In the world of not-so-small servers, more than 4 GByte RAM was necessary years ago. AMD wanted to enter that market with the Opteron and use basically the same architecture for the desktop market, so they needed 64bit because of the servers and used it for marketing on the desktop.
Now I agree they got a bit ahead of the real necesseties there. At the time AMD started promoting 64bit for the desktop, it was about as necessary as a faster CPU for internet surfing (Intels equally stupid marketing campaign a few years ago). But if you wait a few more years, 8 GBytes and more will be typical even for dektop PCs.
There is also the little problem of chip size vs. transistor count. Typical (= affordable) desktop CPUs have a die size (size of the silicon chip) of 100mm to 200mm, making larger chips becomes expensive.
So if you don't want to increase the die size and the prize with it, you have to wait for the next advance in manufacturing technology that allows smaller structures on the chips. Example AMD: They -make dual cores now at 90 nm process technology -are planning to do a quad-core on 65 nm -and maybe will do an 8-core CPU on the next technology (my speculation). But that will take a few more years of waiting.
First, the hardest part is going from 1 to 2 cores. For that, you have to figure out the principle of how to split the workload. Going from 2 cores to n cores will usually be easier. And since dual cores are already becoming mainstream, professional programmers will be forced to take the step from 1 to 2 cores anyway.
Second, the makers of multimedia applications already go ahead with multithreading, because it really works for that type of application. This will drive the market for more cores. In the long run, I expect the mainstream market to settle at the number of cores that works best for multimedia applications and games. I think this will be at least four cores (in that I agree with David Perlmutter) but it may be more, depending on the progress of computer science in parallelization of the above applications. Personally, I would not be surprised to see 16 core CPUs in mainstream computers someday.
For vacuum and zero gravity, you don't need the moon. Build a space station in earth orbit, shipping costs to get your equipment up will be much smaller. Minung on Luna may make sense someday, because lifting stuff out of the moon's gravity is much easier than lifting stuff from earth. But aside from that, I don't see how it would make sense to go all the way to moon.
While I like the idea, any serious colonization will require us to move lots of material and people, not just a handful of astronauts. So the "piles and piles of research" the GP mentioned might be a good idea to invest in transportation first. Start with developing a better shuttle that can fly cheaper, safer and more frequently. That will make it much easier to bring stuff into orbit in the first place. From there, further missions can start as mentioned in the article. Parallel to the development of a better shuttle, some research into lunar mining may be OK, but keep it small until you are ready to start real mass transportation. No prestige projects please that are only meant as part of a pissing contest.
At some point (sorry can't find the link anymore) someone from the open graphics project hinted at "maybe the speed of a Radeon 9600". Which would be an improvement over the chips you mentioned, but not a dramatic one. Of course, there is always the possibility that some other company reconsiders its stance on documentation for their chips. S3 would be a candidate, if they are not blocked from releasing their specs due to third party IP inside. Right now their the marketshare is quite small, and capturing the Linux market with open drivers would make a much greater difference to them than to ATI or NVIDIA.
Getting really offtopic here, but maybe C is not the best language for this particular problem?
Borland Delphi, for instance, offers a compiler switch to activate bounds checking or "range checking" as the Delphi online help calls it. Activating range checking will catch the first of your examples, and there is a convenient checkbox in the project settings to do it. Admittedly, there is no mechanism in Delphi that will catch your second example. But then again, most problems can be solved without pointers. In function calls for example, you can reference original variables in the calling routine(AFAIK the most frequent use of pointers in C) with the keywords VAR and CONST in the parameter list. That will create an implicit call by reference for the variables and the compiler will handle the pointer stuff for you.
And then Delphi is still a relatively low-level language, not that much different from C++. I'm sure others will point out a dozen languages with MUCH better protection against programmer error.
Because dual-core Opterons have been around for a while. The 2xx series will run in systems with two processor sockets, the 8xx in systems up to 8 sockets. Giving you a maximum of 4 or 16 cores. I think the new "4x4" processors will essentially be rebranded Opterons from the 2xx series. So if you really want it and are willing to pay up, you can have a "4x4" AMD system now.
Yep, that sums it up better than grandparent. AMD on 65 nm may be very impressive too, but it seems it will take AMD another 6 months to get there. I hope they don't get hammered too hard in that time.
Once you have Windows + (exclusively) OSS Apps, the step to Linux is not that big. Especially considering that KDE looks a lot like Windows. So I would make that Windows + Apps -> Windows + OSS Apps -> Linux/KDE + OSS Apps
ReactOS may be more interesting for those who have some Windows-only apps they cannot easily replace.
PEBKAC certainly was a factor, and I expect it will remain a factor. Idiots like that russian reactor crew will appear now and then in positions they are not suitable for. For a slightly off-topic example, consider the NASA guys who decided to launch Challenger despite the temperatur being outside spec for the boosters.
So what remains as difference is the inherently safer design of modern reactors. But don't expect the quality of the users to save future reactors.
What really annoying about the whole nuclear fear in the U.S. is that it's really a very green source of energy. You get more radiation released into the atmosphere from a coal-fired plant, not to mention the sludge for your lungs to filter ut of the air. That is only correct as long as there is no major accident. As in Chernobyl. Now the Chernobyl rector was poorly designed from a safety point of view and the crew made stupid experiments, but shit happens. So even with modern designs, I think you have to figure in the "one meltdown per 1000 years" or so in your calculations.
Once the ISP has agreed to deliver a service, he is bound to that contract. IANAL but I think refusing to provide the service would be a valid reason for the customer to cancel the ISP contract immediately and take his business somewhere else. An interesting side note (from Germany, where I live): ISPs frequently offer a nice hardware package (DSL router, often with WLAN) in exchange for a minimum contract duration of 1-2 years. If the provider now breaks the service contract, chances are that you could cancel the ISP contract and keep the goodies. Of course, you'd better NOT do this if you have reason to believe that "they" can prove you have illegaly distributed other peoples' IP.
Now that is weird. Usually, console makers protect their revenue stream by allowing only licensed software to run (for which the game makers pay license fees). If the PS3 comes with Linux pre-installed, this either means that a) the price covers the real costs of the hardware. Which would explain that the PS3 is so expensive. Or b) the pre-installed Linux is seriously crippled somehow, so game makers cannot use it as basis for competitive games without paying license fees to Sony.
The AMD with 4 cores will probably start out being quite expensive, being the new top model. So waiting until January is probably a waste of time. I would, however, wait for the release of the Intel "Conroe" and read some independent reviews on it before deciding on a processor (until now, Intel has pretty much dictated what tests the public gets to see). The AMD X2 has been in the market for a while. You can get independent reviews for that now. At that point, I also expect another round of price cuts, so even if you dislike the Conroe, you will get your AMD X2 cheaper. September/October should be a good time to buy a new PC.
1.5) Working under the "hurry hurry hurry" "boss wants it yesterday" environment means your engineers will cut corners everywhere. When faced with a choice to copy/paste a function in 15 minutes vs. taking some time to refactor and reuse the code in 1 hour, engineers will choose the earlier. In my experience, design debt then accumulates really fast.
A very good observation. Because the 15 minute quickhack can help you to meet your deadline this time. So the engineers will do it the quick way, but somehow the time to re-do it properly is never there. Two or three versions later, you will have to do more changes to the same area of the project. At this point, the lousy design comes around to bite you in the ass.
Only that it does not really work. Overworked programmers are tired programmers who make more mistakes. Long ago, research has shown that the best productivity is reached with a 40 hour week.
A few years ago, I had a combined programmer/support job. One day, a colleague called and said that he had an application on his computer he could not close. So I went over and indeed, one of our programs was in the middle of his screen and did not react to anything. On a hunch, I checked the dektop settings. Lo and behold:
Somehow the guy had made a screenshot while running the application and used that screenshot as Windows wallpaper. Changing the wallpaper got rid of the phantom application;-)
Sure they could work around firewalls that are installed on the Windows computer itself. But eventually someone would get them by monitoring the traffic on his Linux router => back to square one plus accusations for circumventing the Windows firewall application.
5% to 10% who actually switch - I think that would be a realistic number if Microsoft manages to kill all pirated versions. It would also be a nice boost for Linux, compared to the current desktop market share that seems to be maybe 4% - a cursory Google search did not find a current statistic, so I'm going with an extrapolation here.
I agree that it would be a big risk for Microsoft. But never underestimate corporate stupidity. An example from a different business:
At the company I work for, development of a promising new device was canceled two years ago. Because management believed that it would disrupt sales of its predecessor. Since then, the device in question has been mostly sitting in storage while the competition has caught up. Now we will have to catch up because the competition has brought their stuff to market while ours still needs development.
Back to topic: If Microsoft proceeds with a Windows kill switch and nails a few thousand legitimate installations, they would merely match the stupidity of our management;-)
Maybe a community project just needs another form of organization? If interested users form a "Local Infrastructure Corporation", I guess the laws against municipal broadband projects might not apply. Of course, an important difference would be that the "Local Infrastructure Corporation" cannot force anyone to join, so you would have to convince enough interested people to make the undertaking commercially viable.
While I agree the threat with a lawsuit is excessive, "I want to cancel, that's a final decision" should be clear enough to the CSR to make him stop playing games. Personally, I would not threaten him with a lawsuit but with a cancellation letter to the company that includes a complaint about the inept service. The conversation might go like this:
Me: "I want to cancel, that's a final decision" CSR: "You don't want to do this, we have this much better service plan blah blah blah..." Me: "I don't want another service plan, I insist on canceling my account" CSR: "blah blah blah...(still tries avoiding the issue)" Me: "Last time, I want to cancel my account. If you won't do it, I'll send the cancellation as letter to AOL and make it clear that customer service has failed me." CSR: "blah blah blah...(still tries avoiding the issue)" Me: "Goodbye". =>me hangs up and writes nastygram. Preferably as some kind of registered letter where the delivery can be proven. All charges after that point end up as chargebacks.
With LASIK, you have the flap cut and laid to the side, then some tissue is burned off to create the desired change in refraction (what the/. article calls "grinding" and it applies to Lasik too). Afterwards the flap is put back over the treated area. Advantages: -not much pain, -pretty quick results in terms of better vision. Usually even the 1 day postop results look good. Disadvantage: -It takes a long time until the flap is strongly attached to the eye again (1-2 years?). In the meantime, the eye is more vulnerable than usual to accidents.
With PRK or LASEK, only the epithelium (outermost, regenerating layer of the cornea) is removed instead of cutting a flap, then the correction is applied to the cornea surface. After that, it takes a week to heal until the epithelium has regrown Advantages: -Safer because all kinds of flap-related complications are ruled out. Disadvantage: -For the week mentioned above, PRK is said to be a lot more uncomfortable than LASIK. Also, don't expect fully improved vision before the end of the week.
Many, most if you will, are the so called 80% developers, they love adding new stuff and features, but when it comes to making the code clean for others to understand, finish the glitches, work it better into the system, make documentation, they fail. True for some developers. In other cases, you have "80% management" that refuses to invest in cleanup work once the features appear to work. Back to the topic of the article (sort of), this will happen in a typical corporate environment rather than in a community of Open Source volunteers, because there is no management that can threaten them with firing. Which I believe to be the intrinsic advantage of spare-time dveloped Open Source.
Slashdot Mirror
In the world of not-so-small servers, more than 4 GByte RAM was necessary years ago. AMD wanted to enter that market with the Opteron and use basically the same architecture for the desktop market, so they needed 64bit because of the servers and used it for marketing on the desktop.
Now I agree they got a bit ahead of the real necesseties there. At the time AMD started promoting 64bit for the desktop, it was about as necessary as a faster CPU for internet surfing (Intels equally stupid marketing campaign a few years ago).
But if you wait a few more years, 8 GBytes and more will be typical even for dektop PCs.
There is also the little problem of chip size vs. transistor count. Typical (= affordable) desktop CPUs have a die size (size of the silicon chip) of 100mm to 200mm, making larger chips becomes expensive.
So if you don't want to increase the die size and the prize with it, you have to wait for the next advance in manufacturing technology that allows smaller structures on the chips. Example AMD: They
-make dual cores now at 90 nm process technology
-are planning to do a quad-core on 65 nm
-and maybe will do an 8-core CPU on the next technology (my speculation). But that will take a few more years of waiting.
First, the hardest part is going from 1 to 2 cores. For that, you have to figure out the principle of how to split the workload. Going from 2 cores to n cores will usually be easier. And since dual cores are already becoming mainstream, professional programmers will be forced to take the step from 1 to 2 cores anyway.
Second, the makers of multimedia applications already go ahead with multithreading, because it really works for that type of application. This will drive the market for more cores. In the long run, I expect the mainstream market to settle at the number of cores that works best for multimedia applications and games.
I think this will be at least four cores (in that I agree with David Perlmutter) but it may be more, depending on the progress of computer science in parallelization of the above applications. Personally, I would not be surprised to see 16 core CPUs in mainstream computers someday.
For vacuum and zero gravity, you don't need the moon. Build a space station in earth orbit, shipping costs to get your equipment up will be much smaller.
Minung on Luna may make sense someday, because lifting stuff out of the moon's gravity is much easier than lifting stuff from earth. But aside from that, I don't see how it would make sense to go all the way to moon.
While I like the idea, any serious colonization will require us to move lots of material and people, not just a handful of astronauts.
So the "piles and piles of research" the GP mentioned might be a good idea to invest in transportation first. Start with developing a better shuttle that can fly cheaper, safer and more frequently. That will make it much easier to bring stuff into orbit in the first place. From there, further missions can start as mentioned in the article.
Parallel to the development of a better shuttle, some research into lunar mining may be OK, but keep it small until you are ready to start real mass transportation. No prestige projects please that are only meant as part of a pissing contest.
At some point (sorry can't find the link anymore) someone from the open graphics project hinted at "maybe the speed of a Radeon 9600". Which would be an improvement over the chips you mentioned, but not a dramatic one.
Of course, there is always the possibility that some other company reconsiders its stance on documentation for their chips.
S3 would be a candidate, if they are not blocked from releasing their specs due to third party IP inside. Right now their the marketshare is quite small, and capturing the Linux market with open drivers would make a much greater difference to them than to ATI or NVIDIA.
Getting really offtopic here, but maybe C is not the best language for this particular problem?
Borland Delphi, for instance, offers a compiler switch to activate bounds checking or "range checking" as the Delphi online help calls it. Activating range checking will catch the first of your examples, and there is a convenient checkbox in the project settings to do it.
Admittedly, there is no mechanism in Delphi that will catch your second example. But then again, most problems can be solved without pointers. In function calls for example, you can reference original variables in the calling routine(AFAIK the most frequent use of pointers in C) with the keywords VAR and CONST in the parameter list. That will create an implicit call by reference for the variables and the compiler will handle the pointer stuff for you.
And then Delphi is still a relatively low-level language, not that much different from C++. I'm sure others will point out a dozen languages with MUCH better protection against programmer error.
Because dual-core Opterons have been around for a while. The 2xx series will run in systems with two processor sockets, the 8xx in systems up to 8 sockets. Giving you a maximum of 4 or 16 cores.
I think the new "4x4" processors will essentially be rebranded Opterons from the 2xx series. So if you really want it and are willing to pay up, you can have a "4x4" AMD system now.
Yep, that sums it up better than grandparent.
AMD on 65 nm may be very impressive too, but it seems it will take AMD another 6 months to get there. I hope they don't get hammered too hard in that time.
Once you have Windows + (exclusively) OSS Apps, the step to Linux is not that big. Especially considering that KDE looks a lot like Windows.
So I would make that Windows + Apps -> Windows + OSS Apps -> Linux/KDE + OSS Apps
ReactOS may be more interesting for those who have some Windows-only apps they cannot easily replace.
PEBKAC certainly was a factor, and I expect it will remain a factor. Idiots like that russian reactor crew will appear now and then in positions they are not suitable for. For a slightly off-topic example, consider the NASA guys who decided to launch Challenger despite the temperatur being outside spec for the boosters.
So what remains as difference is the inherently safer design of modern reactors. But don't expect the quality of the users to save future reactors.
What really annoying about the whole nuclear fear in the U.S. is that it's really a very green source of energy. You get more radiation released into the atmosphere from a coal-fired plant, not to mention the sludge for your lungs to filter ut of the air.
That is only correct as long as there is no major accident. As in Chernobyl. Now the Chernobyl rector was poorly designed from a safety point of view and the crew made stupid experiments, but shit happens.
So even with modern designs, I think you have to figure in the "one meltdown per 1000 years" or so in your calculations.
Once the ISP has agreed to deliver a service, he is bound to that contract. IANAL but I think refusing to provide the service would be a valid reason for the customer to cancel the ISP contract immediately and take his business somewhere else.
An interesting side note (from Germany, where I live):
ISPs frequently offer a nice hardware package (DSL router, often with WLAN) in exchange for a minimum contract duration of 1-2 years. If the provider now breaks the service contract, chances are that you could cancel the ISP contract and keep the goodies. Of course, you'd better NOT do this if you have reason to believe that "they" can prove you have illegaly distributed other peoples' IP.
Now that is weird.
Usually, console makers protect their revenue stream by allowing only licensed software to run (for which the game makers pay license fees). If the PS3 comes with Linux pre-installed, this either means that
a) the price covers the real costs of the hardware. Which would explain that the PS3 is so expensive. Or
b) the pre-installed Linux is seriously crippled somehow, so game makers cannot use it as basis for competitive games without paying license fees to Sony.
The AMD with 4 cores will probably start out being quite expensive, being the new top model. So waiting until January is probably a waste of time.
I would, however, wait for the release of the Intel "Conroe" and read some independent reviews on it before deciding on a processor (until now, Intel has pretty much dictated what tests the public gets to see). The AMD X2 has been in the market for a while. You can get independent reviews for that now.
At that point, I also expect another round of price cuts, so even if you dislike the Conroe, you will get your AMD X2 cheaper. September/October should be a good time to buy a new PC.
1.5) Working under the "hurry hurry hurry" "boss wants it yesterday" environment means your engineers will cut corners everywhere. When faced with a choice to copy/paste a function in 15 minutes vs. taking some time to refactor and reuse the code in 1 hour, engineers will choose the earlier. In my experience, design debt then accumulates really fast.
A very good observation. Because the 15 minute quickhack can help you to meet your deadline this time. So the engineers will do it the quick way, but somehow the time to re-do it properly is never there.
Two or three versions later, you will have to do more changes to the same area of the project. At this point, the lousy design comes around to bite you in the ass.
Only that it does not really work. Overworked programmers are tired programmers who make more mistakes. Long ago, research has shown that the best productivity is reached with a 40 hour week.
For details, see http://www.igda.org/articles/erobinson_crunch.php
A few years ago, I had a combined programmer/support job. One day, a colleague called and said that he had an application on his computer he could not close.
;-)
So I went over and indeed, one of our programs was in the middle of his screen and did not react to anything. On a hunch, I checked the dektop settings. Lo and behold:
Somehow the guy had made a screenshot while running the application and used that screenshot as Windows wallpaper. Changing the wallpaper got rid of the phantom application
Sure they could work around firewalls that are installed on the Windows computer itself. But eventually someone would get them by monitoring the traffic on his Linux router =>
back to square one plus accusations for circumventing the Windows firewall application.
5% to 10% who actually switch - I think that would be a realistic number if Microsoft manages to kill all pirated versions.
It would also be a nice boost for Linux, compared to the current desktop market share that seems to be maybe 4% - a cursory Google search did not find a current statistic, so I'm going with an extrapolation here.
I agree that it would be a big risk for Microsoft. But never underestimate corporate stupidity. An example from a different business:
;-)
At the company I work for, development of a promising new device was canceled two years ago. Because management believed that it would disrupt sales of its predecessor. Since then, the device in question has been mostly sitting in storage while the competition has caught up. Now we will have to catch up because the competition has brought their stuff to market while ours still needs development.
Back to topic:
If Microsoft proceeds with a Windows kill switch and nails a few thousand legitimate installations, they would merely match the stupidity of our management
Maybe a community project just needs another form of organization? If interested users form a "Local Infrastructure Corporation", I guess the laws against municipal broadband projects might not apply.
Of course, an important difference would be that the "Local Infrastructure Corporation" cannot force anyone to join, so you would have to convince enough interested people to make the undertaking commercially viable.
While I agree the threat with a lawsuit is excessive, "I want to cancel, that's a final decision" should be clear enough to the CSR to make him stop playing games.
Personally, I would not threaten him with a lawsuit but with a cancellation letter to the company that includes a complaint about the inept service. The conversation might go like this:
Me: "I want to cancel, that's a final decision"
CSR: "You don't want to do this, we have this much better service plan blah blah blah..."
Me: "I don't want another service plan, I insist on canceling my account"
CSR: "blah blah blah...(still tries avoiding the issue)"
Me: "Last time, I want to cancel my account. If you won't do it, I'll send the cancellation as letter to AOL and make it clear that customer service has failed me."
CSR: "blah blah blah...(still tries avoiding the issue)"
Me: "Goodbye".
=>me hangs up and writes nastygram. Preferably as some kind of registered letter where the delivery can be proven. All charges after that point end up as chargebacks.
The summary is somewhat misleading.
/. article calls "grinding" and it applies to Lasik too). Afterwards the flap is put back over the treated area.
With LASIK, you have the flap cut and laid to the side, then some tissue is burned off to create the desired change in refraction (what the
Advantages:
-not much pain,
-pretty quick results in terms of better vision. Usually even the 1 day postop results look good.
Disadvantage:
-It takes a long time until the flap is strongly attached to the eye again (1-2 years?). In the meantime, the eye is more vulnerable than usual to accidents.
With PRK or LASEK, only the epithelium (outermost, regenerating layer of the cornea) is removed instead of cutting a flap, then the correction is applied to the cornea surface. After that, it takes a week to heal until the epithelium has regrown
Advantages:
-Safer because all kinds of flap-related complications are ruled out.
Disadvantage:
-For the week mentioned above, PRK is said to be a lot more uncomfortable than LASIK. Also, don't expect fully improved vision before the end of the week.
Many, most if you will, are the so called 80% developers, they love adding new stuff and features, but when it comes to making the code clean for others to understand, finish the glitches, work it better into the system, make documentation, they fail.
True for some developers. In other cases, you have "80% management" that refuses to invest in cleanup work once the features appear to work.
Back to the topic of the article (sort of), this will happen in a typical corporate environment rather than in a community of Open Source volunteers, because there is no management that can threaten them with firing. Which I believe to be the intrinsic advantage of spare-time dveloped Open Source.