That's the classic "appeal to authority" fallacy, combined with a false dichotomy. You don't need to trust either him or me. You can simply refer to any of the numerous publicly available descriptions of the principles used in LCD displays. For example, Pacific Display Technology has an article on their web site "Basics of LCD Operation". That's sufficient to llustrate my point, though far more detailed information is readily available.
If you don't want to bother going to the trouble of reading an article on how LCD displays work, simply thinking about the implications of the specifications of the displays might suggest to you that they move something other than just electrons and photons, given that LCD response time is usually in the tens of milliseconds, with sub-10ms LCD displays being a fairly recent innovation.
What degree I have and when I received it has nothing to do with the physics of the display technology. As Richard Feynman famously observed, "you can't fool nature."
It's not being pedantic. There is a huge difference between a display technology that involves only motion of electrons and photons, and a display technology that involves moving entire molecules. Both LCD and digital paper (e-ink) are in the latter category, and no one that had even a rudimentary grasp of the technology would make that mistake.
The article quotes Len Kawell, a distinguished engineer at Microsoft:
You're moving physical objects around and that takes physical time, not like LCD displays that change the state of electrons.
He obviously has no clue how an LCD works. The applied electric field causes physical movement of the liquid crystal molecules, affecting the polarization of light. Granted the movement is primarily twisting of the molecules, but that is definitely a form of physical movement, not a process in which electrons cause emission or modulation of photons.
Actually, we haven't had a stable liberal party since the late 1700s. At some point (early-to-mid 20th century?) the definition of "liberal" changed dramatically. A classical liberal wanted the government to leave the people alone, while today the so-called liberals want the government to protect us from ourselves. The closest current thing to the original definition of liberal would be "libertarian".
I'll readily concede that the term "intellectual property" is misleading, and it would be best to get rid of the term, but that wouldn't solve the real problems we have with copyrights, patents, and trademarks.
Even if we assume that IP refers to a variety of government-created business monopolies (which is what we are normally talking about),
Perhaps I'm confused. Since there can be no natural monopolies on ideas, what else could we be talking about other than artificial monopolies created by law?
Do you acknowledge the legitimacy of intellectual property to begin with? That is, do you believe that intellectual property is a valid construct equivalent to physical property, or do you think it's illusory?
Your premise that intellectual property has to be either equivalent to physical property or illusory is mistaken. It's entirely possible, perhaps even likely, that intellectual property is a valid construct but NOT equivalent to physical property.
In fact, by its very nature it would have to not be equivalent. For example, if I infringe your intellectual property, I haven't deprived you of the use of it, as would be the case if I stole your physical property. Since the natural consequences of infringement are different, it follows that the rights should not be completely equivalent. However, that's not at all the same as saying that there shouldn't be any intellectual property rights.
I'm not talking about the original SGI machines. There was SGI hardware that implemented full OpenGL with far less than 125,000,000 transistors.
The FPGA is fully expected to do shading, texture mapping, and depth buffering. What it won't have is a large number of parallel functional units of each type. Since it is a development board, not an end-user graphics board, that is acceptable.
No one is trying to get end users to spend $1500 on this card instead of buying an ATI or Nvidia card.
The FPGA tool chain being proprietary isn't any more of a concern than the fact that the net list and microcode of your Intel Core 2 or AMD Opteron are proprietary. That is to say, it's not ideal, but it's also not a huge problem.
Note that the Xilinx FPGA tool chain does let you see the logic that it synthesizes in several low-level forms. At the lowest level, the "FPGA Editor" provides a logical view that essentially shows what has been generated for every node in the FPGA, and allows you to tweak it if desired.
It's not hard to do floating point efficiently in an FPGA, especially one that has dedicated parallel multipliers. Sure, you don't get ASIC performance, but that's not the point. The card is intended to test the logic; once it works you can synthesize an ASIC and get faster floating point and more functional units.
SGI originally made fully-functional OpenGL rendering hardware with far less than 125 million transistors. Maybe that was the minimum for a competitive graphics card at the time of the first GeForce card, but it's definitely possible to render OpenGL with a lot less than that.
The FPGA card is NOT intended to compete with ATI and Nvidia. My understanding is that even the eventual ASIC version isn't intended to compete with the high-end cards.
There aren't any FPGAs, even Virtex 5, that have enough room to compare with the final ASIC. However, testing the design doesn't require hundreds of shaders. The design can be tested with a lesser number of functional units of each type in the FGPA, then synthesized with more of them for the ASIC.
62,208 logic elements is equivalent to 3-4 million gates, or 12-16 million transistors.
This card is an FPGA development card. It is intended for *testing* the rendering pipeline. It won't have as many functional units as the eventual ASIC, but it will demonstrate that the functional units and rendering pipeline work correctly.
Whether Project VGA is "way more practical" depends on what you're trying to do. The OGD1 was intended as a tool for developing a graphics card with a high-performance 3D rendering pipeline. The FPGA won't compete with the ASICs from ATI and Nvidia, but is perfectly suitable for a prototype to demonstrate that the logic is functional. Eventually that logic will be put into an ASIC, and used to make graphics cards that are cost-competitive.
The OGD1 is an FPGA development board. Of course FPGAs are normally used in "these types of applications"; how else could you possibly do FPGA development?
The OGD1 isn't a video card. It's a prototyping tool. Naturally it isn't as inexpensive as a video card, nor is it recommended that people needing a video card puchase one.
I think you're far better off using R500 cards. The OGD1 isn't a video card, it's an FPGA development card that is useful for developing a video card. Eventually the project will produce an ASIC-based video card that might be a reasonable choice for situations similar to what you've described.
The Xilinx tools run fine on Linux, either on 32-bit or 64-bit x86 boxes. Even the no-charge edition (ISE WebPACK) comes with 32-bit Linux binaries, and can be run on 64-bit Linux systems. The main drawback of WebPACK is that it doesn't the largest FPGAs, including the XC3S4000 on this board. Unfortunately you do need the paid version of the tools for that.
If you can find an FPGA for which there are open-source development tools, by all means please let us know about it. Meanwhile those of us that want to get actual work done with FPGAs will make do with the closed-source tools.
People routinely appear in comp.arch.fpga saying that they will write open-source FPGA development software, but none have succeeded at that yet. Perhaps the underestimate the magnitude of the problem. Xilinx has literally thousands of man years invested into developing their tools; it's not something for which one person or a small team can knock out a functional replacement in a year or two.
I try to use open-source software as much as possible, and I release much of the software I write in my spare time under the GPL, but for certain problems, open-source software just isn't going to be practical in the near term.
I've seen Spartan 3 parts used in designs with lots of bandwidth and lots of I/Os. "Lots" is a relative term. Yes, you can get more bandwidth and more I/Os with a Virtex 4 or 5, but for many applications the Spartan 3 (or 3E, or 3A) are perfectly satisfactory.
The Spartan 3 is a professional FPGA. It happens to be relatively inexpensive as it is targeted at ASIC replacement. The performance is lower than a Virtex 4 or 5, but the price/performance ratio is much better. The last time I got quotes, Virtex 4 parts cost about ten times as much per logic element as Spartan 3 parts.
What most people seem to have overlooked is that this isn't an expensive video card. It's a midrange FPGA development card, that happens to be suitable for prototyping video card functionality. It is NOT intended that average users or even power users would buy this to use it as a video card.
The plan is that this card will be used for development of the logic for a video card, which will then be realized in an ASIC in order to produce actual video cards.
It's unclear to me where you're getting your information, but the card uses a 3S4000, which is the second-largest Spartan-3 FPGA. It has over 60,000 logic elements, each of which has a 4-LUT and a FF. The part also has 96 parallel multipliers (18x18 two's complement) and 96 18-Kbit dual-port RAMs.
Slashdot Mirror
They *might* have any number of things on one chip, but most new cell phone chipsets have AGPS at least partially to support the E-911 mandate.
It was my impression that AT&T went directly from EDGE to HSDPA without any intermediate step. However, I could easily be wrong about that.
GSM isn't 3G. AT&T's "3G" service is HSDPA.
Verizon EVDO service works fine with 3G phones. They just have to be EVDO 3G phones, as opposed to HSDPA 3G phones.
If you don't want to bother going to the trouble of reading an article on how LCD displays work, simply thinking about the implications of the specifications of the displays might suggest to you that they move something other than just electrons and photons, given that LCD response time is usually in the tens of milliseconds, with sub-10ms LCD displays being a fairly recent innovation.
What degree I have and when I received it has nothing to do with the physics of the display technology. As Richard Feynman famously observed, "you can't fool nature."
It's not being pedantic. There is a huge difference between a display technology that involves only motion of electrons and photons, and a display technology that involves moving entire molecules. Both LCD and digital paper (e-ink) are in the latter category, and no one that had even a rudimentary grasp of the technology would make that mistake.
Actually, we haven't had a stable liberal party since the late 1700s. At some point (early-to-mid 20th century?) the definition of "liberal" changed dramatically. A classical liberal wanted the government to leave the people alone, while today the so-called liberals want the government to protect us from ourselves. The closest current thing to the original definition of liberal would be "libertarian".
In fact, by its very nature it would have to not be equivalent. For example, if I infringe your intellectual property, I haven't deprived you of the use of it, as would be the case if I stole your physical property. Since the natural consequences of infringement are different, it follows that the rights should not be completely equivalent. However, that's not at all the same as saying that there shouldn't be any intellectual property rights.
The FPGA is fully expected to do shading, texture mapping, and depth buffering. What it won't have is a large number of parallel functional units of each type. Since it is a development board, not an end-user graphics board, that is acceptable.
No one is trying to get end users to spend $1500 on this card instead of buying an ATI or Nvidia card.
Note that the Xilinx FPGA tool chain does let you see the logic that it synthesizes in several low-level forms. At the lowest level, the "FPGA Editor" provides a logical view that essentially shows what has been generated for every node in the FPGA, and allows you to tweak it if desired.
It's not hard to do floating point efficiently in an FPGA, especially one that has dedicated parallel multipliers. Sure, you don't get ASIC performance, but that's not the point. The card is intended to test the logic; once it works you can synthesize an ASIC and get faster floating point and more functional units.
The FPGA card is NOT intended to compete with ATI and Nvidia. My understanding is that even the eventual ASIC version isn't intended to compete with the high-end cards.
There aren't any FPGAs, even Virtex 5, that have enough room to compare with the final ASIC. However, testing the design doesn't require hundreds of shaders. The design can be tested with a lesser number of functional units of each type in the FGPA, then synthesized with more of them for the ASIC.
62,208 logic elements is equivalent to 3-4 million gates, or 12-16 million transistors. This card is an FPGA development card. It is intended for *testing* the rendering pipeline. It won't have as many functional units as the eventual ASIC, but it will demonstrate that the functional units and rendering pipeline work correctly.
Whether Project VGA is "way more practical" depends on what you're trying to do. The OGD1 was intended as a tool for developing a graphics card with a high-performance 3D rendering pipeline. The FPGA won't compete with the ASICs from ATI and Nvidia, but is perfectly suitable for a prototype to demonstrate that the logic is functional. Eventually that logic will be put into an ASIC, and used to make graphics cards that are cost-competitive.
The OGD1 is an FPGA development board. Of course FPGAs are normally used in "these types of applications"; how else could you possibly do FPGA development? The OGD1 isn't a video card. It's a prototyping tool. Naturally it isn't as inexpensive as a video card, nor is it recommended that people needing a video card puchase one.
I think you're far better off using R500 cards. The OGD1 isn't a video card, it's an FPGA development card that is useful for developing a video card. Eventually the project will produce an ASIC-based video card that might be a reasonable choice for situations similar to what you've described.
If you can find an FPGA for which there are open-source development tools, by all means please let us know about it. Meanwhile those of us that want to get actual work done with FPGAs will make do with the closed-source tools.
People routinely appear in comp.arch.fpga saying that they will write open-source FPGA development software, but none have succeeded at that yet. Perhaps the underestimate the magnitude of the problem. Xilinx has literally thousands of man years invested into developing their tools; it's not something for which one person or a small team can knock out a functional replacement in a year or two.
I try to use open-source software as much as possible, and I release much of the software I write in my spare time under the GPL, but for certain problems, open-source software just isn't going to be practical in the near term.
I've seen Spartan 3 parts used in designs with lots of bandwidth and lots of I/Os. "Lots" is a relative term. Yes, you can get more bandwidth and more I/Os with a Virtex 4 or 5, but for many applications the Spartan 3 (or 3E, or 3A) are perfectly satisfactory.
What most people seem to have overlooked is that this isn't an expensive video card. It's a midrange FPGA development card, that happens to be suitable for prototyping video card functionality. It is NOT intended that average users or even power users would buy this to use it as a video card.
The plan is that this card will be used for development of the logic for a video card, which will then be realized in an ASIC in order to produce actual video cards.
It's unclear to me where you're getting your information, but the card uses a 3S4000, which is the second-largest Spartan-3 FPGA. It has over 60,000 logic elements, each of which has a 4-LUT and a FF. The part also has 96 parallel multipliers (18x18 two's complement) and 96 18-Kbit dual-port RAMs.
What do you mean? It's already a 3S4000, which is the second biggest FPGA in the Spartan family.