Sure, they could all work together--but simple game theory predicts they won't.
Not quite. Such collusion would be illegal (between any two companies, not just baby Bills). And the government would surely keep a close watch on MS.
I think breakup is the best solution. Any regulation is hard to enforce and requires lots of supervision from the government. Besides, MS has a history of disregarding government's orders. ___
I actually read the opposite story. AMD already has Spitfires but does not want to release them just yet because they outperform the current Athlons. They want to wait until they have Thunderbirdth, which are even faster. After all, Spitfire is supposed to be the "value" CPU (a la Celeron) but it would be quite funny if the new value CPU outperformed their current offerings and sure as hell beat the crap out of Intel.
Any comment on that? ___
Re:Because Intel has a vastly superior reputation.
on
Athlons Sold Out
·
· Score: 2
I just want to add to the commets that have already been posted on this subject my personal experience. I have 2 machines now, AMD 486dx4-100 and AMD K6/2-300. Both work flawlessly. The 486 box is almost 5 years old now and it still works perfectly. I use it for a firewall and mail/ftp/samba server. The K6/2 is my workstation, about 2 years old. No problems there either. Next year I'm planning to buy a new machine and it will be (you guessed it) K7.
Perhaps your friend had a remarked overclocked CPU or a crappy board or a defective memory or bad video card. Or (*gasp*) perhaps he was running Windows 95 on it;-) You can't just assume it's the CPU's fault. In fact I am sure that had he replaced the AMD CPU with an Intel one, he'd still experience the same behaviour.
And yes, I agree with the post above that some people are indeed too dumb to own computers (not that I'm implying anything...;-)
As the subject says. Intel demoed 1GHz model (what's the deal with this magic number though?), but they can't actually produce them in quantity. In fact, Intel can't produce anything above 800MHz. (And even at 800MHz there are some serious supply issues).
AMD, OTOH, is able to produce 1GHz Athlons. You can even buy one (just check pricewatch.com), but it's pretty useless with its cache set to 1/3 the CPU speed...
I am waiting for Thunderbirds myself. They will have on-die cache (a la Celeron or the new P3). That will boost the speed a lot. Cache is the major bottleneck in Athlon. At 500MHz, Athlon wips P3's ass by like 40% margin, but this margin quickly evaporates as the clock speed is increased -- that is due to half-speed cache of Athlons (2/5 for 800+ and 1/3 for 1000MHz), compared to full-speed cache of P3. ___
I read the documents on the squid homepage and it says that it caches the objects as they pass through it, which speeds up static content by offloading the web server. However, squid cannot cache dynamic content (obviously -- that's why it's called *dynamic*). When a request to a cgi script is made, squid passes it on to the web server. So, I don't see how it can possibly speed up the dynamic content generation.
Also, why would you want to run squid at all, even for static content? Would it not make more sence to just use a bigger box for Apache? Or run 2 Apache boxes clustered instead of Apache + squid?
___
Re:QNX doesn't care about security.
on
QNX Crypt Cracked
·
· Score: 1
QNX is so heavily optimized for high-performance clusters
Would you care to name one QNX high-performance cluster? QNX is an embedded real time OS, no more nor less. ___
How useful is hardware encryption? How does it compare in speed and, more importantly, price/performance to software ecryption? 600+ MHz CPUs are pretty cheap these days, so wouldn't it make more sence to buy a second CPU instead of the crypto card? ___
This is a bit off topic, but I've got a Corel Netwinder and I was wondering how to install a new distribution. It's got RedHat 4.2 right now and I want to install Debian 2.2 instead. Problem is, it does not have a cd-rom or floppy, so the only way you can boot it is off the internal HD. That makes the installation kinda difficult...
As others have said, icecast would be the best solution. It's a Free (GPL'd) mp3 streaming software that works very well indeed.
Some people have suggested Real Player. That is not a viable solution for you. Real costs an arm and a leg and uses proprietary file format for which players are only available for windows. There is a beta version of RealPlayer for Linux, but it is beta and it is not Real's priority to release full-featured version. On the other hand, icecast is Free (so you're not locked into any licensing schemes), and mp3 players are available for every platform (so your listeners are not confined to windows).
Hardware is not an issue here. You can encode 192 Kbit/s mp3's faster than realtime on a Celeron 300 using Lame or BladeEnc. Since you want to stream it on the internet, you'll need to use lower bitrate (like 32Kbit/s), which requires less work of the CPU. I guesstimate that at this bitrate even Pentium 100 will do the job.
Your major issue will be bandwidth. With a T1, you'll have enough bandwidth to have only 48 cuncurrent streams... (assuming 32Kbit/s bitrate)
Actually, you can install it separately if you run "setup/net" (the DOS-style command line switches... uhhh). They describe it in the manual as "network installation" rather than the standard installation on a multi-user OS. You then have to run setup and give it some other options so that it copies about 3 Mb of stuff to your home directory.
But yeah, if you don't use the "/net" when you install it, it will copy everything to your home directory, which is completely stupid. No, it's beyond stupid. The StarOffice people must be shot for that! ___
Remember the previous DeCSS case? The judge said specifically that linking to DeCSS is allowed. So does this case actually stand any chance or is MPAA just trying the scare tactic? ___
I'm tired of all you people saying that faster procs are useless because of other bottlenecks.
I'm also tired of people such nonsense. The fact of the matter is that the CPU bus is not a bottleneck (it was proven in multitude of tests running 3d games on 66 and 100 MHz bus systems -- performance difference was less than 1%) and the memory bandwidth is not a bottleneck (it was proven in benchmarks comparing RDRAM to regualar PC100 SDRAM). However, for the 3d games, the CPU is the bottleneck at low resolutions. At higher resolutions (1024x768 and up), the video card becomes the bottleneck. (However, at any resolution the 3d card speed is much more important than the CPU speed). That is why a lowly Celeron equipped with GeForce DDR will kick a 1GHz machine with RDRAM and say a ATI card any time.
Linux ipchains model does not have separate chains for input and output. (I got a bit confused too when I first looked at it). Every packet arrives at the input chain, then passes through the forward chain (if the box forwards/masquerades packets), then leaves from the output chain. Each chain applies its filters to the packets. If a packet in accepted in the input chain, it can still be denied downstream (in the forward or output chain). However, if a packet is denied in the input chain, it does not continue to forward, etc. Why would you need 3 chains? Well, it's actually *very* convenient. All packets arrive intact to the input chain. If they are accepted, they go to the forward chain where they get forwarded/masqueraded. Forwarding changes the interface of the packet (e.g. eth0 may be external interface while eth1 internal. An incoming packet from the outside will change its interface in the forward chane from eth0 to eth1). Also, if you are using masquerading, packets get masqueraded/demasqueraded in the forward chain. So, when packets arrive to the output chain, they have already been forwarded and their headers have been rewritten by masq! I find it extremely convenient. On my home firewall (which is also a samba/nfs server for the lack of another box), I can block the smb and nfs traffic with just 1 rule (for each), inserted in the output chain. Oh yeah, also ipchains allows you to match packets by interface. For example, if eth0 is your external interface, you know ips 10.0.0.0/8, etc. cannot come from there. ipchains allows you to block these addresses coming from eth0, but allow them if they come from eth1 (where eth1 is the local interface). I'm not sure how FreeBSD handles it. Also, does FreeBSD support port forwarding? ___
All this stuff (and more) will be included in the next release of Debian (woody), which is currently (quite appropriately) marked unstable. Even RedHat didn't include XFree 4.0 in 6.2 because it's still too unstable. You can't just stick the latest and greatest stuff in the distribution. You have to let it mature a little.
Distributions with kernel 2.4, XFree 4.0, Apache 2.0, etc. will come out by the end of the year. And if you are really desperate, you can just install all this stuff yourself -- nobody is stopping you. But I prefer to wait until everything has been well-tested. Remember, Debian is considered to be the most stable distribution because they freeze it for at least 3 month and spend a lot of time on testing. It comes out rock-solid in the end. And that's why I love Debian. That's the only distro I'd put on a server. ___
khttp really bothers me. The obvious question is why? Just to get a few more points on benchmarks? (or shall I say "benchcrafts"). And risk the reliability for the sake of a little bit of speed? I *know* it can be un-compiled, but that's not the point. The point is that Linux kernel developers have fallen for the Mindcraft's "benchcrafts" and decided to speed up http by putting the web server in the kernel (a la Microsoft). Everyone knows that the major reason IIS is so unreliable is because it runs in the kernel mode. Haven't we been bashing it all along just for that? And believe me, in the "Enterprise", reliability is much *MUCH* more important than speed.
So, what do we want to put in the kernel today? ___
speaking of which, what ever happened to Duke Nukem forever? I haven't heard of it in a loooong while. Duke Nukem 3d was a *very* awesome game 3 years ago. Way cooler then Doom 2 / Heretic / Hexen / whatever at the time. So are they still making DNF? Last I heard they licensed Q2 engine for it but then decided to switch to Unreal engine... ahh ___
First of all, Linux is not Unix, and nobody claims it is. Linux was modeled after Unix and it looks & feels like Unix but it is still a Unix clone. All the Unix flavours you mentioned, and the ones you didn't are all descendants of the original AT&T Unix. (i.e. *BSD, AIX, Solaris, Irix, HP-UX were all derived from AT&T Unix at some point).
Programs written for one flavor of UNIX typically cannot be ported to another without considerable effort.
Huh? Who told you that, smart guy? Programs written for one flavor of UNIX typically can be ported to another by simply recompiling them.
The differences between the different kinds of UNIX seem to be as great as the differences between any particular implementation and other OSs.
Now this statement does not make any sence. *BSD is much closer to Solaris than it is to Windows. And even Linux is much closer to Irix than it is to MacOS
Could one port all the standard command line utilities to NT, clone one or two of the popular shells, set up the directory structure in the standard UNIX layout and call it Microsoft UNIX?"
uhhm, no. btw, do you work for Microsoft? It's just that your questions seem to suggest that you are planning to port a few shells to NT and rebrand it as Unix. hmmm, back in NT 3.1 days, MS was marketing is as "better Unix than Unix". Is that the plan again?
Is there a trial version of the Corel Office for Linux? I an not a free-software zealot and I wouldn't mind paying for Corel Office. If it's a good product that is. But I cannot justify spending even a penny for something that just barely runs. And that's why I want a trial version.
Corel WP8 for Linux was a complete joke. I could not believe they were actually trying to sell the thing. Corel Linux is also a joke. I used it for about 2 days and that was enough for me. Once again, I could not believe somebody would actually pay money for this garbage. And now, several posters have said that the long-awaited Corel Office for Linux is nothing more than the win32 version linked against WINE with little or no testing done. So, is there a trial version? I want to see for myself. Also, please reply with your experience using Corel Office. (But don't bother to reply if you've never used it before).
it literally costs 2X the Microsoft solution if we want to use KDE (the licensing of the Qt libs is more than twice the cost of W2K + Visual C++ pro).
But QT is cross-platform. It can't be emphasized enough. If you link your program against QT, you'll be able to compile it on Windoze 95/98/NT/2000/or whatever the flavour of the month is, as well as any flavour of Unix and (I believe) Mac. Just think of how much development time and money it saves! Besides, QT2.0 has lots of features that MFC does not support, like themes for instance. Have you seen KDE 2.0 screenshots? It's an eyecandy.
This question does not make any sence. 2.3 is a development tree (much like 2.1). 2.2 and 2.4 are the stable trees.
By convention, Linux kernels with the odd minor number (as in 2.1, 2.3, etc.) are development kernels (read: not stable, constantly changing). Kernels with even minor number (as in 2.0, 2.2, etc.) are stable (only changed due to bug fixes). So the "stable 2.3" is 2.4.
Slashdot Mirror
Not quite. Such collusion would be illegal (between any two companies, not just baby Bills). And the government would surely keep a close watch on MS.
I think breakup is the best solution. Any regulation is hard to enforce and requires lots of supervision from the government. Besides, MS has a history of disregarding government's orders.
___
I actually read the opposite story. AMD already has Spitfires but does not want to release them just yet because they outperform the current Athlons. They want to wait until they have Thunderbirdth, which are even faster. After all, Spitfire is supposed to be the "value" CPU (a la Celeron) but it would be quite funny if the new value CPU outperformed their current offerings and sure as hell beat the crap out of Intel.
Any comment on that?
___
I just want to add to the commets that have already been posted on this subject my personal experience. I have 2 machines now, AMD 486dx4-100 and AMD K6/2-300. Both work flawlessly. The 486 box is almost 5 years old now and it still works perfectly. I use it for a firewall and mail/ftp/samba server. The K6/2 is my workstation, about 2 years old. No problems there either. Next year I'm planning to buy a new machine and it will be (you guessed it) K7.
;-) You can't just assume it's the CPU's fault. In fact I am sure that had he replaced the AMD CPU with an Intel one, he'd still experience the same behaviour.
;-)
Perhaps your friend had a remarked overclocked CPU or a crappy board or a defective memory or bad video card. Or (*gasp*) perhaps he was running Windows 95 on it
And yes, I agree with the post above that some people are indeed too dumb to own computers (not that I'm implying anything...
___
megasword actually sounds quite reasonable (but still weird). Why don't we call it bigasssore.
___
Idiot. AGP is *64* bit. get your tinkin facts straight.
___
As the subject says. Intel demoed 1GHz model (what's the deal with this magic number though?), but they can't actually produce them in quantity. In fact, Intel can't produce anything above 800MHz. (And even at 800MHz there are some serious supply issues).
AMD, OTOH, is able to produce 1GHz Athlons. You can even buy one (just check pricewatch.com), but it's pretty useless with its cache set to 1/3 the CPU speed...
I am waiting for Thunderbirds myself. They will have on-die cache (a la Celeron or the new P3). That will boost the speed a lot. Cache is the major bottleneck in Athlon. At 500MHz, Athlon wips P3's ass by like 40% margin, but this margin quickly evaporates as the clock speed is increased -- that is due to half-speed cache of Athlons (2/5 for 800+ and 1/3 for 1000MHz), compared to full-speed cache of P3.
___
I read the documents on the squid homepage and it says that it caches the objects as they pass through it, which speeds up static content by offloading the web server. However, squid cannot cache dynamic content (obviously -- that's why it's called *dynamic*). When a request to a cgi script is made, squid passes it on to the web server. So, I don't see how it can possibly speed up the dynamic content generation.
Also, why would you want to run squid at all, even for static content? Would it not make more sence to just use a bigger box for Apache? Or run 2 Apache boxes clustered instead of Apache + squid?
___
Would you care to name one QNX high-performance cluster? QNX is an embedded real time OS, no more nor less.
___
How useful is hardware encryption? How does it compare in speed and, more importantly, price/performance to software ecryption? 600+ MHz CPUs are pretty cheap these days, so wouldn't it make more sence to buy a second CPU instead of the crypto card?
___
This is a bit off topic, but I've got a Corel Netwinder and I was wondering how to install a new distribution. It's got RedHat 4.2 right now and I want to install Debian 2.2 instead. Problem is, it does not have a cd-rom or floppy, so the only way you can boot it is off the internal HD. That makes the installation kinda difficult...
Any ideas?
___
ha-ha!
Seriously though, I couldn't feel any less sympathetic to either party.
___
As others have said, icecast would be the best solution. It's a Free (GPL'd) mp3 streaming software that works very well indeed.
Some people have suggested Real Player. That is not a viable solution for you. Real costs an arm and a leg and uses proprietary file format for which players are only available for windows. There is a beta version of RealPlayer for Linux, but it is beta and it is not Real's priority to release full-featured version. On the other hand, icecast is Free (so you're not locked into any licensing schemes), and mp3 players are available for every platform (so your listeners are not confined to windows).
Hardware is not an issue here. You can encode 192 Kbit/s mp3's faster than realtime on a Celeron 300 using Lame or BladeEnc. Since you want to stream it on the internet, you'll need to use lower bitrate (like 32Kbit/s), which requires less work of the CPU. I guesstimate that at this bitrate even Pentium 100 will do the job.
Your major issue will be bandwidth. With a T1, you'll have enough bandwidth to have only 48 cuncurrent streams... (assuming 32Kbit/s bitrate)
___
Actually, you can install it separately if you run /net" (the DOS-style command line switches... uhhh). They describe it in the manual as "network installation" rather than the standard installation on a multi-user OS. You then have to run setup and give it some other options so that it copies about 3 Mb of stuff to your home directory.
"setup
But yeah, if you don't use the "/net" when you install it, it will copy everything to your home directory, which is completely stupid. No, it's beyond stupid. The StarOffice people must be shot for that!
___
Remember the previous DeCSS case? The judge said specifically that linking to DeCSS is allowed. So does this case actually stand any chance or is MPAA just trying the scare tactic?
___
I'm also tired of people such nonsense. The fact of the matter is that the CPU bus is not a bottleneck (it was proven in multitude of tests running 3d games on 66 and 100 MHz bus systems -- performance difference was less than 1%) and the memory bandwidth is not a bottleneck (it was proven in benchmarks comparing RDRAM to regualar PC100 SDRAM). However, for the 3d games, the CPU is the bottleneck at low resolutions. At higher resolutions (1024x768 and up), the video card becomes the bottleneck. (However, at any resolution the 3d card speed is much more important than the CPU speed). That is why a lowly Celeron equipped with GeForce DDR will kick a 1GHz machine with RDRAM and say a ATI card any time.
___
Linux ipchains model does not have separate chains for input and output. (I got a bit confused too when I first looked at it). Every packet arrives at the input chain, then passes through the forward chain (if the box forwards/masquerades packets), then leaves from the output chain. Each chain applies its filters to the packets. If a packet in accepted in the input chain, it can still be denied downstream (in the forward or output chain). However, if a packet is denied in the input chain, it does not continue to forward, etc.
Why would you need 3 chains? Well, it's actually *very* convenient. All packets arrive intact to the input chain. If they are accepted, they go to the forward chain where they get forwarded/masqueraded. Forwarding changes the interface of the packet (e.g. eth0 may be external interface while eth1 internal. An incoming packet from the outside will change its interface in the forward chane from eth0 to eth1). Also, if you are using masquerading, packets get masqueraded/demasqueraded in the forward chain. So, when packets arrive to the output chain, they have already been forwarded and their headers have been rewritten by masq!
I find it extremely convenient. On my home firewall (which is also a samba/nfs server for the lack of another box), I can block the smb and nfs traffic with just 1 rule (for each), inserted in the output chain.
Oh yeah, also ipchains allows you to match packets by interface. For example, if eth0 is your external interface, you know ips 10.0.0.0/8, etc. cannot come from there. ipchains allows you to block these addresses coming from eth0, but allow them if they come from eth1 (where eth1 is the local interface). I'm not sure how FreeBSD handles it.
Also, does FreeBSD support port forwarding?
___
All this stuff (and more) will be included in the next release of Debian (woody), which is currently (quite appropriately) marked unstable. Even RedHat didn't include XFree 4.0 in 6.2 because it's still too unstable. You can't just stick the latest and greatest stuff in the distribution. You have to let it mature a little.
Distributions with kernel 2.4, XFree 4.0, Apache 2.0, etc. will come out by the end of the year. And if you are really desperate, you can just install all this stuff yourself -- nobody is stopping you. But I prefer to wait until everything has been well-tested. Remember, Debian is considered to be the most stable distribution because they freeze it for at least 3 month and spend a lot of time on testing. It comes out rock-solid in the end. And that's why I love Debian. That's the only distro I'd put on a server.
___
khttp really bothers me. The obvious question is why? Just to get a few more points on benchmarks? (or shall I say "benchcrafts"). And risk the reliability for the sake of a little bit of speed? I *know* it can be un-compiled, but that's not the point. The point is that Linux kernel developers have fallen for the Mindcraft's "benchcrafts" and decided to speed up http by putting the web server in the kernel (a la Microsoft). Everyone knows that the major reason IIS is so unreliable is because it runs in the kernel mode. Haven't we been bashing it all along just for that? And believe me, in the "Enterprise", reliability is much *MUCH* more important than speed.
So, what do we want to put in the kernel today?
___
speaking of which, what ever happened to Duke Nukem forever? I haven't heard of it in a loooong while. Duke Nukem 3d was a *very* awesome game 3 years ago. Way cooler then Doom 2 / Heretic / Hexen / whatever at the time. So are they still making DNF? Last I heard they licensed Q2 engine for it but then decided to switch to Unreal engine... ahh
___
A stupid question for the hardware gurus out there: what's a vector register?
___
Linux was modeled after Unix and it looks & feels like Unix but it is still
a Unix clone. All the Unix flavours you mentioned, and the ones you didn't are
all descendants of the original AT&T Unix. (i.e. *BSD, AIX, Solaris, Irix,
HP-UX were all derived from AT&T Unix at some point).
Programs written for one flavor of UNIX
typically cannot be ported to another without considerable effort.
Huh? Who told you that, smart guy? Programs written for one flavor of UNIX
typically can be ported to another by simply recompiling them.
The differences between the
different kinds of UNIX seem to be as great as the differences between
any
particular implementation and other OSs.
Now this statement does not make any sence. *BSD is much closer to Solaris
than it is to Windows. And even Linux is much closer to Irix than it is to
MacOS
Could one port all the standard
command line utilities to NT, clone one or two of the popular shells,
set up
the directory structure in the standard UNIX layout and call
it Microsoft
UNIX?"
uhhm, no. btw, do you work for Microsoft? It's just that your questions seem to
suggest that you are planning to port a few shells to NT and rebrand it as
Unix. hmmm, back in NT 3.1 days, MS was marketing is as "better Unix than
Unix". Is that the plan again?
___
Is there a trial version of the Corel Office for Linux? I an not a free-software zealot and I wouldn't mind paying for Corel Office. If it's a good product that is. But I cannot justify spending even a penny for something that just barely runs. And that's why I want a trial version.
Corel WP8 for Linux was a complete joke. I could not believe they were actually trying to sell the thing. Corel Linux is also a joke. I used it for about 2 days and that was enough for me. Once again, I could not believe somebody would actually pay money for this garbage. And now, several posters have said that the long-awaited Corel Office for Linux is nothing more than the win32 version linked against WINE with little or no testing done. So, is there a trial version? I want to see for myself. Also, please reply with your experience using Corel Office. (But don't bother to reply if you've never used it before).
___
But QT is cross-platform. It can't be emphasized enough. If you link your program against QT, you'll be able to compile it on Windoze 95/98/NT/2000/or whatever the flavour of the month is, as well as any flavour of Unix and (I believe) Mac. Just think of how much development time and money it saves! Besides, QT2.0 has lots of features that MFC does not support, like themes for instance. Have you seen KDE 2.0 screenshots? It's an eyecandy.
___
This question does not make any sence. 2.3 is a development tree (much like 2.1). 2.2 and 2.4 are the stable trees.
By convention, Linux kernels with the odd minor number (as in 2.1, 2.3, etc.) are development kernels (read: not stable, constantly changing). Kernels with even minor number (as in 2.0, 2.2, etc.) are stable (only changed due to bug fixes). So the "stable 2.3" is 2.4.
___
2.4 will support UDMA/66
right now, you can get a patch for 2.2 kernels to support UDMA/66, but it's not part of the official kernel tree.
___