Here's Motorola's G4 fact sheet. The real lowdown on the G4 is here. Especially check out the hardware spec. (The link seems to be broken or something, though. I looked at it a few weeks ago:(
The TotalImpact page doesn't say what speed they run the L2 cache at. (The PDF spec sheet link is broken:( G4s support a range of clock divisors for the external L2 cache SRAMs, from 1:1 to 4:1. Apple uses 2:1 in their towers. (BTW, the cache RAM is external, but the control logic and stuff is all on chip.) #define X(x,y) x##y
dcypher.net or SETI@home is not a _clever_ use. It is an obvious use. So let's not waste space yacking about it.
OTOH, clusters are better when they have faster interconnections, so what if you got a mobo with a lot of PCI slots, and put a bunch of cards in it? PCI beats ethernet any day:) #define X(x,y) x##y
(I should check users.pl for replies more often. I didn't realize that you'd already replied to me on the same topic in another story, or that I'd bugged you about it so recently!)
If you post as Foogle, apparently it is possible to turn off the default score=2 bit. (I don't have that much karma, but that's what I hear.)
I think it would be more entertaining if you made your sig say "Fake Bruce" or something. I just don't like the idea of trying to fool people into thinking you're someone you're not. If you think Real Bruce sucks, why would you want to raise people's opinions of him?
Oh well, I suppose everyone has their little schemes and weirdnesses.
>... for any programmer to come from Windows to Unix it is simply not documentated as much.
You haven't found libc.info yet, have you. Open an xterm and run info libc. (or use emacs's info reader mode.) Press i to search the index for e.g. socket or getopt or thread. The info page has _good_ explanations and understandable examples that help you figure out how to do what you want to do. Of course, it only covers libc, not graphics programming. Most of the info is applicable to C in general. I do scientific computing at work, but JDS Uniphase has a tight-ass IT dept that won't let me use Linux. I do C programming on windoze 98, and I find the libc info page _very_ useful. I compile with mingw, so my code is linked again MS's C library. (info runs under cygwin, so I usually use that. I never got used to using emacs's info mode.) I would go insane without cygwin and emacs! If I wasn't just working here for the summer, I'd quit unless they let me use Linux.) #define X(x,y) x##y
I personally would pick Linux for my desktop, since I know how to use it really well. Other than that, I agree with you. Get a new account so the sensible stuff you have to say isn't posted at -1. #define X(x,y) x##y
Wow, that is a really interesting perspective! I bet you'd love to have a variable-bitrate format where you could turn up the bandwidth around the critical important moments, like that guitar note.
I got the impression that you are used to setting the controls a certain way, and leaving them for the whole track. With computers, it should be fairly easy to have an eq profile for a song, so the settings could change whenever you want. It could all be set, similar to a midi sequencer I think, so you wouldn't need to manually move sliders while encoding or any monkey business like that.
All this could make things very interesting: people who rip CDs would produce crap compared to official mastered compressed music distributed with a groups permission... This is good. #define X(x,y) x##y
The important thing to realize here is that anything they downloaded would not be compressed with their profile. All they could use the format for would be to compress stuff they had gotten through non-lossy methods (e.g. from a CD). If stuff they do hear was lost when it was compressed before they downloaded, it won't come back if they decompress and then recompress according to their "profile".
If computers improve their number crunching a lot faster than the internet grows in bandwidth, then maybe we will see sites where audiophiles can have music compressed for them personally on demand, according to their profile. Otherwise, if internet bandwidth grows a lot, we might start seeing losslessly compressed (e.g. gzip), or totally uncompressed, audio data (e.g. PCM) being sent. #define X(x,y) x##y
This (my post, _not_ the main article) is simple undergrad physics, sorry. You forgot about radiant heat, i.e. photons emitted by hot plasma, absorbed by everything with line-of-sight. This is the same effect that lets you feel a camp fire without the air around you being above normal temperature. The plasma is quite a bit hotter than any camp fire, though!
I'm wondering the same thing as the original poster, how are they going to stop the plasma's heat from warming the rest of the craft, and heating the magnets? #define X(x,y) x##y
There is another reason you left out. With a low -thrust engine, the only way to get up would be to circle around the earth gaining speed and altitude. However, you have to get through the air resistance for this to work.
Hmm, maybe a ship with low-speed wings would work, so it would just climb steeply at low speed. Still, don't forget about the fuel that hovering takes. If you're hovering, your engine is 0% efficient, since you aren't gaining any kinetic or potential energy, but you're expending fuel energy. I think this is why the space shuttle blasts off straight up with a whole lot of power, instead of going up more gradually. #define X(x,y) x##y
There is no drag in a vacuum. Space is a pretty good vacuum, so you don't have to worry about the odd hydrogen atom every now and then. IIRC, interstellar space has a density of something like 1 H atom/cubic meter. (that sounds too low, but it might be right...) Out there, you could go _really_ fast relative to the drifting H atoms before having to worry!
Therefore, the only real speed limit is the speed of light in vacuum, c = 299792458 meters per second. (This speed limit is with respect to the observer, of course.) For objects with mass, it is only possible to assymptotically approach c, and I'm sure you would run out of fuel before you were close. OTOH, pick any speed less than c, and theoretically you could build a ship with enough fuel to reach it. (If you pick anything more than a small fraction of c, you would need an absurdly large amount of fuel!) #define X(x,y) x##y
Re:Lessons to be learned from Napster....
on
Napster Wars
·
· Score: 1
You're right, but it would have been smart to post from an account that isn't trying to fool people into thinking you're the real Bruce. Get another account you can use for serious posts. Your "Bruce Perens." posts default to -1. #define X(x,y) x##y
If you buy DVDs from the US, I think you would need a modified player (or decent software on your computer) to play it, because the UK is a different region than the US. I don't know, since I'm holding off on DVDs until the corporate jugular-grab surrounding them is resolved. #define X(x,y) x##y
don't look a gift-horse in the mouth; use rpmunpack (find it on metalab or something) to get the cpio.gz out of the RPM. (RPM is just a cpio.gz with some headers, so rpmunpack just copies starting at a certain offset. It's a tiny program, and will compile on anything, since it just uses stdio, IIRC. I know I've used it to check out RPMs on a SPARC Solaris machine.)
Or, if you want a program that handles all the major Linux package formats _well_, then get alien. It does deb, rpm, slp, and (of course) tgz. You can also just tell it to unpack.
Most programs will work on other distros, even if they were packaged with RPM. I use Debian myself, but there are a lot of people that have an easier time dealing with an RPM than with anything else, so I accept the fact that people distribute stuff in RPMs. Just be glad that RPM is a well-defined standard, and that there are several Free programs to deal with them. Sounds pretty good to me! #define X(x,y) x##y
you're thinking of the POSIX.4 (realtime functions) mlock and munlock functions. NT has a POSIX subsystem, and win32 probably has an equivalent function that works on w9x and NT. #define X(x,y) x##y
> hey, GNU forms most of what we know of unices > today
Only on GNU systems! Sunx has their own ls, ps, grep, etc. for Solaris. So does IBM for AIX. I'm pretty sure the *BSDs don't use the GNU tools, for the most part, either. It is true that a lot of admins install at least bash on their commercial Unix boxes, and many install other GNU tools too. #define X(x,y) x##y
exactly. What needs to happen is that the companies that distribute a single binary for linux need to make binaries linked against glibc2.0, and glibc2.1, (libc =5 can be forgotten about, IMHO. People with libc5 systems can install libc6 and have it coexist.)
Then there's the architecture problem. Linux is not x86 Linux, but I can't think of the last time I saw a commercial demo download present more than an x86 linux option. #define X(x,y) x##y
Some companies don't even bother to provide binaries linked against libraries that are commonly used on other systems. e.g. some companies ship binaries linked against glibc 2.1, and did so when rh 6.0 was one of the only distros using 2.1 (They were early to adopt it, right?) apps linked against glib2.1 won't run on 2.0 systems, but usually 2.0 linked binaries will run on 2.1 systems.
And that's only libc. The biggest problem is probably other library versions. (X, gtk, jpeg, etc.) I don't really know, having mostly avoided commercial stuff these days, to see what it's like getting stuff done with libre apps. (I've only got a couple commercial/demo versions of things, and 18 packages from non-free installed, according to vrms, but I hardly ever use half of those. (The whole concept of vrms is hilarious! try it:) )
As an aside, it seems that to get by without commercial apps, you have to know LaTeX and some programming, to replace word processors and easy-to-use spreadsheets. I usually use AWK for physics data manipulation that most people would use a spreadsheet for. #define X(x,y) x##y
The libc info pages, and other GNU docs, often say "this is a GNU extension, and only available on GNU systems". I see it as Debian GNU/Linux is an implementation of the GNU operating system, using the Linux kernel, GNU userspace tools, with a Debian-flavoured package management/system setup style. It makes sense to me, but that's just me...:) #define X(x,y) x##y
In case you're not up on your quantum mechanics, read the recent scientific american article about quantum entanglement. It's exactly the principle used here.
Quantum entanglement provides a method for creating a one-time pad shared between two parties that are (in theory) arbitrarily far apart. All you need is a source of entangled photon pairs that is directed toward both parties. If quantum mechanics works the way we think it does, there is no, even in theory with infinite computational power, for an evesdropper to find out the secret key.
This quantum entanglement-encryption works by creating a secret key shared between two parties. This is the same as RSA or DH. The difference is in the nature of the key and the possible attacks. Quantum entanglement can generate lots of key bits, enough, in fact, that the key can be used to XOR the data. Moreover, there is _no_ way for an evesdropper to measure photons from either path without being detected. This makes even brute force attacks impossible, even in theory given infinite time. The key length equals the message length, so you would end up generating all possible messages of a given length if you tried brute force.
(sorry, last two paragraphs are a lot the same:( #define X(x,y) x##y
I've got a P200MMX, which I got ~2.5 years ago. It still runs great, especially after getting a new HD which gives me 14MB/s sequential read with UDMA:) Including some NICs, that's 300$ Canadian that I've spent on the machine in 2.5 years. I'm not using relatively old hardware on purpose because it is old, I'm just using it until I feel it is worth my money to upgrade. I can understand you being happy with your machine, but I don't understand the fact that you seem to think upgrading would be bad, even if it were free! (Hint for low memory/slow X window system usage: try wm2 (apt-get install wm2) ~50kB of data+text, rest of RSS is shared library code. I don't use it, since I don't need to be quite that minimal, but it comes in handy:). #define X(x,y) x##y
> The problem is that I have been betrayed by so called "hardware > companies" and have been given crap. If intel made things backwards > compatable with adapter modules or something I could really make this > puppy scream.
uh huh. Right. You do realize that memory bandwidth is a very important limitation, don't you? You have no choice but to upgrade the motherboard to get a faster system bus and memory controllers. You don't seriously think it's possible to make an adapter that lets you slap even 66MHz EDO RAM into an old 386 with 32pin SIMM slots, do you? Or to plug a K6-3@400MHz through an adapter into you 386 mobo, and have it do more than crawl? Even one really old component will drag the whole system down, be it RAM, vid card, disk, or (to a lesser extent) CPU. (A slow CPU can still throw the bytes around reasonably well, but waiting for a slow hard drive is _slow_)
At least you have the good sense to run Debian:) #define X(x,y) x##y
I'm planning to make my next computer a G4 tower. I'll run Linux on it, of course:) Modern Macs now have an easy-to-open case, so they obviously aren't trying to make it hard to upgrade! They have a PCI bus, so you can just throw any card in and have it work. (and work correctly, becauseISA PnP is where the praying happens, PCI PnP seems to work well. I'm sick of running a computer that is backwards compatible with an 8086, which wasn't designed with scalability in mind! BTW, anyone know where I can buy nice G4 system without a load of multi-media stuff? I just want a mobo, processor, and RAM, and maybe a new hard drive. It would be extremely cool if anyone could point me to a place where I could get a decent SMP G4 system for a decent price. (BTW, I'm in Canada) #define X(x,y) x##y
Slashdot Mirror
Here's Motorola's G4 fact sheet. The real lowdown on the G4 is here. Especially check out the hardware spec. (The link seems to be broken or something, though. I looked at it a few weeks ago :(
:( G4s support a range of clock divisors for the external L2 cache SRAMs, from 1:1 to 4:1. Apple uses 2:1 in their towers. (BTW, the cache RAM is external, but the control logic and stuff is all on chip.)
The TotalImpact page doesn't say what speed they run the L2 cache at. (The PDF spec sheet link is broken
#define X(x,y) x##y
dcypher.net or SETI@home is not a _clever_ use. It is an obvious use. So let's not waste space yacking about it.
:)
OTOH, clusters are better when they have faster interconnections, so what if you got a mobo with a lot of PCI slots, and put a bunch of cards in it? PCI beats ethernet any day
#define X(x,y) x##y
(I should check users.pl for replies more often. I didn't realize that you'd already replied to me on the same topic in another story, or that I'd bugged you about it so recently!)
:)
If you post as Foogle, apparently it is possible to turn off the default score=2 bit. (I don't have that much karma, but that's what I hear.)
I think it would be more entertaining if you made your sig say "Fake Bruce" or something. I just don't like the idea of trying to fool people into thinking you're someone you're not. If you think Real Bruce sucks, why would you want to raise people's opinions of him?
Oh well, I suppose everyone has their little schemes and weirdnesses.
Happy hacking
#define X(x,y) x##y
> ... for any programmer to come from Windows to Unix it is simply not documentated as much.
You haven't found libc.info yet, have you. Open an xterm and run info libc. (or use emacs's info reader mode.) Press i to search the index for e.g. socket or getopt or thread. The info page has _good_ explanations and understandable examples that help you figure out how to do what you want to do. Of course, it only covers libc, not graphics programming. Most of the info is applicable to C in general. I do scientific computing at work, but JDS Uniphase has a tight-ass IT dept that won't let me use Linux. I do C programming on windoze 98, and I find the libc info page _very_ useful. I compile with mingw, so my code is linked again MS's C library.
(info runs under cygwin, so I usually use that. I never got used to using emacs's info mode.) I would go insane without cygwin and emacs! If I wasn't just working here for the summer, I'd quit unless they let me use Linux.)
#define X(x,y) x##y
I personally would pick Linux for my desktop, since I know how to use it really well. Other than that, I agree with you. Get a new account so the sensible stuff you have to say isn't posted at -1.
#define X(x,y) x##y
Wow, that is a really interesting perspective! I bet you'd love to have a variable-bitrate format where you could turn up the bandwidth around the critical important moments, like that guitar note.
I got the impression that you are used to setting the controls a certain way, and leaving them for the whole track. With computers, it should be fairly easy to have an eq profile for a song, so the settings could change whenever you want. It could all be set, similar to a midi sequencer I think, so you wouldn't need to manually move sliders while encoding or any monkey business like that.
All this could make things very interesting: people who rip CDs would produce crap compared to official mastered compressed music distributed with a groups permission... This is good.
#define X(x,y) x##y
The important thing to realize here is that anything they downloaded would not be compressed with their profile. All they could use the format for would be to compress stuff they had gotten through non-lossy methods (e.g. from a CD). If stuff they do hear was lost when it was compressed before they downloaded, it won't come back if they decompress and then recompress according to their "profile".
If computers improve their number crunching a lot faster than the internet grows in bandwidth, then maybe we will see sites where audiophiles can have music compressed for them personally on demand, according to their profile. Otherwise, if internet bandwidth grows a lot, we might start seeing losslessly compressed (e.g. gzip), or totally uncompressed, audio data (e.g. PCM) being sent.
#define X(x,y) x##y
> Yes - humans lack the physical and psychological power to view the awesome site of Mars.
Mars has an awesome site? mars.org? no, I guess not. Maybe you were talking about the sight of mars.
#define X(x,y) x##y
This (my post, _not_ the main article) is simple undergrad physics, sorry. You forgot about radiant heat, i.e. photons emitted by hot plasma, absorbed by everything with line-of-sight. This is the same effect that lets you feel a camp fire without the air around you being above normal temperature. The plasma is quite a bit hotter than any camp fire, though!
I'm wondering the same thing as the original poster, how are they going to stop the plasma's heat from warming the rest of the craft, and heating the magnets?
#define X(x,y) x##y
There is another reason you left out. With a low -thrust engine, the only way to get up would be to circle around the earth gaining speed and altitude. However, you have to get through the air resistance for this to work.
Hmm, maybe a ship with low-speed wings would work, so it would just climb steeply at low speed. Still, don't forget about the fuel that hovering takes. If you're hovering, your engine is 0% efficient, since you aren't gaining any kinetic or potential energy, but you're expending fuel energy. I think this is why the space shuttle blasts off straight up with a whole lot of power, instead of going up more gradually.
#define X(x,y) x##y
There is no drag in a vacuum. Space is a pretty good vacuum, so you don't have to worry about the odd hydrogen atom every now and then. IIRC, interstellar space has a density of something like 1 H atom/cubic meter. (that sounds too low, but it might be right...) Out there, you could go _really_ fast relative to the drifting H atoms before having to worry!
Therefore, the only real speed limit is the speed of light in vacuum, c = 299792458 meters per second. (This speed limit is with respect to the observer, of course.) For objects with mass, it is only possible to assymptotically approach c, and I'm sure you would run out of fuel before you were close. OTOH, pick any speed less than c, and theoretically you could build a ship with enough fuel to reach it. (If you pick anything more than a small fraction of c, you would need an absurdly large amount of fuel!)
#define X(x,y) x##y
You're right, but it would have been smart to post from an account that isn't trying to fool people into thinking you're the real Bruce. Get another account you can use for serious posts. Your "Bruce Perens." posts default to -1.
#define X(x,y) x##y
If you buy DVDs from the US, I think you would need a modified player (or decent software on your computer) to play it, because the UK is a different region than the US. I don't know, since I'm holding off on DVDs until the corporate jugular-grab surrounding them is resolved.
#define X(x,y) x##y
don't look a gift-horse in the mouth; use rpmunpack (find it on metalab or something) to get the cpio.gz out of the RPM. (RPM is just a cpio.gz with some headers, so rpmunpack just copies starting at a certain offset. It's a tiny program, and will compile on anything, since it just uses stdio, IIRC. I know I've used it to check out RPMs on a SPARC Solaris machine.)
Or, if you want a program that handles all the major Linux package formats _well_, then get alien. It does deb, rpm, slp, and (of course) tgz. You can also just tell it to unpack.
Most programs will work on other distros, even if they were packaged with RPM. I use Debian myself, but there are a lot of people that have an easier time dealing with an RPM than with anything else, so I accept the fact that people distribute stuff in RPMs. Just be glad that RPM is a well-defined standard, and that there are several Free programs to deal with them. Sounds pretty good to me!
#define X(x,y) x##y
you're thinking of the POSIX .4 (realtime functions) mlock and munlock functions. NT has a POSIX subsystem, and win32 probably has an equivalent function that works on w9x and NT.
#define X(x,y) x##y
> Instead of
> fighting against RPM files, why not put an RPM handler in other
> distributions by default?
It's already there: alien(1)
#define X(x,y) x##y
> hey, GNU forms most of what we know of unices
> today
Only on GNU systems! Sunx has their own ls, ps, grep, etc. for Solaris. So does IBM for AIX. I'm pretty sure the *BSDs don't use the GNU tools, for the most part, either. It is true that a lot of admins install at least bash on their commercial Unix boxes, and many install other GNU tools too.
#define X(x,y) x##y
no fair, somebody probably voted with every node of their turbocluster or something :)
#define X(x,y) x##y
exactly. What needs to happen is that the companies that distribute a single binary for linux need to make binaries linked against glibc2.0, and glibc2.1, (libc =5 can be forgotten about, IMHO. People with libc5 systems can install libc6 and have it coexist.)
Then there's the architecture problem. Linux is not x86 Linux, but I can't think of the last time I saw a commercial demo download present more than an x86 linux option.
#define X(x,y) x##y
Some companies don't even bother to provide binaries linked against libraries that are commonly used on other systems. e.g. some companies ship binaries linked against glibc 2.1, and did so when rh 6.0 was one of the only distros using 2.1 (They were early to adopt it, right?) apps linked against glib2.1 won't run on 2.0 systems, but usually 2.0 linked binaries will run on 2.1 systems.
:) )
And that's only libc. The biggest problem is probably other library versions. (X, gtk, jpeg, etc.) I don't really know, having mostly avoided commercial stuff these days, to see what it's like getting stuff done with libre apps. (I've only got a couple commercial/demo versions of things, and 18 packages from non-free installed, according to vrms, but I hardly ever use half of those. (The whole concept of vrms is hilarious! try it
As an aside, it seems that to get by without commercial apps, you have to know LaTeX and some programming, to replace word processors and easy-to-use spreadsheets. I usually use AWK for physics data manipulation that most people would use a spreadsheet for.
#define X(x,y) x##y
The libc info pages, and other GNU docs, often say "this is a GNU extension, and only available on GNU systems". I see it as Debian GNU/Linux is an implementation of the GNU operating system, using the Linux kernel, GNU userspace tools, with a Debian-flavoured package management/system setup style. It makes sense to me, but that's just me... :)
#define X(x,y) x##y
In case you're not up on your quantum mechanics, read the recent scientific american article about quantum entanglement. It's exactly the principle used here.
Quantum entanglement provides a method for creating a one-time pad shared between two parties that are (in theory) arbitrarily far apart. All you need is a source of entangled photon pairs that is directed toward both parties. If quantum mechanics works the way we think it does, there is no, even in theory with infinite computational power, for an evesdropper to find out the secret key.
This quantum entanglement-encryption works by creating a secret key shared between two parties. This is the same as RSA or DH. The difference is in the nature of the key and the possible attacks. Quantum entanglement can generate lots of key bits, enough, in fact, that the key can be used to XOR the data. Moreover, there is _no_ way for an evesdropper to measure photons from either path without being detected. This makes even brute force attacks impossible, even in theory given infinite time. The key length equals the message length, so you would end up generating all possible messages of a given length if you tried brute force.
(sorry, last two paragraphs are a lot the same :(
#define X(x,y) x##y
I've got a P200MMX, which I got ~2.5 years ago. It still runs great, especially after getting a new HD which gives me 14MB/s sequential read with UDMA :) Including some NICs, that's 300$ Canadian that I've spent on the machine in 2.5 years. I'm not using relatively old hardware on purpose because it is old, I'm just using it until I feel it is worth my money to upgrade. I can understand you being happy with your machine, but I don't understand the fact that you seem to think upgrading would be bad, even if it were free! (Hint for low memory/slow X window system usage: try wm2 (apt-get install wm2) ~50kB of data+text, rest of RSS is shared library code. I don't use it, since I don't need to be quite that minimal, but it comes in handy :).
#define X(x,y) x##y
> The problem is that I have been betrayed by so called "hardware
:)
> companies" and have been given crap. If intel made things backwards
> compatable with adapter modules or something I could really make this
> puppy scream.
uh huh. Right. You do realize that memory bandwidth is a very important limitation, don't you? You have no choice but to upgrade the motherboard to get a faster system bus and memory controllers. You don't seriously think it's possible to make an adapter that lets you slap even 66MHz EDO RAM into an old 386 with 32pin SIMM slots, do you? Or to plug a K6-3@400MHz through an adapter into you 386 mobo, and have it do more than crawl? Even one really old component will drag the whole system down, be it RAM, vid card, disk, or (to a lesser extent) CPU. (A slow CPU can still throw the bytes around reasonably well, but waiting for a slow hard drive is _slow_)
At least you have the good sense to run Debian
#define X(x,y) x##y
I'm planning to make my next computer a G4 tower. I'll run Linux on it, of course :) Modern Macs now have an easy-to-open case, so they obviously aren't trying to make it hard to upgrade! They have a PCI bus, so you can just throw any card in and have it work. (and work correctly, becauseISA PnP is where the praying happens, PCI PnP seems to work well. I'm sick of running a computer that is backwards compatible with an 8086, which wasn't designed with scalability in mind!
BTW, anyone know where I can buy nice G4 system without a load of multi-media stuff? I just want a mobo, processor, and RAM, and maybe a new hard drive. It would be extremely cool if anyone could point me to a place where I could get a decent SMP G4 system for a decent price. (BTW, I'm in Canada)
#define X(x,y) x##y