Can someone explain the name? 'Lone Gunman' would make sense, and so would 'Formerly lone until they got together Gunmen', but if there are several gunmen how can they be Lone?
Pesky 'standards' getting you down? Consumers switching willy-nilly from one provider to another? Don't like the idea of this new-fangled 'interoperability'? There's an easy answer! Just make your own additional rules and refuse to allow devices that don't meet them. And the best part is, you can do this while still claiming to comply with the standard!
Yes, keep the existing format, just do something clever in the encoder.
One way would be to have a smoothing step before compression, but that's not ideal. The code doing the smoothing wouldn't have any particular knowledge of what pixel sequences compress well or badly. Better to make a hacked version of zlib which tolerates some amount of error when looking to see whether a pixel sequence has occurred previously.
Exactly. Schneier complains that the market prefers quickly-released software to secure software. He may think this is foolish. But since when was it up to him to dictate what people should and should not be able to buy? Currently you have the choice between cheap software with no liability and very expensive software sufficient checking. Some like NASA and the military may choose the expensive option, but the cheap option should be available too.
Most Slashdot readers may think it unfortunate that the market prefers Windows and MS Office to more capable alternatives, but few would argue for the more popular choice to be banned as a way of 'correcting' the market's decision.
The article talks a lot about alternatives to gasoline but I've heard that the easiest way to reduce dependence on oil imports is to adapt diesel engines. On TV they said that for a few hundred pounds any diesel car can be modified to run off vegetable oil. You can even use the leftover dirty oil from frying. If this is true I'm surprised it hasn't taken off yet - maybe we'll just have to wait for the next oil shock.
JNG isn't what I had in mind though - it's just JPEG compression underneath. I meant create an ordinary (zlib-compressed) PNG file which is almost, but not quite identical to the original image and so can compress better.
Yeah I said LZH not LZW. It used to be LZW (SEA ARC and all that) then the 'Deflate' algorithm was introduced. It's a two-stage process: look for sequences in the sliding window (Lempel-Ziv), and then if you ouput a position+length to refer back to an earlier sequence, use Huffman coding to reduce the size of these backreferences. Hence LZH.
It's true, JPEG sucks for screenshots. Yet it can be a lot smaller than GIF or even PNG, so people may use it despite the loss of quality.
I'd like to see some kind of 'lossy PNG'. PNG files are compressed using the LZH algorithm first used in pkzip, and now used in zlib, gzip and all sorts of places as well as PNG. This works by having a 'sliding window' where the last 32Kbyte or so of data is stored, and then looking back through that window to see if the current sequence of bytes has been output previously. If so, just output a referral back to where the sequence occurred before.
But what if you modified zlib so it didn't require an exact match? You might say that having one pixel in every ten out of place, or set to a 'close enough' colour (eg dark grey instead of black), is good enough to find a match. Then you could generate PNG images which lost some of the original information, but wouldn't look quite as sucky as JPEG. And any ordinary PNG viewer would be able to display them.
I don't know whether this would work well in practice but it's worth trying. Probably for dithered images it would be great but screenshots wouldn't be helped much without losing noticeable quality.
The biggest bandwidth saving on many sites would be to generate PNGs with an appropriate number of colours. Very often you can reduce your image to 8, or even 4 colours with very little loss of quality but a big reduction in image size.
What surprises me is that this is the first we (Slashdot readers) have heard about it. There have been several headlines saying 'new supercomputer planned' with a story 'it will be quite fast, and finished in 2004'... but this new world's-fastest-computer just suddenly appeared without being preannounced.
Are any of the supercomputer projects in the pipeline expected to be faster than this?
You are right - even basic PC uses require spurts of high performance, even if average utilization is low. But supposing there were a way to farm out these occasional computations to the network? Instead of bringing out new PC models with faster, shinier CPUs, manufacturers might tout the speed and low-latency of their network connections.
Ironic but statistically very unlikely. I suppose it depends on the key length. If you have a 100 bit message and a 10000 bit key, then by the pigeonhole principle there exists at least one key to 'decrypt' the message to anything you like. There are 2**100 messages, a given key can send each message to one of 2**100 encrypted strings, so there are (2**100) * (2**100) possible keys, at most. (Less than that in fact, since two different plaintexts should encrypt to different ciphertexts.)
Haven't checked this thoroughly, could well be wrong. But it sounds roughtly sensible.
What I meant was you could generate random data and they could 'break' the encryption on it just as well. In either case for any key chosen it'll decrypt to garbage (almost certainly).
If the encryption system is any good, it's not possible to distinguish between 'some random data' and 'some random data which has been encrypted with a key I'm not telling you'. They are effectively the same thing. IANAC.
Yes, a desktop does have to worry about power consumption and *definitely* space limitations. We're always being told how computers are getting smaller and smaller, yet the desktop PC hasn't shrunk at all recently, if anything it is bigger than ten years ago (minitower cases, larger CRTs).
You don't really need AGP, it's not essential for a video card (unless you play certain games). Not having PCI support is more contentious, but if the stuff you need is on the motherboard you could build a desktop PC without PCI slots. I mean what does the average desktop have beyond video, sound and Ethernet?
I'm thinking of something like the IBM PS/2 E, which was essentially a laptop in a desktop case. It had four PCMCIA slots, an LCD screen and trackpoint keyboard. Video and IDE interfaces were on the motherboard. The machine is very small and, when the HD spins down, completely silent. With its 50MHz 486SLC2 processor it wasn't a speed demon even at the time (1992), but quietness, reliability and a small footprint are sometimes more important than raw speed.
It sounds like this chip will become 'unexpectedly' popular for desktop systems as well as portables. Things like wireless Ethernet will be useful on the desktop too, and if the chip runs cool then it won't need a noisy fan. (And reduced power consumption means you can use a fan-less PSU as well.)
I wouldn't be surprised to see Intel making a desktop version of this chip as the Celeron replacement, depending on what AMD come up with.
If it's Unixlike does that mean there is a port of the X Window System? You could just get rid of that whole icky Aqua nonsense and run an X server with something nice and minimal like icewm.
- The Motherhood and Apple Pie Act - The God Bless America Act - The Will Somebody Please Think About The Children Act - The Digital Millenium... oh no they've used that one already.
Of course it's good for the computer to respond instantly. Most users would be perfectly happy with a 50MHz machine for web browsing and email (if the software got de-bloated a bit) but I'm not going to argue that cheap availability of faster processors is a bad thing.
What I meant was, people are saying that power users and specialist applications will drive the development and adoption of faster processors, as happened in the past. But does this pattern still hold if every machine is networked? Do you need a faster CPU, or just a faster broadband connection?
I'm not arguing against any particular theory, just pointing out a truism about scientific theories in general. I'm sure crackpots do the same - it doesn't mean they don't have a valid point.
Really, it shouldn't need to be pointed out that science is not truth, just the best approximation to it we have at the moment.
BTW - for this purpose I count myself as a 'Grandma' since I am just reading Slashdot and typing right now. Mozilla is bloated but still the CPU usage is tiny for most types of browsing.
Negative publicity in the wake of the Brilliant-Kazaa controversy has some industry veterans worried that consumers will switch from mindlessly clicking "I agree" to staunchly refusing to accept terms of service.
But for every power user tapping his fingers waiting for the compiler, there are a hundred Grandmas typing things into MS Word or reading web pages. Which mean a CPU utilization of about one per cent, since the computer spends most of its time waiting for human input.
Now I know there are already lots of projects to try and tap unused computing power, but it doesn't seem to have gone as far as it could. Imagine something like MOSIX distributed worldwide over the net - so when you run 'make' all sorts of random people you've never met will execute part of the job on their PCs. The protections against sabotage would be quite difficult to work out, but I'm sure it's possible.
What I'm saying is that in the past, there was always a need for faster CPUs in the individual PC. But if networks get faster and more widespread, it might turn out that individual PCs are fast enough and more effort should go into harnessing them together.
Of course, if an efficient global market did develop in computing power, then it might be worth developing faster processors just for that reason, to 'farm' them.
Slashdot Mirror
Can someone explain the name? 'Lone Gunman' would make sense, and so would 'Formerly lone until they got together Gunmen', but if there are several gunmen how can they be Lone?
Pesky 'standards' getting you down? Consumers switching willy-nilly from one provider to another? Don't like the idea of this new-fangled 'interoperability'? There's an easy answer! Just make your own additional rules and refuse to allow devices that don't meet them. And the best part is, you can do this while still claiming to comply with the standard!
Yes, keep the existing format, just do something clever in the encoder.
One way would be to have a smoothing step before compression, but that's not ideal. The code doing the smoothing wouldn't have any particular knowledge of what pixel sequences compress well or badly. Better to make a hacked version of zlib which tolerates some amount of error when looking to see whether a pixel sequence has occurred previously.
Exactly. Schneier complains that the market prefers quickly-released software to secure software. He may think this is foolish. But since when was it up to him to dictate what people should and should not be able to buy? Currently you have the choice between cheap software with no liability and very expensive software sufficient checking. Some like NASA and the military may choose the expensive option, but the cheap option should be available too.
Most Slashdot readers may think it unfortunate that the market prefers Windows and MS Office to more capable alternatives, but few would argue for the more popular choice to be banned as a way of 'correcting' the market's decision.
The article talks a lot about alternatives to gasoline but I've heard that the easiest way to reduce dependence on oil imports is to adapt diesel engines. On TV they said that for a few hundred pounds any diesel car can be modified to run off vegetable oil. You can even use the leftover dirty oil from frying. If this is true I'm surprised it hasn't taken off yet - maybe we'll just have to wait for the next oil shock.
JNG isn't what I had in mind though - it's just JPEG compression underneath. I meant create an ordinary (zlib-compressed) PNG file which is almost, but not quite identical to the original image and so can compress better.
Yeah I said LZH not LZW. It used to be LZW (SEA ARC and all that) then the 'Deflate' algorithm was introduced. It's a two-stage process: look for sequences in the sliding window (Lempel-Ziv), and then if you ouput a position+length to refer back to an earlier sequence, use Huffman coding to reduce the size of these backreferences. Hence LZH.
It's true, JPEG sucks for screenshots. Yet it can be a lot smaller than GIF or even PNG, so people may use it despite the loss of quality.
I'd like to see some kind of 'lossy PNG'. PNG files are compressed using the LZH algorithm first used in pkzip, and now used in zlib, gzip and all sorts of places as well as PNG. This works by having a 'sliding window' where the last 32Kbyte or so of data is stored, and then looking back through that window to see if the current sequence of bytes has been output previously. If so, just output a referral back to where the sequence occurred before.
But what if you modified zlib so it didn't require an exact match? You might say that having one pixel in every ten out of place, or set to a 'close enough' colour (eg dark grey instead of black), is good enough to find a match. Then you could generate PNG images which lost some of the original information, but wouldn't look quite as sucky as JPEG. And any ordinary PNG viewer would be able to display them.
I don't know whether this would work well in practice but it's worth trying. Probably for dithered images it would be great but screenshots wouldn't be helped much without losing noticeable quality.
The biggest bandwidth saving on many sites would be to generate PNGs with an appropriate number of colours. Very often you can reduce your image to 8, or even 4 colours with very little loss of quality but a big reduction in image size.
Imagine a Beowulf cluster of these!
What surprises me is that this is the first we (Slashdot readers) have heard about it. There have been several headlines saying 'new supercomputer planned' with a story 'it will be quite fast, and finished in 2004'... but this new world's-fastest-computer just suddenly appeared without being preannounced.
Are any of the supercomputer projects in the pipeline expected to be faster than this?
You are right - even basic PC uses require spurts of high performance, even if average utilization is low. But supposing there were a way to farm out these occasional computations to the network? Instead of bringing out new PC models with faster, shinier CPUs, manufacturers might tout the speed and low-latency of their network connections.
Ironic but statistically very unlikely. I suppose it depends on the key length. If you have a 100 bit message and a 10000 bit key, then by the pigeonhole principle there exists at least one key to 'decrypt' the message to anything you like. There are 2**100 messages, a given key can send each message to one of 2**100 encrypted strings, so there are (2**100) * (2**100) possible keys, at most. (Less than that in fact, since two different plaintexts should encrypt to different ciphertexts.)
Haven't checked this thoroughly, could well be wrong. But it sounds roughtly sensible.
What I meant was you could generate random data and they could 'break' the encryption on it just as well. In either case for any key chosen it'll decrypt to garbage (almost certainly).
If the encryption system is any good, it's not possible to distinguish between 'some random data' and 'some random data which has been encrypted with a key I'm not telling you'. They are effectively the same thing. IANAC.
What is the difference between encrypted random data and random data?
Yes, a desktop does have to worry about power consumption and *definitely* space limitations. We're always being told how computers are getting smaller and smaller, yet the desktop PC hasn't shrunk at all recently, if anything it is bigger than ten years ago (minitower cases, larger CRTs).
You don't really need AGP, it's not essential for a video card (unless you play certain games). Not having PCI support is more contentious, but if the stuff you need is on the motherboard you could build a desktop PC without PCI slots. I mean what does the average desktop have beyond video, sound and Ethernet?
I'm thinking of something like the IBM PS/2 E, which was essentially a laptop in a desktop case. It had four PCMCIA slots, an LCD screen and trackpoint keyboard. Video and IDE interfaces were on the motherboard. The machine is very small and, when the HD spins down, completely silent. With its 50MHz 486SLC2 processor it wasn't a speed demon even at the time (1992), but quietness, reliability and a small footprint are sometimes more important than raw speed.
It sounds like this chip will become 'unexpectedly' popular for desktop systems as well as portables. Things like wireless Ethernet will be useful on the desktop too, and if the chip runs cool then it won't need a noisy fan. (And reduced power consumption means you can use a fan-less PSU as well.)
I wouldn't be surprised to see Intel making a desktop version of this chip as the Celeron replacement, depending on what AMD come up with.
If it's Unixlike does that mean there is a port of the X Window System? You could just get rid of that whole icky Aqua nonsense and run an X server with something nice and minimal like icewm.
Coming soon from your friendly legislator:
- The Motherhood and Apple Pie Act
- The God Bless America Act
- The Will Somebody Please Think About The Children Act
- The Digital Millenium... oh no they've used that one already.
Hammer & Athlon? Isn't that a brand of baking soda or something?
Of course it's good for the computer to respond instantly. Most users would be perfectly happy with a 50MHz machine for web browsing and email (if the software got de-bloated a bit) but I'm not going to argue that cheap availability of faster processors is a bad thing.
What I meant was, people are saying that power users and specialist applications will drive the development and adoption of faster processors, as happened in the past. But does this pattern still hold if every machine is networked? Do you need a faster CPU, or just a faster broadband connection?
I'm not arguing against any particular theory, just pointing out a truism about scientific theories in general. I'm sure crackpots do the same - it doesn't mean they don't have a valid point.
Really, it shouldn't need to be pointed out that science is not truth, just the best approximation to it we have at the moment.
BTW - for this purpose I count myself as a 'Grandma' since I am just reading Slashdot and typing right now. Mozilla is bloated but still the CPU usage is tiny for most types of browsing.
But for every power user tapping his fingers waiting for the compiler, there are a hundred Grandmas typing things into MS Word or reading web pages. Which mean a CPU utilization of about one per cent, since the computer spends most of its time waiting for human input.
Now I know there are already lots of projects to try and tap unused computing power, but it doesn't seem to have gone as far as it could. Imagine something like MOSIX distributed worldwide over the net - so when you run 'make' all sorts of random people you've never met will execute part of the job on their PCs. The protections against sabotage would be quite difficult to work out, but I'm sure it's possible.
What I'm saying is that in the past, there was always a need for faster CPUs in the individual PC. But if networks get faster and more widespread, it might turn out that individual PCs are fast enough and more effort should go into harnessing them together.
Of course, if an efficient global market did develop in computing power, then it might be worth developing faster processors just for that reason, to 'farm' them.