Apple and Microsoft make us believe that the GUI defines the OS, oh well... I think the OS will become irrelevant. Why do so many computers still run an old system like XP? Because there's simply not enough need for "yet another great OS". It's the new hardware and slowly the newer software that drives the move to windows 7, while the recession slowed down investment in new hardware.
But the time that you had all your information stored on a PC and maybe some backup drive, is becoming history. We don't want have our data and it's associated applications tight to one location. We want have it on the road, at home, everywhere. Which is why I think the "cloud computing" trend is real, and that means that the OS for many applications will become totally irrelevant.
Unfortunately most home use PSU's have no self test build in, so you need to verify it yourself. Are the voltages in range? also under the heavy load? Almost all other units in our PC do have a POST (power on self test). So if a PC boots fine, what remains are mainly intermittent failures, which are a lot harder to uncover.
Apart from hard drive failures, the three most common types of failures you'll find on a mother board are:
1) faulty supply capacitors. These cause supply noise, or even worse, make the supply on the board itself unstable. The small ceramic caps on a board rarely fail, but the bigger elcos often do. The result can be that digital communication on the communication busses fails, or parts become unstable.Not so easy to diagnose by software. What is a clear sign, when more then one part on your board seems to fail. With a scope you can check for supply noise. Some bad elco series were easy to see, their packages were ballooning and they were leaking some brown stuff.
2) intermittent connections. This happens more often then you would guess, but certainly in modern PC's where big (memory) chips are soldered down on ball grid arrays, failing connection do occur. To diagnose it, you could try tapping on the board with the handle of a screwdriver, while your PC is running some heavy program. But that does not reveal all bad connection of the ball grid arrays. More effective is "cold spray", cooling the suspected parts down will widen the gaps of bad connections.
3) Faulty chips. I think this is actually rare, chips die pretty hard (even memory), most parts suffer from failing connections. But if a chip is on the edge of failing, that is revealed by heating them up. Give them a stress test using a heavy load, place your PC in a warm place, direct sunlight and run a stress test. This is probably best captured by software tools.
I'm very surprised about all negative comments. Yes, up to 4.1 a lot of things didn't work, were unstable. Currently I use the 4.2 RC2 release (on Gentoo), and to be honest, it just works really nice. Ok, in preferences I replaced conquerer for FF3 (those build in browser are horrible). But the configuration menu is easy and good (seems a copy of apple's) , dolphin also seems a copy from Apple's and is perfect if you ask me. I haven't seen any crashes with this latest release so far. I like how it looks and feels. At work we have gnome, which works fine too, but starts looking out dated, like XP.
I think from 4.2 and higher it will be an attractive and useful desktop, with a lot of potential.
That's how we use them. Our cad software runs on Linux, so we buy workstation that we plug into our network, and we're ready to play. All we need to do is set the network setting, NIS, and a few security setting and we're ready. Just like the old days when we had Solaris work stations.
Vista came with my laptop, and it sure does work fine. The start was a bumpy road, blue screens, slow response. But now a few patches and a service pack later, it works fine, better then XP: Disk partitioning on the fly, less problems with broken user profiles, much less clumsy handling of multiple networks while traveling.
Even though I am a hardcore Linux user, I am satisfied with Vista. To my opinion, XP is really outdated.
My thoughts exactly. And for that reason, Yahoo's engineers are not really compatible with Microsoft either. The engineers actually have the technology in their heads, if they run away, what's left of Yahoo's value?
Yes indeed, compression is the worst part, it eliminates a big part of the emotion from music. However, compression was already standard for FM broadcasting, but back then that didn't cause distortion. The biggest problem today is that they crank up the levels towards clipping. If you listen to a CD from the British band "Muse" (not even MP3), you wonder if either your amplifier or your CD player is broken. Even the "soft" parts are highly distorted. Open a track in a wave editor, and you'll see that it's constantly clipping. That's really annoying. Probably the MP3 version sounds better, as MP3 smooths the sharp square waves.
I don't agree with the article that the differences between the compression formats are small. I have no experience with AAC, but the OGG and WMA formats sound a lot more transparent then the older MP3 at equal bit rates. Certainly for classical music, with trumpets, horns or vocals, MP3 "colors" the sound a lot more.
One could argue that Windows equals communism: the same outdated desktop for everyone. With Linux you have capitalism: e.g. freedom of choice.
Anyway. I wonder the majority of instability problems on any system are related to hardware drivers. Windows depends very much on Vendors to supply these drivers. For almost all UNIX flavors however the drivers are written by the same guys that developed to OS. And for Linux most of the drivers are written by the kernel developers. Could that explain the difference in stability???
That makes only sense if the hardware specifications were open. The major problem is that they are not, and the OSS community depends on reverse engineering of Vendor created Windows drivers to create a driver.
Of course that is against the GPL-licence. But the companies have their reasons for not showing their specifications.
In circuit design it already exist, that you can pack your design as "black box module" and give it to others to compile into their designs. They cannot see the inside but they can compile and test it.
Wouldn't this be a desirable feature of GCC as well? Not only for binary drivers, but also for propriety software distribution. Then companies can even distribute it as source, and let the automake tools do their work on each platform.
Wouldn't that make the Linux/BSD platforms a lot more attractive to software vendors that nowadays need to support all kinds of flavors of platforms?
It's open source, so the driver code would be visible to everyone. So how do they keep their NDA? I think that will determine if this will become a succes.
Will they be binary drivers? Or can gcc compile scrambled code? Any ideas?
RFI: Actually the smaller te circuit, the better it's RF Immunity, because the smaller the wiring the less effective its antenna efficiency is. But I can assure you, since we crossed the 100MHz barrier, a lot has been done to improve RF immunity. Todays system boards and chips are RF-designs, also to keep reflections small and maintain signal integrity. All traces are transmission lines, which have good RF-Immunity as well.
Electro migration: This is the reason why switching currents (also known as shoot through current / overlap current) have to be reduced with feature size. Wiring already occupies a lot of die area.
That depends very much on the kind of people working in that company. If they're not too computer illetarate, it shouldn't be a problem. Linux isn't that complicated.You can easily maintain it yourself. I've worked in two companies that both switched to Linux. And after some initial setup problems, that was in both cases a succes. It is a steep learning curve, that is true. But the big advantuge of Linux is that it requires much less maintenance then Windows.
If PDF is royalty free, why not? Guaranteed WHYSIWYG, and todays printers of enough processing power on board to process PDF.
I would certainly be interested.
(Please don't bother my poor spelling, I'm no native speaker)
I'm working for a large international company (about 9000 employees world wide) which is phasing out Redmond, because it lacks proffesionalism, and they constantly change their own standards. We use open source, because of it's better (also not perfect) consistancy and much lower maintenance cost. We don't develop software ourselves, we're just users. I must admit, not standard users, all employees are engineers. We are not interested in a shiny glammer interface, the thing just needs to work. Redmond is only compatible with Redmond and nothing else, so we cannot glue applications together. That is the main reason it is phased out. What makes Redmond so expensive is that with every update something else gets broken. Often, our Sys. Op. thought he had tested the latest patch good enough, rolls it out and "bang" the network goes down again in an area he had overlooked. Due to the lack of good technical documentation, it takes a lot of time to get it up again.
With OSS the technical information is available on the internet and we know much better what each patch does. Moreover, because OSS obeys open standards much better (also not always perfectly), we can glue applicaltions together. Currently we are working with a system that is far more powerful than the shiny Redmond system. And the system downtime is reduced considerably.
While I doubt if that is really going to happen, you are certainly right that the landscape is changing. During the clockfrequency race, the same programs just ran faster. In future speed improvement has to come from parrallelism and specialized execution units, meaning that the code has to be adapted to the hardware much more then in the past. And this is certainly interersting as now we have to see who is able to utilize the new architecture faster: open source ore closed source? Especially since probably the hardware per user could differe a lot.
However, for desktop usage, it is really the applications that make the value, not the OS.
without people telling us that we're a freak.
A fat girl can be very beautiful, but that thought seems to be forbidden in our world. We geeks don't mind to be different, but it's amazing how negative people respond if your girlfriend is fat.
Slashdot Mirror
Apple and Microsoft make us believe that the GUI defines the OS, oh well... I think the OS will become irrelevant. Why do so many computers still run an old system like XP? Because there's simply not enough need for "yet another great OS". It's the new hardware and slowly the newer software that drives the move to windows 7, while the recession slowed down investment in new hardware. But the time that you had all your information stored on a PC and maybe some backup drive, is becoming history. We don't want have our data and it's associated applications tight to one location. We want have it on the road, at home, everywhere. Which is why I think the "cloud computing" trend is real, and that means that the OS for many applications will become totally irrelevant.
Unfortunately most home use PSU's have no self test build in, so you need to verify it yourself. Are the voltages in range? also under the heavy load? Almost all other units in our PC do have a POST (power on self test). So if a PC boots fine, what remains are mainly intermittent failures, which are a lot harder to uncover.
Apart from hard drive failures, the three most common types of failures you'll find on a mother board are:
1) faulty supply capacitors. These cause supply noise, or even worse, make the supply on the board itself unstable. The small ceramic caps on a board rarely fail, but the bigger elcos often do. The result can be that digital communication on the communication busses fails, or parts become unstable.Not so easy to diagnose by software. What is a clear sign, when more then one part on your board seems to fail. With a scope you can check for supply noise. Some bad elco series were easy to see, their packages were ballooning and they were leaking some brown stuff.
2) intermittent connections. This happens more often then you would guess, but certainly in modern PC's where big (memory) chips are soldered down on ball grid arrays, failing connection do occur. To diagnose it, you could try tapping on the board with the handle of a screwdriver, while your PC is running some heavy program. But that does not reveal all bad connection of the ball grid arrays. More effective is "cold spray", cooling the suspected parts down will widen the gaps of bad connections.
3) Faulty chips. I think this is actually rare, chips die pretty hard (even memory), most parts suffer from failing connections. But if a chip is on the edge of failing, that is revealed by heating them up. Give them a stress test using a heavy load, place your PC in a warm place, direct sunlight and run a stress test. This is probably best captured by software tools.
I'm very surprised about all negative comments. Yes, up to 4.1 a lot of things didn't work, were unstable. Currently I use the 4.2 RC2 release (on Gentoo), and to be honest, it just works really nice. Ok, in preferences I replaced conquerer for FF3 (those build in browser are horrible). But the configuration menu is easy and good (seems a copy of apple's) , dolphin also seems a copy from Apple's and is perfect if you ask me. I haven't seen any crashes with this latest release so far. I like how it looks and feels. At work we have gnome, which works fine too, but starts looking out dated, like XP. I think from 4.2 and higher it will be an attractive and useful desktop, with a lot of potential.
That's how we use them. Our cad software runs on Linux, so we buy workstation that we plug into our network, and we're ready to play. All we need to do is set the network setting, NIS, and a few security setting and we're ready. Just like the old days when we had Solaris work stations.
Vista came with my laptop, and it sure does work fine. The start was a bumpy road, blue screens, slow response. But now a few patches and a service pack later, it works fine, better then XP: Disk partitioning on the fly, less problems with broken user profiles, much less clumsy handling of multiple networks while traveling. Even though I am a hardcore Linux user, I am satisfied with Vista. To my opinion, XP is really outdated.
My thoughts exactly. And for that reason, Yahoo's engineers are not really compatible with Microsoft either. The engineers actually have the technology in their heads, if they run away, what's left of Yahoo's value?
Yes indeed, compression is the worst part, it eliminates a big part of the emotion from music. However, compression was already standard for FM broadcasting, but back then that didn't cause distortion. The biggest problem today is that they crank up the levels towards clipping. If you listen to a CD from the British band "Muse" (not even MP3), you wonder if either your amplifier or your CD player is broken. Even the "soft" parts are highly distorted. Open a track in a wave editor, and you'll see that it's constantly clipping. That's really annoying. Probably the MP3 version sounds better, as MP3 smooths the sharp square waves.
I don't agree with the article that the differences between the compression formats are small. I have no experience with AAC, but the OGG and WMA formats sound a lot more transparent then the older MP3 at equal bit rates. Certainly for classical music, with trumpets, horns or vocals, MP3 "colors" the sound a lot more.
One could argue that Windows equals communism: the same outdated desktop for everyone. With Linux you have capitalism: e.g. freedom of choice.
Anyway. I wonder the majority of instability problems on any system are related to hardware drivers. Windows depends very much on Vendors to supply these drivers. For almost all UNIX flavors however the drivers are written by the same guys that developed to OS. And for Linux most of the drivers are written by the kernel developers. Could that explain the difference in stability???
That makes only sense if the hardware specifications were open. The major problem is that they are not, and the OSS community depends on reverse engineering of Vendor created Windows drivers to create a driver.
Of course that is against the GPL-licence. But the companies have their reasons
for not showing their specifications.
In circuit design it already exist, that you can pack your design as "black box module"
and give it to others to compile into their designs. They cannot see the inside but they
can compile and test it.
Wouldn't this be a desirable feature of GCC as well? Not only for binary drivers, but also
for propriety software distribution. Then companies can even distribute it as source, and
let the automake tools do their work on each platform.
Wouldn't that make the Linux/BSD platforms a lot more attractive to software vendors that
nowadays need to support all kinds of flavors of platforms?
It's open source, so the driver code would be visible to everyone. So how
do they keep their NDA? I think that will determine if this will become
a succes.
Will they be binary drivers? Or can gcc compile scrambled code?
Any ideas?
RFI: Actually the smaller te circuit, the better it's RF Immunity, because the smaller the wiring the less effective its antenna efficiency is.
But I can assure you, since we crossed the 100MHz barrier, a lot has been done to improve RF immunity. Todays system boards and chips are RF-designs, also to keep reflections small and maintain signal integrity. All traces are transmission lines, which have good RF-Immunity as well.
Electro migration: This is the reason why switching currents (also known as shoot through current / overlap current) have to be reduced with feature size. Wiring already occupies a lot of die area.
That depends very much on the kind of people working in that company. If they're not too computer illetarate, it shouldn't be a problem. Linux isn't that complicated.You can easily maintain it yourself.
I've worked in two companies that both switched to Linux. And after some initial setup problems, that was in both cases a succes. It is a steep learning curve, that is true. But the big advantuge of Linux is that it requires much less maintenance then Windows.
If PDF is royalty free, why not? Guaranteed WHYSIWYG, and todays printers of enough processing power on board to process PDF. I would certainly be interested.
(Please don't bother my poor spelling, I'm no native speaker)
I'm working for a large international company (about 9000 employees world wide) which is phasing out Redmond, because it lacks proffesionalism, and they constantly change their own standards.
We use open source, because of it's better (also not perfect) consistancy and much lower maintenance cost. We don't develop software ourselves, we're just users. I must admit, not standard users, all employees are engineers. We are not interested in a shiny glammer interface, the thing just needs to work. Redmond is only compatible with Redmond and nothing else, so we cannot glue applications together. That is the main reason it is phased out.
What makes Redmond so expensive is that with every update something else gets broken. Often, our Sys. Op. thought he had tested the latest patch good enough, rolls it out and "bang" the network goes down again in an area he had overlooked. Due to the lack of good technical documentation, it takes a lot of time to get it up again.
With OSS the technical information is available on the internet and we know much better what each patch does. Moreover, because OSS obeys open standards much better (also not always perfectly), we can glue applicaltions together. Currently we are working with a system that is far more powerful than the shiny Redmond system. And the system downtime is reduced considerably.
While I doubt if that is really going to happen, you are certainly right that the landscape is changing. During the clockfrequency race, the same programs just ran faster. In future speed improvement has to come from parrallelism and specialized execution units, meaning that the code has to be adapted to the hardware much more then in the past.
And this is certainly interersting as now we have to see who is able to utilize the new architecture faster: open source ore closed source? Especially since probably the hardware per user could differe a lot.
However, for desktop usage, it is really the applications that make the value, not the OS.
without people telling us that we're a freak. A fat girl can be very beautiful, but that thought seems to be forbidden in our world. We geeks don't mind to be different, but it's amazing how negative people respond if your girlfriend is fat.
True. But be aware that their Quad opteron boards require a minimum of two processors to run!