So you think manufacturers just want to have that extra bit tacked on top? Don't you think they would rather realize extra profit from the same product by people using for things other than 3-d graphics?
I don't understand your first statement. The fact that these GPUs exist and are being used to do so many things would imply that its actually not that specialized. It just has a fat pipeline. Matrix operations are very common and many common tasks, such as web browsing even, can easily take advantage of them for image decompression and video/audio streaming. And maybe in the future if we get the whole "we don't need a dedicated coprocessor" idea out of our heads, it could be used for things like Neural Network Assistants, faster/better speech recognition, and other more complex tasks which are only not commonplace on the desktop right now because the desktop can't effectively handle it right now.
For the positioning and cooling, well there is one in there right now. There is enough space more than likely even for more than one.
Also, I'm not saying lets not give the sucker a cache. It would more than likely need a cache of its own dedicated memory to effectively operate just like any processor.
When I was about 15 and I first started reading about the first GPUs, all I could think about was, "Boy is this a step in the wrong direction." I believe in hardware whose purposes are cleanly seperated. Well, the GPU thing has had its hayday, why not start making general purpose coprocessors now so every application can get a nice boost (well a lot of applications). The instructions already resemble a normal processors anyways, so why not.
Doesn't anybody find it annoying that 3-D operation is being hardwired into the video card to begin with? Why aren't we making 200million transistor math coprocessors with high bus speeds, uncoupled from the video card. This way we wouldn't have to keep getting a new video card every time we want to upgrade our systems 3-d performance. Since these operations are highly symmetric, you could put in an array of these into one machine to incrementally upgrade. Also, this would make the issue of how to access your GPU to use for other purposes irrelevant, as it would be a math coprocessor expected to be used as such anyways. And the best reason for doing it this way: OpenGL (and DirectX too) could become more of a thick software layer on top of the generic coprocessor, and since the coprocessors would eventually standardize on common instruction set, you wouldn't need a new version of OpenGL or DirectX for every new coprocessor release. What do you guys think?
As long as everybody insists on cloning then we will never break free from Microsoft. As long as the files are called 'Microsoft Excel' files, then Microsoft will set the standard. What we need is a multiplatform, superior, competing open source product (OpenOffice is off to a good start on this). A product that makes people say, "Hey I have an OpenOffice spreadsheet to send you." not "Hey I have a Microsoft Excel file to send you." Once we quit letting Microsoft set the standards is the only time we will be free. But to do that, everything made must strive to be multiplatform INCLUDING EXCELLENT WINDOWS COMPATABILITY (NOT EXCEPTIONAL FUNCTIONALITY THOUGH), AND it must be a superior product (not about as good).
Give me a break. I have been tracking since the early 90s, and programming since I was 8. For fun I used to try to track convincing depictions of battlefields, ufo landings, explosions, etc. When I was 12 I was thinking about how to make a program I called sound stage which lets your place emitters in a 3-dimensional space that keeps track of the audio's propogation and reflections. I got bored eventially and never finished it (I guess it could've been a good thing to pursue). Lets just say, I know all about this crap:-P.
But, just the same...this would be 'cheating', it wouldn't be simulating real world audio dynamincs so much as attempting to mimic it.
A simulation is cheating, otherwise it would be called existence. I don't see how you can draw the line between simulation and mimicry as they are one and the same thing. Also, there can be just as good a realtime physical model for sound as we currently have for audio. Its not like current realtime visuals even begin to account for radiosity (I guess its possible with pixel shaders, but not pheasible at the moment). It will be a while before that can happen.
Re:Where are the "Sound Acceleration" cards?
on
Cinematic Game Graphics
·
· Score: 2, Interesting
Samples are a good analogy to Textures. 3-D positional audio is a good anology to the rendering of textured meshes, where EAX or something like what would compare to Pixel Shaders.
I don't think they need many textures/samples though. What they need is a more generic way to reproduce uniqueness using a small subset of samples. Such as having multiresolution layered textures or procedural textures, and increasing the number of simultaneous voices to allow for the combination of several samples to simulate several phases in an action at different intervals/frequencies/intensities.
Standardize on the base, common functionality, at least. And adopt a standardized linux distribution config file in/etc that describes where the distro puts different things and how it does some things. So when you install an rpm instead of an ebuild for example, it won't look like a fish out of water as far as where it puts things.
The thing keeping Linux from running everything at my workplace is:
* The prevalence of MS SQL in the workplace, a lot of proprietary systems use Jet (I guess out of pure ignorance of the alternatives)
* The necessity for an open source Small Business Server (with A/R, and Sales Orders / Invoicing / Inventory)
* The pain in the ass that is multiple packaging standards
So have a Linux Packaging Conference, and make everyone agree on a standard, and then let all those redundant repackagers package something that otherwise wouldn't have an easy way of being installed. No standard is, in this case, a very big pain.
Mod this man up. He brings attention to a fact that everyone ignores. Every new packaging format makes up their own packaging standards, and requires completely seperate repositories and users working on it. Come on now, how hard do we have to make this?
The exports to India will benefit businesses and our economy, but not necessarily me. People confuse a good economy with a good job market. They do not really correlate.
That being said, it is the process of a capitalist nation going through globalization, in an overpopulated world, that is causing this grief. Work becomes cheap when the job market is saturated with applicants (read USA+INDIA+globalization=saturated job market). The effect of this will be an initial downturn until the quality of life in both areas reach equilibrium at which point it will no longer be cheaper to pay Indian workers. This is because both sides will have equal motivation for attaining any particular job. Although, in order for this to work, universal population controls would HAVE TO BE put into place: force a 2 kid max per person law into every country that wishes to trade with us. This will ensure that the population does not grow anymore out of control than it already is, and it will increase the quality of living for everyone.
Of course the alternative is resistance. Tax everything imported until the playing field is equal. A sort of artificial imposition of equilibrium. Of course, you get the downside that there would be no 2 kid limitation so population would still explode, other countries would get pissed at you and tax your products all to hell, thereby killing exports, and if the other countries globalize without us, chances are they would have formed some really strong bonds and alliances that we will probably be strongly biased against us. So it would be harder to get in, when we realized how wrong we were, and it would also make us look like that big evil empire that everybody in the middle east seems to think we are. Also, it would make prices for things we've come to expect as being cheap bounce all around, because now we'd be buying American. So it would make our quality of life drop anyways, because we don't have enough reserves of cash and everything is suddenly more expensive. By thats just my predictions, your mileage may vary.
Your logic is assuming that there is a large enough potential viewer base to financially justify producing that form of programming over another. In other words, assuming that it is popular. All the varied, marginally popular programming (aka viewed as unique and interesting to small specific groups of people) would be snuffed out, because businesses want to make maximum profit. And max profit just won't happen for the smaller companies who don't produce the most popular forms of shows. This decision could potentially kill whats left of the uniquely interesting programming.
Does nobody think this will increase the price of the service to a point where you would probably be worse off now, than you were before. Now there are going to be a couple bazillion filters required. That and the cable company doesn't like to suddenly not make as much money, so they are going to raise prices to recoup potential losses. This will only hurt the consumer in the end.
Obviously, if they would just work with each other to create standard modularization points for different components of the system. And if they would just use standardized locations and types for shortcuts and other common things, this wouldn't even be a problem. It would be a feature. If the system was properly componentized then we would all win. I think the two projects need to merge, but the ability to select which feature from which one you want should be included. Or better yet, have a generic component interface defined for every feature and have whatever program fits the bill, fill the need.
Slashdot Mirror
So you think manufacturers just want to have that extra bit tacked on top? Don't you think they would rather realize extra profit from the same product by people using for things other than 3-d graphics?
I don't understand your first statement. The fact that these GPUs exist and are being used to do so many things would imply that its actually not that specialized. It just has a fat pipeline. Matrix operations are very common and many common tasks, such as web browsing even, can easily take advantage of them for image decompression and video/audio streaming. And maybe in the future if we get the whole "we don't need a dedicated coprocessor" idea out of our heads, it could be used for things like Neural Network Assistants, faster/better speech recognition, and other more complex tasks which are only not commonplace on the desktop right now because the desktop can't effectively handle it right now.
For the positioning and cooling, well there is one in there right now. There is enough space more than likely even for more than one.
Also, I'm not saying lets not give the sucker a cache. It would more than likely need a cache of its own dedicated memory to effectively operate just like any processor.
When I was about 15 and I first started reading about the first GPUs, all I could think about was, "Boy is this a step in the wrong direction." I believe in hardware whose purposes are cleanly seperated. Well, the GPU thing has had its hayday, why not start making general purpose coprocessors now so every application can get a nice boost (well a lot of applications). The instructions already resemble a normal processors anyways, so why not.
Doesn't anybody find it annoying that 3-D operation is being hardwired into the video card to begin with? Why aren't we making 200million transistor math coprocessors with high bus speeds, uncoupled from the video card. This way we wouldn't have to keep getting a new video card every time we want to upgrade our systems 3-d performance. Since these operations are highly symmetric, you could put in an array of these into one machine to incrementally upgrade. Also, this would make the issue of how to access your GPU to use for other purposes irrelevant, as it would be a math coprocessor expected to be used as such anyways. And the best reason for doing it this way: OpenGL (and DirectX too) could become more of a thick software layer on top of the generic coprocessor, and since the coprocessors would eventually standardize on common instruction set, you wouldn't need a new version of OpenGL or DirectX for every new coprocessor release. What do you guys think?
As long as everybody insists on cloning then we will never break free from Microsoft. As long as the files are called 'Microsoft Excel' files, then Microsoft will set the standard. What we need is a multiplatform, superior, competing open source product (OpenOffice is off to a good start on this). A product that makes people say, "Hey I have an OpenOffice spreadsheet to send you." not "Hey I have a Microsoft Excel file to send you." Once we quit letting Microsoft set the standards is the only time we will be free. But to do that, everything made must strive to be multiplatform INCLUDING EXCELLENT WINDOWS COMPATABILITY (NOT EXCEPTIONAL FUNCTIONALITY THOUGH), AND it must be a superior product (not about as good).
Very astute of you...even if by accident ;)
:-P.
Give me a break. I have been tracking since the early 90s, and programming since I was 8. For fun I used to try to track convincing depictions of battlefields, ufo landings, explosions, etc. When I was 12 I was thinking about how to make a program I called sound stage which lets your place emitters in a 3-dimensional space that keeps track of the audio's propogation and reflections. I got bored eventially and never finished it (I guess it could've been a good thing to pursue). Lets just say, I know all about this crap
But, just the same...this would be 'cheating', it wouldn't be simulating real world audio dynamincs so much as attempting to mimic it.
A simulation is cheating, otherwise it would be called existence. I don't see how you can draw the line between simulation and mimicry as they are one and the same thing. Also, there can be just as good a realtime physical model for sound as we currently have for audio. Its not like current realtime visuals even begin to account for radiosity (I guess its possible with pixel shaders, but not pheasible at the moment). It will be a while before that can happen.
Samples are a good analogy to Textures. 3-D positional audio is a good anology to the rendering of textured meshes, where EAX or something like what would compare to Pixel Shaders.
I don't think they need many textures/samples though. What they need is a more generic way to reproduce uniqueness using a small subset of samples. Such as having multiresolution layered textures or procedural textures, and increasing the number of simultaneous voices to allow for the combination of several samples to simulate several phases in an action at different intervals/frequencies/intensities.
BIND has always been buggy. Use djbdns if you want simplicity, stability, and security.
Also, there should be a standard registry between all of the installers that describes what is installed and where.
Standardize on the base, common functionality, at least. And adopt a standardized linux distribution config file in /etc that describes where the distro puts different things and how it does some things. So when you install an rpm instead of an ebuild for example, it won't look like a fish out of water as far as where it puts things.
The thing keeping Linux from running everything at my workplace is: * The prevalence of MS SQL in the workplace, a lot of proprietary systems use Jet (I guess out of pure ignorance of the alternatives) * The necessity for an open source Small Business Server (with A/R, and Sales Orders / Invoicing / Inventory) * The pain in the ass that is multiple packaging standards
So have a Linux Packaging Conference, and make everyone agree on a standard, and then let all those redundant repackagers package something that otherwise wouldn't have an easy way of being installed. No standard is, in this case, a very big pain.
Mod this man up. He brings attention to a fact that everyone ignores. Every new packaging format makes up their own packaging standards, and requires completely seperate repositories and users working on it. Come on now, how hard do we have to make this?
The exports to India will benefit businesses and our economy, but not necessarily me. People confuse a good economy with a good job market. They do not really correlate.
That being said, it is the process of a capitalist nation going through globalization, in an overpopulated world, that is causing this grief. Work becomes cheap when the job market is saturated with applicants (read USA+INDIA+globalization=saturated job market). The effect of this will be an initial downturn until the quality of life in both areas reach equilibrium at which point it will no longer be cheaper to pay Indian workers. This is because both sides will have equal motivation for attaining any particular job. Although, in order for this to work, universal population controls would HAVE TO BE put into place: force a 2 kid max per person law into every country that wishes to trade with us. This will ensure that the population does not grow anymore out of control than it already is, and it will increase the quality of living for everyone.
Of course the alternative is resistance. Tax everything imported until the playing field is equal. A sort of artificial imposition of equilibrium. Of course, you get the downside that there would be no 2 kid limitation so population would still explode, other countries would get pissed at you and tax your products all to hell, thereby killing exports, and if the other countries globalize without us, chances are they would have formed some really strong bonds and alliances that we will probably be strongly biased against us. So it would be harder to get in, when we realized how wrong we were, and it would also make us look like that big evil empire that everybody in the middle east seems to think we are. Also, it would make prices for things we've come to expect as being cheap bounce all around, because now we'd be buying American. So it would make our quality of life drop anyways, because we don't have enough reserves of cash and everything is suddenly more expensive. By thats just my predictions, your mileage may vary.
People just don't know how to Google for an answer here?
I personally would much rather see GTK adopted instead of QT, just because of licensing costs alone.
Your logic is assuming that there is a large enough potential viewer base to financially justify producing that form of programming over another. In other words, assuming that it is popular. All the varied, marginally popular programming (aka viewed as unique and interesting to small specific groups of people) would be snuffed out, because businesses want to make maximum profit. And max profit just won't happen for the smaller companies who don't produce the most popular forms of shows. This decision could potentially kill whats left of the uniquely interesting programming.
No, it mainly leads to strictly popular oppinion led programming, i.e. more reality craporama.
Does nobody think this will increase the price of the service to a point where you would probably be worse off now, than you were before. Now there are going to be a couple bazillion filters required. That and the cable company doesn't like to suddenly not make as much money, so they are going to raise prices to recoup potential losses. This will only hurt the consumer in the end.
So they are renaming DirectX, and then dubbing it cross-platform (because it runs on windows of different types). This is stupid.
Unless, you actually know what you are doing. If linux gains marketshare, then new devices might actually be supported.
Obviously, if they would just work with each other to create standard modularization points for different components of the system. And if they would just use standardized locations and types for shortcuts and other common things, this wouldn't even be a problem. It would be a feature. If the system was properly componentized then we would all win. I think the two projects need to merge, but the ability to select which feature from which one you want should be included. Or better yet, have a generic component interface defined for every feature and have whatever program fits the bill, fill the need.
The big misconception in this, is that people believed they were anonymous in the first place.
Exactly what "jet" can operate in a vacuum. Its creating a jet of what?
Planet of the Mice, just isn't quite as catchy.
Everybody, now this is your chance. Support Intel in their decision to open-source a driver, by buying their product. They are a rare breed.