Absolutely. You don't have to recompile the code everytime you want to turn on an FPGA system. With Xilinx FPGAs, you store the object code in a reprogrammable PROM and on power-up, the FPGA just reads the PROM to find out what it's supposed to do. Altera chips integrate the PROM and keep their programming when turned off so they startup faster.
Yeah, I laughed at the reconfiguring 1000s of times per second thing too. In fact, I can't think of a single instance where I've heard of anyone dynamically reprogramming the FPGAs while the system is running. In real life, it's much more comparable to flashing the BIOS on a computer. You just don't do it all that often. But the fact that you can just might save your project from failure if you find a bug.
Do the newer Xilinx chips really take that long to load? I'm still waiting for my new boards with Spartan IIs on them to be assembled. I've got plenty of designs with 5200 series parts that take about a quarter of a second / FPGA to load since I chain them together in series.
I'm certainly not misled by marketing info and my design group at Argonne National Laboratory extensively uses FPGAs from Xilinx and Altera in our work. We use them to create special purpose systems for use in detector grids and particle accelerators. They are MUCH better than any standard computer system when it comes to processing and storing the massive quantities of data the scientists around here generate with their experiments. The ability to reprogram the FPGAs rather than designing a new piece of hardware for each experiment saves lots of time and keeps our design costs low.
As for the clock distribution, I can't speak for Altera, but Xilinx's newest chips (the Spartan II and Virtex II series) defeat this problem with several high performance DLLs (Delay Lock Loops) and special routing networks that provide clocks with essentially zero delay. (+/-60ps jitter for those that care.) So clock skew isn't even something I have to worry about as a designer anymore.
Of course, performance is only as good as the quality of the design that is downloaded to the FPGA. So simple playing probably isn't the best way to judge the speed. You have to really know how to fine tune your code to show the true horsepower in these things.
Re:The emperor looks great in those new clothes
on
FPGA Supercomputers
·
· Score: 1
Well along those lines, the newest FPGAs from Xilinx (the Virtex II series) have special features built-in to support DSP applications. And they actually outperform dedicated DSP chips from TI, etc. I could go get the numbers to back that up if you'd like but it could take a while to find in this messy office.
You know, my friends and I have joked about MS (and other companies) having backdoors in their software for years. We've always known the NSA has incredible powers when it comes to finding out what they want to know. It's only a quick leap of logic to assume they've got systems that make Carnivore look like a toy. Besides MS, I'd say the next biggest target is Cisco's routers. Think about the incredible potential if the NSA can just reroute anything they want through their own network.
You may be required by law to sell a product at the advertised price, I don't know. But the point here is that the consumer put the "wrong" price on the product. If the merchant actually advertises a $3 laptop, that's one thing. But if the merchant advertises it as $1999 and the consumer tricks them into charging $3, then it's bait and switch on the part of the consumer.
But as a grad student the CE courses were generally easier and more "dot-com whore" oriented than the CS courses, oddly enough.
Weird, where did you go to school? The grad CE courses at U of Illinois (my alma mater) and other schools I'm familiar with tend to be very hardware oriented. Things like measuring ground bounce in a circuit board, EM interference, and crosstalk. That's about as far from "dot-com whore" as you can get.
I strongly agree with your first two statements. But the third is opposite of my experience. I was a CEng major and spent most of my time in courses involving logic design, circuit design, and programming at the hardware level (assembly, VHDL, and some C for embedded systems). There isn't a whole lot of difficult math involved unless you think boolean logic is hard. My CSci major friends took courses in OS and application level programming and software testing/verification. That last part is VERY math intensive. It means you have to be able to mathematically prove that your software does what it is supposed to do, and does NOT do what it is not supposed to do. Generally their classes involved a lot more theory while mine had more hands on practical experience building and testing things. The end result is that while the CSci majors can theoretically program circles around me, their efforts are worthless if I don't do my job right. Hmm, perhaps that explains why you think we CEngineers have a huge ego?
and, if I'm in any way wrong, feel free to shoot me down
You are in fact wrong. The poster never changed their assertion that CD singles are 1% of the RIAA's base. He did say that full albums were 92% of sales and they went up. He said the other 8% went down. But that 8% includes CD singles PLUS vinyl PLUS cassete singles PLUS cassete albums PLUS music videos PLUS anything else they sell.
As for singles, I've never puchased one because at $4-5 for a single song w/ (maybe) some remixes, it just doesn't make sense. The average CD costs $13 (more if you're silly enough to buy at Sam Goody or another of those mall stores) and has 12 songs at a little over a $1 per song. So you only need to like 3-4 of them to be better off buying the full album.
Well, if you're really paranoid, you could start by putting white-noise generators on your windows. Of course, if Big Brother notices they can't read your computer screen, they might try to convince a judge to grant a search warrant because you are clearly hiding criminal activity. Does anyone know if that would work?
For instance, I have a ps1 w/ 30+ games right now, do I like any of them? no? I have an n64 w/ 3 games, do i absolutely love all of them? Yes. I would play Zelda, Mario Kart, and StarFox every day if I had the time.
You talk about wanting to get rid of Sony because they've got lots of bad games compared to Nintendo and you think they'll disappear. Yet the above quote shows you've given a heck of a lot more money to Sony than Nintendo. So if Sony's marketing is good enough to make you buy lots of games you don't like, they'll be financially sound and survive just fine, I think.
This does not pose the slightest threat to Russia or China. The laser's range is under 200 miles, and a big, slow 747 could never get close enough to their missile fields to shoot down their ICBMs. This only threatens rogue states with missiles, like Iraq and N. Korea.
Where exactly did you get that 200 mile range?. The article refers to shooting down missles "hundreds of miles away at the speed of light" which puts pretty much anyone in range of at least one US military base.
Well of course it sounds like the FBI stole an innocent individual's computer. The "innocent" individual wrote the article. How do we know he isn't lying like crazy? Besides, the guy's point was that we need to be more paranoid so we should be questioning what he says.
Authors should release "freebies" to the net which have embedded links to their website. It would be really cool to be browsing in a Napster-like-program, find a song you like, and find a link to the artists website attached to it.
Ah, but if they did this, Digital Convergence would probably sue them claiming they invented the concept of including links with products. BTW, haven't heard much about DC lately. Are they still stupid?
And in fact one aspect of MP3 encoding is to remove frequencies beyond normal human hearing so it will strip away the data you mention making it a really poor place to hide the SDMI watermark since then ANY idiot could remove it:-)
I understand the argument you are making. Here's the additional information you need to fully see what happens.
In your example, you convert analog 0-4 to digital 0 and digital 0 to analog 2. This would seem the smart thing to do because 2 is at the center of the range of possible values. Unfortunately, the DAC will turn digital 0 into analog 0. (Continuing your example, analog 5-9 becomes digital 1 and digital 1 become analog 5). Now in a perfect world, things would still work as you propose. The problem is that in a real world application the difference between analog 4 (digital 0) and analog 5 (digital 1) is on the order of a few microvolts. Needless to say any noise in the system (and there always is a little) is most likely also on the order of a few microvolts meaning that analog 5 might turn into a 4 and then it will be a digital 0.
One way to reduce the error is to use DACs and ADCs with different bit accuracies but always greater than the resolution you really want. Then throw away the lower bits. Of course, in the case of the SDMI watermarks, we don't want to compensate like this since we want to destroy the watermarks.
I'm currently working on a project that uses analog to digital to analog conversions. From my experience, the degradation of this process is equal to at most 1 LSB (least significant bit) of the ADC. Now depending on how many bits your ADC uses, that may amount to only a trivial degradation. For a 24bit ADC, that's 1/16777216 of the signal's range. Now I don't claim to be an expert in audio systems, but since the application in question involves converting the result of this ADC into an MP3 with a lossy compression of 12:1, I seriously doubt it would be a detectable degradation.
It's impossible because DACs turn a specific digital value into an exact, corresponding analog value (+/- the error rating of the DAC) and an ADC turns a RANGE of analog values into a corresponding digital value. This means the result of ANY signal that undergoes an A->D->A or D->A->D conversion is not the original signal. It's close but the lowest bits of precision of each sample are slighty modified. The watermarking that SDMI proposes must be contained in those low bits or users would hear distorted audio. So if one uses this simple process on an SDMI compliant audio file, the watermark will become unrecognizable to an SDMI compliant player. The player will then be forced to assume the audio file is pre-SDMI and will play it fine because SDMI players are supposed to play ALL non-SDMI audio.
actually what was supposed to be blatantly obvious is that running fdisk on the mbr is the REALLY fast way to format a hard drive. forget dual-booting, DOS, and everything else because once you clear the mbr, the computer is convinced you've got a blank hard drive. I once had a hard drive with a couple different partitions on it that I needed to move and resize before installing a new OS. rather than wait for Partition Magic to safely move all the data in those partitions, I just cleared the mbr, and created new partitions. Total time was about 2 minutes.
hopefully you do this several times in a row since there is still a magnetic trace of the previous data present on the hard drive. the FBI/CIA/NSA/etc can just read what used to be there, and what was there before that, and what was there... I think I remember reading somewhere that to truly obliterate data you need to write over it with zeros, then again with ones, and repeat that 6 times!
the article says their magnetic source is about three times stronger than a refrigerator magnet. somehow I don't think that will cause much trouble for any electronic or biological systems present in this type of spacecraft.
and even worse...if you notice you're on a collision course with some asteroid, comet, or other space thing how do you change course to avoid while traveling at that kind of velocity?
Well, I admit I haven't checked into how they plan to implement the watermarking. However, what you've described is also my understanding of how MP3s are created - remove frequencies that are supposedly inaudible to humans. Therefore, it sounds like just taking a watermarked song from a SDMI CD and encoding it as an MP3 would destroy the watermark and you could trade it without fear that the silly RIAA people could track it. Now on the other hand, there are plenty of people who can actually hear the difference between an MP3 and a CD. I'm betting these same people will be pretty pissed if they purchase a SDMI CD and can hear the watermark!
Re:Some applications need the fastest cpu
on
Pentium 4 Delayed
·
· Score: 1
I know exactly what you're talking about. The FPGA design for the current project I'm working on takes almost a half hour to compile on my PIII 600. One of my coworkers recently suggested I consider switching to Altera FPGAs for future projects. He said they have a system where customers can submit their designs to one of Altera's dedicated supercomputers for compiling really large chips. I have no idea what it costs to do this though.
Slashdot Mirror
Absolutely. You don't have to recompile the code everytime you want to turn on an FPGA system. With Xilinx FPGAs, you store the object code in a reprogrammable PROM and on power-up, the FPGA just reads the PROM to find out what it's supposed to do. Altera chips integrate the PROM and keep their programming when turned off so they startup faster.
Yeah, I laughed at the reconfiguring 1000s of times per second thing too. In fact, I can't think of a single instance where I've heard of anyone dynamically reprogramming the FPGAs while the system is running. In real life, it's much more comparable to flashing the BIOS on a computer. You just don't do it all that often. But the fact that you can just might save your project from failure if you find a bug.
Do the newer Xilinx chips really take that long to load? I'm still waiting for my new boards with Spartan IIs on them to be assembled. I've got plenty of designs with 5200 series parts that take about a quarter of a second / FPGA to load since I chain them together in series.
I'm certainly not misled by marketing info and my design group at Argonne National Laboratory extensively uses FPGAs from Xilinx and Altera in our work. We use them to create special purpose systems for use in detector grids and particle accelerators. They are MUCH better than any standard computer system when it comes to processing and storing the massive quantities of data the scientists around here generate with their experiments. The ability to reprogram the FPGAs rather than designing a new piece of hardware for each experiment saves lots of time and keeps our design costs low. As for the clock distribution, I can't speak for Altera, but Xilinx's newest chips (the Spartan II and Virtex II series) defeat this problem with several high performance DLLs (Delay Lock Loops) and special routing networks that provide clocks with essentially zero delay. (+/-60ps jitter for those that care.) So clock skew isn't even something I have to worry about as a designer anymore. Of course, performance is only as good as the quality of the design that is downloaded to the FPGA. So simple playing probably isn't the best way to judge the speed. You have to really know how to fine tune your code to show the true horsepower in these things.
Well along those lines, the newest FPGAs from Xilinx (the Virtex II series) have special features built-in to support DSP applications. And they actually outperform dedicated DSP chips from TI, etc. I could go get the numbers to back that up if you'd like but it could take a while to find in this messy office.
You know, my friends and I have joked about MS (and other companies) having backdoors in their software for years. We've always known the NSA has incredible powers when it comes to finding out what they want to know. It's only a quick leap of logic to assume they've got systems that make Carnivore look like a toy. Besides MS, I'd say the next biggest target is Cisco's routers. Think about the incredible potential if the NSA can just reroute anything they want through their own network.
You may be required by law to sell a product at the advertised price, I don't know. But the point here is that the consumer put the "wrong" price on the product. If the merchant actually advertises a $3 laptop, that's one thing. But if the merchant advertises it as $1999 and the consumer tricks them into charging $3, then it's bait and switch on the part of the consumer.
But as a grad student the CE courses were generally easier and more "dot-com whore" oriented than the CS courses, oddly enough.
Weird, where did you go to school? The grad CE courses at U of Illinois (my alma mater) and other schools I'm familiar with tend to be very hardware oriented. Things like measuring ground bounce in a circuit board, EM interference, and crosstalk. That's about as far from "dot-com whore" as you can get.
I strongly agree with your first two statements. But the third is opposite of my experience. I was a CEng major and spent most of my time in courses involving logic design, circuit design, and programming at the hardware level (assembly, VHDL, and some C for embedded systems). There isn't a whole lot of difficult math involved unless you think boolean logic is hard. My CSci major friends took courses in OS and application level programming and software testing/verification. That last part is VERY math intensive. It means you have to be able to mathematically prove that your software does what it is supposed to do, and does NOT do what it is not supposed to do. Generally their classes involved a lot more theory while mine had more hands on practical experience building and testing things. The end result is that while the CSci majors can theoretically program circles around me, their efforts are worthless if I don't do my job right. Hmm, perhaps that explains why you think we CEngineers have a huge ego?
and, if I'm in any way wrong, feel free to shoot me down
You are in fact wrong. The poster never changed their assertion that CD singles are 1% of the RIAA's base. He did say that full albums were 92% of sales and they went up. He said the other 8% went down. But that 8% includes CD singles PLUS vinyl PLUS cassete singles PLUS cassete albums PLUS music videos PLUS anything else they sell.
As for singles, I've never puchased one because at $4-5 for a single song w/ (maybe) some remixes, it just doesn't make sense. The average CD costs $13 (more if you're silly enough to buy at Sam Goody or another of those mall stores) and has 12 songs at a little over a $1 per song. So you only need to like 3-4 of them to be better off buying the full album.
Well, if you're really paranoid, you could start by putting white-noise generators on your windows. Of course, if Big Brother notices they can't read your computer screen, they might try to convince a judge to grant a search warrant because you are clearly hiding criminal activity. Does anyone know if that would work?
For instance, I have a ps1 w/ 30+ games right now, do I like any of them? no? I have an n64 w/ 3 games, do i absolutely love all of them? Yes. I would play Zelda, Mario Kart, and StarFox every day if I had the time.
You talk about wanting to get rid of Sony because they've got lots of bad games compared to Nintendo and you think they'll disappear. Yet the above quote shows you've given a heck of a lot more money to Sony than Nintendo. So if Sony's marketing is good enough to make you buy lots of games you don't like, they'll be financially sound and survive just fine, I think.
This does not pose the slightest threat to Russia or China. The laser's range is under 200 miles, and a big, slow 747 could never get close enough to their missile fields to shoot down their ICBMs. This only threatens rogue states with missiles, like Iraq and N. Korea.
Where exactly did you get that 200 mile range?. The article refers to shooting down missles "hundreds of miles away at the speed of light" which puts pretty much anyone in range of at least one US military base.
Well of course it sounds like the FBI stole an innocent individual's computer. The "innocent" individual wrote the article. How do we know he isn't lying like crazy? Besides, the guy's point was that we need to be more paranoid so we should be questioning what he says.
Authors should release "freebies" to the net which have embedded links to their website. It would be really cool to be browsing in a Napster-like-program, find a song you like, and find a link to the artists website attached to it.
Ah, but if they did this, Digital Convergence would probably sue them claiming they invented the concept of including links with products. BTW, haven't heard much about DC lately. Are they still stupid?
The correct URL for this story is here
And in fact one aspect of MP3 encoding is to remove frequencies beyond normal human hearing so it will strip away the data you mention making it a really poor place to hide the SDMI watermark since then ANY idiot could remove it :-)
I understand the argument you are making. Here's the additional information you need to fully see what happens.
In your example, you convert analog 0-4 to digital 0 and digital 0 to analog 2. This would seem the smart thing to do because 2 is at the center of the range of possible values. Unfortunately, the DAC will turn digital 0 into analog 0. (Continuing your example, analog 5-9 becomes digital 1 and digital 1 become analog 5). Now in a perfect world, things would still work as you propose. The problem is that in a real world application the difference between analog 4 (digital 0) and analog 5 (digital 1) is on the order of a few microvolts. Needless to say any noise in the system (and there always is a little) is most likely also on the order of a few microvolts meaning that analog 5 might turn into a 4 and then it will be a digital 0.
One way to reduce the error is to use DACs and ADCs with different bit accuracies but always greater than the resolution you really want. Then throw away the lower bits. Of course, in the case of the SDMI watermarks, we don't want to compensate like this since we want to destroy the watermarks.
I'm currently working on a project that uses analog to digital to analog conversions. From my experience, the degradation of this process is equal to at most 1 LSB (least significant bit) of the ADC. Now depending on how many bits your ADC uses, that may amount to only a trivial degradation. For a 24bit ADC, that's 1/16777216 of the signal's range. Now I don't claim to be an expert in audio systems, but since the application in question involves converting the result of this ADC into an MP3 with a lossy compression of 12:1, I seriously doubt it would be a detectable degradation.
It's impossible because DACs turn a specific digital value into an exact, corresponding analog value (+/- the error rating of the DAC) and an ADC turns a RANGE of analog values into a corresponding digital value. This means the result of ANY signal that undergoes an A->D->A or D->A->D conversion is not the original signal. It's close but the lowest bits of precision of each sample are slighty modified. The watermarking that SDMI proposes must be contained in those low bits or users would hear distorted audio. So if one uses this simple process on an SDMI compliant audio file, the watermark will become unrecognizable to an SDMI compliant player. The player will then be forced to assume the audio file is pre-SDMI and will play it fine because SDMI players are supposed to play ALL non-SDMI audio.
actually what was supposed to be blatantly obvious is that running fdisk on the mbr is the REALLY fast way to format a hard drive. forget dual-booting, DOS, and everything else because once you clear the mbr, the computer is convinced you've got a blank hard drive. I once had a hard drive with a couple different partitions on it that I needed to move and resize before installing a new OS. rather than wait for Partition Magic to safely move all the data in those partitions, I just cleared the mbr, and created new partitions. Total time was about 2 minutes.
hopefully you do this several times in a row since there is still a magnetic trace of the previous data present on the hard drive. the FBI/CIA/NSA/etc can just read what used to be there, and what was there before that, and what was there... I think I remember reading somewhere that to truly obliterate data you need to write over it with zeros, then again with ones, and repeat that 6 times!
the article says their magnetic source is about three times stronger than a refrigerator magnet. somehow I don't think that will cause much trouble for any electronic or biological systems present in this type of spacecraft.
and even worse...if you notice you're on a collision course with some asteroid, comet, or other space thing how do you change course to avoid while traveling at that kind of velocity?
Well, I admit I haven't checked into how they plan to implement the watermarking. However, what you've described is also my understanding of how MP3s are created - remove frequencies that are supposedly inaudible to humans. Therefore, it sounds like just taking a watermarked song from a SDMI CD and encoding it as an MP3 would destroy the watermark and you could trade it without fear that the silly RIAA people could track it. Now on the other hand, there are plenty of people who can actually hear the difference between an MP3 and a CD. I'm betting these same people will be pretty pissed if they purchase a SDMI CD and can hear the watermark!
I know exactly what you're talking about. The FPGA design for the current project I'm working on takes almost a half hour to compile on my PIII 600. One of my coworkers recently suggested I consider switching to Altera FPGAs for future projects. He said they have a system where customers can submit their designs to one of Altera's dedicated supercomputers for compiling really large chips. I have no idea what it costs to do this though.