a "compiler for a safe C-like language". One thing which they pointed really caught my eye though.
"we couldn't help but notice that 12 out of the 16 programs written in unsafe languages failed to meet the basic correctness criteria whereas only 4 out of the remaining 21 safe language programs were rejected."
Hmmmm...
None of my claims was shot down :) I never said the trademark cost money.
Was simply way too much FUD being spread on this thread. Noone took the trouble to read the USB 2.0 spec and the adopters agreement... simple facts were that patent licensing for USB 2.0 is free and it isnt for Firewire, and speed mixing does not lower the overall throughput.
"You're not going to convince anyone that USB 2.0 isn't a joke compared to 1394."
I wasnt trying to, I was stating mere fact and pointing out FUD. Frankly the eagerness with which people try make clearly false assumptions to proove USB 2.0's inferiority only makes me doubt the validity of the arguments they do seem to back up a little better.
I havent got the background to make technical assesment's on either, but I can read unlike some people here apperently.
You can leave all your USB 1.x hubs and devices like they are, you can use the free port on your shiny new USB 2.x host for 1 USB 2.0 device... but if you want more than 1 yes you are going to have to shell out for a new hub yeah.
There is an IEEE-1394 patent pool for which you have to pay per system, whereas the USB 2.0 adopter's agreement costs you nothing apart from a license to technique's used in your implementation of the standard.
As long as I am pointing it out in other threads might as well remind people here :)
They assume the high speed bus will kick down for low/fast speeds (those are USB 1.x speeds). Instead they use translators, all communications between high speed devices occur at high speed.
Yes I have seen articles linked below which said as much... but their respective writers were apparently a bit too much firewire zealots to point out the obvious alternative which the USB forum has chosen. The USB hub's and controller's have translator's, all communications between high speed devices occur at high speed.
You are quite correct in that it has been reported, but those reporters were quite wrong. Pity that they couldnt be content with just arguing Firewire's strongpoint but had to assume weaknesses on USB 2.0's without proper facts to back the assumption up... if they had at least pointed out the obvious alternative I would have had some respect for them, but as it is those articles were just partisan propoganda.
They assumed the same scheme of low/fast speed bus sharing would be used with low-fast/high speed bus sharing... or in other words they were assuming Intel and the rest of the USB forum are morons.
:)
If there were morons it was them for making such stupid assumptions. Intel uses a translator to translate low-fast to high speed communication and vice versa.
As for complexity in hubs... gates are cheap, I am willing to bet you could loose the cost of the extra gates (manufacturing costs) in the IEEE-1394 patent pool licensing fee's for a single system. As for cables, fully USB 1.x compliant cable's are USB 2.0 compliant too. That said, I have no doubt what percentage of cables is fully compliant
A couple of cents is still more than no cents.
And I dont think 50 cents on a 10$ mouse is ridiculously low for instance so there was a real need for USB 1.x. And from there its only a small step to seeing the market for USB 2.0, backwards compatibility and being able to use the old connectors and cabling are both very good reasons... and you save yourself a couple of cents, given how low margins tend to be in the computer biz its not unimportant.
Firewire never really got over the 1$ a port fiasco, the relatively sane licensing terms were too little too late apparently.
"Mixing USB 1.X and 2.0 devices will not work well, the USB 1.X devices drag the average bus speed way down."
Why is this so? Shouldnt the hub isolate the USB 1.X devices and translate their communication to high speed USB 2.0?
First develop it for ages putting it in the public eye, and then all of a sudden reveal its patent encumbered. Nice...
Thats not what I wanted to talk about though, Linux the trademark is under control of Linus. So if the community was to shun Linus, they'd have to rename their fork... and anyone who thinks the name is unimportant they are terribly naive.