If you think about it, Intel tried to kill the x86 line at least twice (Itanium, i860), both times with absolutely no success due to market and competitor forces.
The only chip out that that's truly competed with the x86 instruction set for general purpose computing has been the PPC, but that's really only because Mac decided to go that route when they switched away from the 68000 processors. How many people ran Windows NT on Alpha, PPC, or MIPS? If Mac had offered both x86 and PPC hardware, do you think PPC would of survived?
The simple fact of the matter is, that even with all the legacy warts on the instruction set, it gets the job done. Nobody's going to give up affordability and backwards compatability in exchange for "engineering elegance" and perhaps a small performance gain, especially when you can just stick with x86 for 6 months and wait for Moore's law to catch up.
I've got a 12inch powerbook; no wireless, no external video, and no sleep. I can do without wireless and sleep, but not having external video is a real killer... I still have to use my ancient Dell when I do presentations.
And I think the same goes for almost every powerbook that's been released in the last year and a half.
So, what major linux distributions, BSD variants, or other operating systems are still using the XFree86 code base? Is the transition essentially complete?
Potassium 40 is a naturally occuring radioactive isotope of potassium. A small percentage of the potassium in your salt substitute (potassium chloride) is potassium 40, hence the low levels of radioactivity.
If you spilled potassium chloride on anything, it would not be considered a radioactive waste, because the radioactivity levels are at natural levels.
If you have potassium chloride where the levels of potassium 40 have been enriched, that would be considered radioactive waste.
I've had Fedora Core Test 3 installed for about a week now, and I gotta say, I love Gnome 2.6. It's very clean, polished, and the gnome bundled apps are consistent with each other.
That being said, I still haven't decided if I like the spatial file navigation of nautilus, although I'm trying to give it more time. I'm a command line guy, so I tend to think in "browser" mode, and I think most of the people here on/. are probably command line/browser mode entrained people.
For people who started their computer experience on Mac's, they'll probably love the new nautilus, but I started on DOS 2.0, so I might be to old of a dog to teach.
For a better rebuttal of Petreley's article (and how to access "browser" mode in Gnome 2.6), check out: Crack Pipes for Everyone!
They do ship a "Developer's" or "Xcode" CD with the retail boxed version of the OS. But if you have a preloaded version of OS X, or if you would rather be using the latest tools (patches come out all the time), you gotta sign up as a developer and download it from the Apple Developer's web site.
So yes, I would concur that 300,000 number is grossly inflated. Including me, I know at least 3 "Apple Developers" who only signed up so that they could get gcc and get fink to work. And I know precisely 0 other "Apple Developers". Doing a little math, (0/3)*300,000 = 0 real Apple developers.
I've got both OS X and Yellow Dog linux installed on my powerbook, and I've found after over a year of trying to like OS X, I very rarely boot into OS X.
I just find that OS X is slow, clunky, and annoying. And I'm comparing this to Gnome 2.4, which is an interface not widely known for its blazing speed.
And OS X has trashed itself twice... requiring a complete reinstall both times... talk about annoying as the OS X install is slow as molasses.
How much time are you wasting trying to eak out a 5% performance increase? Now multiply that time by your hourly wage, and I think you'll quickly prove that getting a processor that's 5% faster is generally going to be the more economical route.
If it's your own time, go ahead, compile from source if it makes you happy. But if you're screwing around, compiling everything from source when you're on your professor's time, you're probably wasting his money.
That's funny, because the NetBSD version for AMD64 is still experimental.... while there have been at least two stable AMD64 linux distributions for the past couple of months (Redhat Enterprise, Suse).
Maybe you'll want to give one of the stable AMD64 linux ports a spin?
As to the 100,000 number, you might want to check out the register's version.
Somewhere around 10,000 CPU's is a more realistic system (with 4000 of them coming from a single system).
I think this is actually an instance where a little proprietary software could vastly increase the use of free and open source software.
For a large number of people, the only real limiting factor for them to using Linux is that they absolutely have to be compatible with Microsoft Office. Star Office is close, but there are often subtle format changes when going between MS and Star, and if you're doing real business (and your employer is paying for MS Office) why screw around with the free solution? Not to mention Excel (I don't use spreadsheats, but last time I tried Star Office's excel alike, it sucked, correct me if it's gotten significantly better).
Look at it this way, if Star Office is a good enough option, why aren't more people running it on Windows?
Having MS Office available on Linux will make using Linux on corporate desktops a real option. Let's face it, MS Office is THE killer app for the vast majority of desktops out there. Without MS Office, most people will never try Linux.
And once people can safely transition to Linux, they always have the option of transitioning to Star Office later.
And further, once people are comfortable with using Linux as a desktop at work, they'll start to use Linux as a desktop at home. But at home, people are much more cost sensitive, and I think they'll be willing to put up with slight difference between MS Office and Star Office to save a couple hundred bucks.
Or could it be that huge ass 6MB cache that intel has been forced to put onto the itanium in order to get decent performance?
You talk about performance per clock speed, well allright, a 1.5GHz Itanium2 will beat a 1.5GHz Xeon.
What we really should be talking about is performance per transistor. 1.5 GHz Itanium2 - 410 million 3.2 GHz Xeon - 120 million?
And if you look at benchmarks, these processors perform close to identically, so the Xeon clearly wins.
Pretty amazing, considering the Itanium was suppose to offload computational workload from the processor to the compiler, so in theory it should need fewer transistors than a comparable i386 design.
And I upgraded a Redhat 8.0 machine to Fedora Core 1 from 500 miles away with one reboot.
I am seriously considering Debian for future servers though. Fedora has been stable, but I'd like to have something on the server that doesn't need to be upgraded every 6 months.
The ultra 5 is really just a PC with an ultrasparc processor. Not saying that's necessarily a bad thing, but an Ultra 5 is about as nostalgic as a Dell with a 500 MHz Pentium II.
Most modern CPU's handle floating point as 64 bits internally (with a couple extra bits for rounding stuff off, etc.). If the CPU is compliant to the IEEE specs (almost everyone is) you should get the exact same results running on one architecture as another.
The main exception is Intel, whose math processing uses 80 bit internally and is not IEEE compliant. While technically Intel's floating point math is a little more accurate (we're talking about the last bit here), it can be annoying that your results from a IA32 processor will be slightly different than from a Sparc...
Note that while Intels are 80 bit internally, as soon as you move off the CPU, you're back down to 64 bits. Someone correct me if I'm wrong, but I believe there's an option in gcc to force IA32 code to be IEEE compliant, and it works by inserting a load and save before every floating point operation (forcing the value off chip, which kills performance but causes things to get truncated to 64 bits).
If my memory serves me correctly, this was several years ago already.
And what basically happened, is the XFree86 guys did a big "fuck you very much, we'll stick with X11R6.3".
The X Consortium, realizing they were no longer in the driver's seat, had to change their licensing so that XFree86 would go along, and it would appear like the Consortium still had authority.
If anyone else recalls the actual events better, please pipe up. But the take away message is that for all intents and purposes, XFree86 is X.
And Stallman is so rabid about his ideology that he often hurts his own cause.
If you notice, the researchers picked heparin as their trial drug. There's a good reason for this, heparin cannot be taken orally and must be injected several times a day. Compared to several injections a day, you quickly see how an implanted microchip would win out. Insulin for diabetics would be another great application.
In the real world, few people would pick something that's implanted over an oral pill. How many women you know are on Norplant? Nope, they're all still taking the daily oral pill.
Slashdot Mirror
I'm in agreement with the above poster.
If you think about it, Intel tried to kill the x86 line at least twice (Itanium, i860), both times with absolutely no success due to market and competitor forces.
The only chip out that that's truly competed with the x86 instruction set for general purpose computing has been the PPC, but that's really only because Mac decided to go that route when they switched away from the 68000 processors. How many people ran Windows NT on Alpha, PPC, or MIPS? If Mac had offered both x86 and PPC hardware, do you think PPC would of survived?
The simple fact of the matter is, that even with all the legacy warts on the instruction set, it gets the job done. Nobody's going to give up affordability and backwards compatability in exchange for "engineering elegance" and perhaps a small performance gain, especially when you can just stick with x86 for 6 months and wait for Moore's law to catch up.
I've got a 12inch powerbook; no wireless, no external video, and no sleep. I can do without wireless and sleep, but not having external video is a real killer... I still have to use my ancient Dell when I do presentations.
And I think the same goes for almost every powerbook that's been released in the last year and a half.
So, what major linux distributions, BSD variants, or other operating systems are still using the XFree86 code base? Is the transition essentially complete?
Potassium 40 is a naturally occuring radioactive isotope of potassium. A small percentage of the potassium in your salt substitute (potassium chloride) is potassium 40, hence the low levels of radioactivity.
If you spilled potassium chloride on anything, it would not be considered a radioactive waste, because the radioactivity levels are at natural levels.
If you have potassium chloride where the levels of potassium 40 have been enriched, that would be considered radioactive waste.
Depends on the strength of the beta.
Tritium's beta has a mean of 6KeV... you need a beta with an energy of at least 70KeV to penetrate the dead layer of cells in your skin.
I've had Fedora Core Test 3 installed for about a week now, and I gotta say, I love Gnome 2.6. It's very clean, polished, and the gnome bundled apps are consistent with each other.
That being said, I still haven't decided if I like the spatial file navigation of nautilus, although I'm trying to give it more time. I'm a command line guy, so I tend to think in "browser" mode, and I think most of the people here on /. are probably command line/browser mode entrained people.
For people who started their computer experience on Mac's, they'll probably love the new nautilus, but I started on DOS 2.0, so I might be to old of a dog to teach.
For a better rebuttal of Petreley's article (and how to access "browser" mode in Gnome 2.6), check out: Crack Pipes for Everyone!
"I have to resort to opening the Home icon then going 'up' a level, but that feels really hokey"
/home/user directory.
In 2.6, the "home" icon really does open into your
I agree, the behaviour pre 2.6 was pretty brain dead.
Or have you tried downloading the XServer for 10.2
That's right, it's no longer available, they pulled it. You now have to buy 10.3 to be able to run the free XServer.
Interesting claim. You got benchmarks? I haven't seen anything where the G5 can consistently beat a Pentium IV, much less an Opteron.
no, the question was why run linux on mac hardware when you can run linux on cheaper i386 hardware.
and i also run linux on a mac portable (12" powerbook). i do it 'cause i got the powerbook for free, and i didn't like os x.
They do ship a "Developer's" or "Xcode" CD with the retail boxed version of the OS. But if you have a preloaded version of OS X, or if you would rather be using the latest tools (patches come out all the time), you gotta sign up as a developer and download it from the Apple Developer's web site.
So yes, I would concur that 300,000 number is grossly inflated. Including me, I know at least 3 "Apple Developers" who only signed up so that they could get gcc and get fink to work. And I know precisely 0 other "Apple Developers". Doing a little math, (0/3)*300,000 = 0 real Apple developers.
I gotta agree with Quinkin.
I've got both OS X and Yellow Dog linux installed on my powerbook, and I've found after over a year of trying to like OS X, I very rarely boot into OS X.
I just find that OS X is slow, clunky, and annoying. And I'm comparing this to Gnome 2.4, which is an interface not widely known for its blazing speed.
And OS X has trashed itself twice... requiring a complete reinstall both times... talk about annoying as the OS X install is slow as molasses.
I look at it pragmatically.
How much time are you wasting trying to eak out a 5% performance increase? Now multiply that time by your hourly wage, and I think you'll quickly prove that getting a processor that's 5% faster is generally going to be the more economical route.
If it's your own time, go ahead, compile from source if it makes you happy. But if you're screwing around, compiling everything from source when you're on your professor's time, you're probably wasting his money.
Uh, 2^64 is only 1.84E19.. which isn't even enough to address the atoms in a milligram of water...
2^128 is 3.4E38... which isn't even close to addressing the number of atoms in the earth.
You would need somewhere around 2^168 to address all the atoms in the earth.
256 bits would, however, be about right for addressing all the atoms in the universe.
That's funny, because the NetBSD version for AMD64 is still experimental.... while there have been at least two stable AMD64 linux distributions for the past couple of months (Redhat Enterprise, Suse).
Maybe you'll want to give one of the stable AMD64 linux ports a spin?
As to the 100,000 number, you might want to check out the register's version. Somewhere around 10,000 CPU's is a more realistic system (with 4000 of them coming from a single system).
I think this is actually an instance where a little proprietary software could vastly increase the use of free and open source software.
For a large number of people, the only real limiting factor for them to using Linux is that they absolutely have to be compatible with Microsoft Office. Star Office is close, but there are often subtle format changes when going between MS and Star, and if you're doing real business (and your employer is paying for MS Office) why screw around with the free solution? Not to mention Excel (I don't use spreadsheats, but last time I tried Star Office's excel alike, it sucked, correct me if it's gotten significantly better).
Look at it this way, if Star Office is a good enough option, why aren't more people running it on Windows?
Having MS Office available on Linux will make using Linux on corporate desktops a real option. Let's face it, MS Office is THE killer app for the vast majority of desktops out there. Without MS Office, most people will never try Linux.
And once people can safely transition to Linux, they always have the option of transitioning to Star Office later.
And further, once people are comfortable with using Linux as a desktop at work, they'll start to use Linux as a desktop at home. But at home, people are much more cost sensitive, and I think they'll be willing to put up with slight difference between MS Office and Star Office to save a couple hundred bucks.
It's all about lowering the transition barrier.
Or could it be that huge ass 6MB cache that intel has been forced to put onto the itanium in order to get decent performance?
You talk about performance per clock speed, well allright, a 1.5GHz Itanium2 will beat a 1.5GHz Xeon.
What we really should be talking about is performance per transistor.
1.5 GHz Itanium2 - 410 million
3.2 GHz Xeon - 120 million?
And if you look at benchmarks, these processors perform close to identically, so the Xeon clearly wins.
Pretty amazing, considering the Itanium was suppose to offload computational workload from the processor to the compiler, so in theory it should need fewer transistors than a comparable i386 design.
Personally though, I'm buying an opteron.
And I upgraded a Redhat 8.0 machine to Fedora Core 1 from 500 miles away with one reboot.
I am seriously considering Debian for future servers though. Fedora has been stable, but I'd like to have something on the server that doesn't need to be upgraded every 6 months.
Just another (single) data point to add, for the image reconstruction software I use routinely, I get these performances:
intel pentium IV, 3.2 GHz: 5.0 minutes
athlon XP, 1.533 GHz: 5.7 minutes
intel pentium III 733 MHz: 8.1 minutes
From the PIII to the PIV, a 340% increase in processor speed, I get 60% increase in performance...
Yeah, and the ultra 5 isn't one of them.
The ultra 5 is really just a PC with an ultrasparc processor. Not saying that's necessarily a bad thing, but an Ultra 5 is about as nostalgic as a Dell with a 500 MHz Pentium II.
uh, no.
floats are 32 bits.
doubles are 64 bits.
Most modern CPU's handle floating point as 64 bits internally (with a couple extra bits for rounding stuff off, etc.). If the CPU is compliant to the IEEE specs (almost everyone is) you should get the exact same results running on one architecture as another.
The main exception is Intel, whose math processing uses 80 bit internally and is not IEEE compliant. While technically Intel's floating point math is a little more accurate (we're talking about the last bit here), it can be annoying that your results from a IA32 processor will be slightly different than from a Sparc...
Note that while Intels are 80 bit internally, as soon as you move off the CPU, you're back down to 64 bits. Someone correct me if I'm wrong, but I believe there's an option in gcc to force IA32 code to be IEEE compliant, and it works by inserting a load and save before every floating point operation (forcing the value off chip, which kills performance but causes things to get truncated to 64 bits).
And I've never heard of anything using 96 bits.
If my memory serves me correctly, this was several years ago already.
And what basically happened, is the XFree86 guys did a big "fuck you very much, we'll stick with X11R6.3".
The X Consortium, realizing they were no longer in the driver's seat, had to change their licensing so that XFree86 would go along, and it would appear like the Consortium still had authority.
If anyone else recalls the actual events better, please pipe up. But the take away message is that for all intents and purposes, XFree86 is X.
And Stallman is so rabid about his ideology that he often hurts his own cause.
Well, uh, know of them. Can't say I "know" them...
The article is mostly hype.
If you notice, the researchers picked heparin as their trial drug. There's a good reason for this, heparin cannot be taken orally and must be injected several times a day. Compared to several injections a day, you quickly see how an implanted microchip would win out. Insulin for diabetics would be another great application.
In the real world, few people would pick something that's implanted over an oral pill. How many women you know are on Norplant? Nope, they're all still taking the daily oral pill.