Actually, the design of Windows is largely the same as the design of OS X and KDE in this regard. If you remove khtml from KDE, lots of stuff breaks. Sure, you can use Gnome or Busybox or something, but a distro designed around KDE would break if you removed khtml.
Similarly, Lots of stuff in OSX is designed around WebKit. Take away WebKit and tons of stuff breaks.
IE is no more or less "integral to the OS" as either of those. There's nothing magical about it.
Microsoft said that removing IE from Windows 98, as it was designed, would break the OS and make it unusable.
This is totally different from designing a new version of the OS in a different way. It's like taking a car with a central wiring harness, removing that harness and the car breaks compared to designing a new car with a distributed wiring schematic with no central harness. The former doesn't preclude the latter, and designing the new car doesn't make the designers liars for saying that removing the central wiring harness would break the car.
Wherever you heard that XP was a 'rewrite', you heard wrong. Vista isn't going to be a 'rewrite' either. I'm not sure where you get that information, but it's very wrong.
This entire story was about Intel Mac's, and I was responding to someone that was asking why Apple didn't just "leapfrog" Intel and go AMD. My point was, that Yonah currently beats any Turion out there, so it wouldn't be a 'leapfrog", and we're talking about MacOS X on Yonah, not Windows.
As far as I know, Turion isn't dual core. You might be implying that Turion single core would beat out Yonah single core, but that's really beside the point. Apple has to ship a product, and can't wait around for AMD to catch up.
Image manipulation, video, audio, archiving... all rely on large math.
No, not really. These tasks are typically handled by SSE instructions, which haven't change between 32 and 64 bits. Large integer operations are far less common.
You can't find i386 kernel practicaly in any distro.
i386 kernels are still VERY common. I just installed Kubuntu 5.10 and it defaulted to a 386 kernel, for instance.
Tests I pointed you were done on Windows. What are Windows 32-bit optimized for?
I believe that Windows has been optimized for 586 for a number of years. There's not a huge difference between 586 and 686 in terms of performance optimization. Windows is NOT optimized for AMD, though, that's for sure.
The only tests that showed any major improvements were large integer math tests, mostly the PKI stuff, which according to the article was hand optimized for AMD, which is really quite a bit different from general optimizations.
Very, very bad arguments. It almost seems like you lost all your sane arguments
You seem to completley misunderstand. I said nothing about how MUCH memory or disk space. I was talking about memory bandwidth, and the amount of throughput. When you double the data without doubling your throughput, you lose performance. Now, granted, the AMD's integrated memory controller really reduces memory latency, but currently that's limited to DDR400. Intel systems can run at DDR2-667 speeds (though Yonah is significantly below that right now, being 32 bit).
I should have been a bit clearer. Yes, if you're doing large math, like PKI encryption/decryption, then you're going to see very good results with 64 bit, but that's not a common task for most people.
I really doubt you would see that great an improvement in responsiveness between 32 and 64 bit, unless the 32 bit version were optimized for 386 rather than 686 (which is pretty common). If you're going to benchmark, you should at least make sure you're using a kernel that's optimized for the 32 bit version of the CPU you're using.
One reason I doubt the improvement is that you also have a tradeoff of nearly doubling your memory load. 64 bit pointers are all twice as large, and ints will be 64 bit by default, which will mean the average unit of storage is also double.
64 bit doesn't improve perfomance by that much, certainly not the 25% or so that Yonah has over the Turion. At best, you get 3-6% improvement from 64 bit optimized code.
I've never seen any benchmarks for 64 bit code over 32 bit (even under Linux) that are anywhere near the 25% performance gain that Yonah has over Turion.
While it's correct to say that Modern Unix based OS's were designed with networking in mind, so was modern Windows. NT (the basis of 2000 and XP) was written with networking in mind from the get go.
The problem is not the design of the OS, but rather the applications and libraries that sit on top of it. The same goes (largely) for Unix based OS's.
One reason is that there's already a ton of infrastructure in place, and that infrastructure is not designed to handle high speeds, which means replacing it over time. In newly developed countries, high speed infrastructure was put in place initially.
Another reason is that our government incurs a HUGE amount of administrative overhead in telecommunications, with all kinds of rules and regulations, policies, paperwork, requirements, etc... Not that all those are necessarily bad things, especially in a competitive telecom market, but it does add to cost.
That's all well and good, but it doesn't really explain why the Wine team didn't either a) notice the flaw, or b) report it.
If the flaw was obvious, as is claimed, then the Wine team would have noticed the flaw and failed to notify anyone about it. That would be irresponsible.
If the flaw were not obvious, then the Wine team would have dutifully duplicated the functionality without noticing, but that really invalidates the argument that a security audit would have noticed it among the 50+ million lines of code.
Seriously, please tell me how if this flaw was so "obvious", that the Wine team managed to implement it without noticing as well? That's really the major flaw in all these arguments about how "obvious" (in hindsight) it is.
OTOH, the *nix OSs were designed from the start for a network environment with appropriate security.
Why do people like to rewrite history like this? How quickly people forget that Unix didn't even have networking until more than 10 years after it was created. How quickly people forget the era of the Morris worm which resulted because common applications didn't have any regard for security. How quickly people forget that even today, security is still a major problem in plenty of apps on Unix.
Please, do yourself a favor, and realize that security will never mean flaws will not happen. Security is about reducing and finding what you can, and dealing with what slipped through your fingers in a timely manner.
"A decent security audit"? Seriously. Have you ever tried to audit 50 million lines of code? How come the Wine developers didn't catch this? Why did they implement the same flaw? If it was that obvious, that wouldn't have happened.
Dude. Do you HONESTLY think that virtually *ANY* non-trivial piece of software will ever be completely defect free? Ever? Even Donald Knuth paid a good chunk of money in "bounties" on his supposed bug free software, though there hasn't been one found in a while now. And TeX is orders of magnitude less complex than a typical OS distribution, such as Windows, Mac OSX, or any version of Linux.
It's utterly brain dead to "complain" that flaws are found and fixed, regardless of how seriously security is being taken.
You want to know why.NET is growing at a fast pace? I can download VS.NET Express for free, and inside of 10 minutes, have my first web app (hello world, or whatever) running right there on the box, with all of about 10 clicks total from download to running app.
Now, how do you do that with Java? It's a bunch of disparate technogies with all their own quirks that need to be all working in unison with magic incantations to get everything to work. You need to get and configure Apache and Tomcat. You need to install the JVM. You need to download and install Eclipse or something similar. You need to build your skeleton and get it all going. You really have to be an ubergeek to get started in this environment. VS.NET makes it so ANYONE can get started if they want to.
Apple just makes it REALLY easy for users to use iTunes to pay for legitimate music. Or, maybe iPod users are typically less computer savvy and don't know where to download music. Or, maybe iPod users are MORE deceptive, and were less likely to admit to a stranger that they break the law. Or, maybe iPod users were under represented in the survey? Or... any number of other reasons.
Actually, no. The CIA is fobidden to spy on americans, not the NSA. And, while the NSA isn't supposed to spy on americans without court orders, they are certainly allowed to with them.
The fact is, the majority of the people making claims about this don't even understand what it does. The majority of the speculation isn't possible. It doesn't give anyone (Not even Microsoft, much less the NSA) a backdoor into your computer.
Another explanation is that banging out a patch to fix the symptom is faster than fixing the problem. When there's exploit code running in the wild, the former is what happens. When there is no evidence that the vulnerability is being exploited, the latter is what happens.
Stac sued Microsoft for patent infringement and won. Microsoft counter-sued Stac for Copyright infringement, and also won (for integrating it into DOS).
I don't know of a single case where Microsoft has sued anyone initially for patent infringement, though there are several where Microsoft has counter-sued after first being sued. Come to think of it, I don't think those were patent suits either, but copyright infringement (MS v. Stac, for instance)
Seriously, Why would you even care about the difference between the kernel and the platform?
Do you really think your boss will accept the excuse "Oh yeah, our entire servers were compromised by hackers, all our customer data was stolen, and pornography was placed on our home pages, but it's Ok! The vulernabilities weren't in the kernel".
It's splitting hairs to argue that the software that runs on your platform isn't just as important as the OS itself. The vulnerability of your system as a WHOLE is what's important, not whether the flaws are in this or that part of it.
Slashdot Mirror
Actually, the design of Windows is largely the same as the design of OS X and KDE in this regard. If you remove khtml from KDE, lots of stuff breaks. Sure, you can use Gnome or Busybox or something, but a distro designed around KDE would break if you removed khtml.
Similarly, Lots of stuff in OSX is designed around WebKit. Take away WebKit and tons of stuff breaks.
IE is no more or less "integral to the OS" as either of those. There's nothing magical about it.
Microsoft said that removing IE from Windows 98, as it was designed, would break the OS and make it unusable.
This is totally different from designing a new version of the OS in a different way. It's like taking a car with a central wiring harness, removing that harness and the car breaks compared to designing a new car with a distributed wiring schematic with no central harness. The former doesn't preclude the latter, and designing the new car doesn't make the designers liars for saying that removing the central wiring harness would break the car.
Wherever you heard that XP was a 'rewrite', you heard wrong. Vista isn't going to be a 'rewrite' either. I'm not sure where you get that information, but it's very wrong.
This entire story was about Intel Mac's, and I was responding to someone that was asking why Apple didn't just "leapfrog" Intel and go AMD. My point was, that Yonah currently beats any Turion out there, so it wouldn't be a 'leapfrog", and we're talking about MacOS X on Yonah, not Windows.
As far as I know, Turion isn't dual core. You might be implying that Turion single core would beat out Yonah single core, but that's really beside the point. Apple has to ship a product, and can't wait around for AMD to catch up.
Image manipulation, video, audio, archiving... all rely on large math.
No, not really. These tasks are typically handled by SSE instructions, which haven't change between 32 and 64 bits. Large integer operations are far less common.
You can't find i386 kernel practicaly in any distro.
i386 kernels are still VERY common. I just installed Kubuntu 5.10 and it defaulted to a 386 kernel, for instance.
Tests I pointed you were done on Windows. What are Windows 32-bit optimized for?
I believe that Windows has been optimized for 586 for a number of years. There's not a huge difference between 586 and 686 in terms of performance optimization. Windows is NOT optimized for AMD, though, that's for sure.
The only tests that showed any major improvements were large integer math tests, mostly the PKI stuff, which according to the article was hand optimized for AMD, which is really quite a bit different from general optimizations.
Very, very bad arguments. It almost seems like you lost all your sane arguments
You seem to completley misunderstand. I said nothing about how MUCH memory or disk space. I was talking about memory bandwidth, and the amount of throughput. When you double the data without doubling your throughput, you lose performance. Now, granted, the AMD's integrated memory controller really reduces memory latency, but currently that's limited to DDR400. Intel systems can run at DDR2-667 speeds (though Yonah is significantly below that right now, being 32 bit).
I should have been a bit clearer. Yes, if you're doing large math, like PKI encryption/decryption, then you're going to see very good results with 64 bit, but that's not a common task for most people.
I really doubt you would see that great an improvement in responsiveness between 32 and 64 bit, unless the 32 bit version were optimized for 386 rather than 686 (which is pretty common). If you're going to benchmark, you should at least make sure you're using a kernel that's optimized for the 32 bit version of the CPU you're using.
One reason I doubt the improvement is that you also have a tradeoff of nearly doubling your memory load. 64 bit pointers are all twice as large, and ints will be 64 bit by default, which will mean the average unit of storage is also double.
64 bit doesn't improve perfomance by that much, certainly not the 25% or so that Yonah has over the Turion. At best, you get 3-6% improvement from 64 bit optimized code.
I've never seen any benchmarks for 64 bit code over 32 bit (even under Linux) that are anywhere near the 25% performance gain that Yonah has over Turion.
While it's correct to say that Modern Unix based OS's were designed with networking in mind, so was modern Windows. NT (the basis of 2000 and XP) was written with networking in mind from the get go.
The problem is not the design of the OS, but rather the applications and libraries that sit on top of it. The same goes (largely) for Unix based OS's.
One reason is that there's already a ton of infrastructure in place, and that infrastructure is not designed to handle high speeds, which means replacing it over time. In newly developed countries, high speed infrastructure was put in place initially.
Another reason is that our government incurs a HUGE amount of administrative overhead in telecommunications, with all kinds of rules and regulations, policies, paperwork, requirements, etc... Not that all those are necessarily bad things, especially in a competitive telecom market, but it does add to cost.
Actually, the new Yonah processor is significantly faster than the Turion. Check out these benchmarks:
d uo_notebooks_trade_battery_life_for_quicker_respon se/
http://www.tomshardware.com/2006/01/16/will_core_
That's all well and good, but it doesn't really explain why the Wine team didn't either a) notice the flaw, or b) report it.
If the flaw was obvious, as is claimed, then the Wine team would have noticed the flaw and failed to notify anyone about it. That would be irresponsible.
If the flaw were not obvious, then the Wine team would have dutifully duplicated the functionality without noticing, but that really invalidates the argument that a security audit would have noticed it among the 50+ million lines of code.
Seriously, please tell me how if this flaw was so "obvious", that the Wine team managed to implement it without noticing as well? That's really the major flaw in all these arguments about how "obvious" (in hindsight) it is.
OTOH, the *nix OSs were designed from the start for a network environment with appropriate security.
Why do people like to rewrite history like this? How quickly people forget that Unix didn't even have networking until more than 10 years after it was created. How quickly people forget the era of the Morris worm which resulted because common applications didn't have any regard for security. How quickly people forget that even today, security is still a major problem in plenty of apps on Unix.
Please, do yourself a favor, and realize that security will never mean flaws will not happen. Security is about reducing and finding what you can, and dealing with what slipped through your fingers in a timely manner.
"A decent security audit"? Seriously. Have you ever tried to audit 50 million lines of code? How come the Wine developers didn't catch this? Why did they implement the same flaw? If it was that obvious, that wouldn't have happened.
Dude. Do you HONESTLY think that virtually *ANY* non-trivial piece of software will ever be completely defect free? Ever? Even Donald Knuth paid a good chunk of money in "bounties" on his supposed bug free software, though there hasn't been one found in a while now. And TeX is orders of magnitude less complex than a typical OS distribution, such as Windows, Mac OSX, or any version of Linux.
It's utterly brain dead to "complain" that flaws are found and fixed, regardless of how seriously security is being taken.
You're still missing the point. All of those "multitude" of servers are complex to set up in the first place.
You want to know why .NET is growing at a fast pace? I can download VS.NET Express for free, and inside of 10 minutes, have my first web app (hello world, or whatever) running right there on the box, with all of about 10 clicks total from download to running app.
Now, how do you do that with Java? It's a bunch of disparate technogies with all their own quirks that need to be all working in unison with magic incantations to get everything to work. You need to get and configure Apache and Tomcat. You need to install the JVM. You need to download and install Eclipse or something similar. You need to build your skeleton and get it all going. You really have to be an ubergeek to get started in this environment. VS.NET makes it so ANYONE can get started if they want to.
Apple just makes it REALLY easy for users to use iTunes to pay for legitimate music. Or, maybe iPod users are typically less computer savvy and don't know where to download music. Or, maybe iPod users are MORE deceptive, and were less likely to admit to a stranger that they break the law. Or, maybe iPod users were under represented in the survey? Or... any number of other reasons.
Actually, no. The CIA is fobidden to spy on americans, not the NSA. And, while the NSA isn't supposed to spy on americans without court orders, they are certainly allowed to with them.
Actually, Bruce Schneier's analysis is somewhat different.
K eyinMicrosoftCryptoAPI
http://www.schneier.com/crypto-gram-9909.html#NSA
The fact is, the majority of the people making claims about this don't even understand what it does. The majority of the speculation isn't possible. It doesn't give anyone (Not even Microsoft, much less the NSA) a backdoor into your computer.
Another explanation is that banging out a patch to fix the symptom is faster than fixing the problem. When there's exploit code running in the wild, the former is what happens. When there is no evidence that the vulnerability is being exploited, the latter is what happens.
Stac sued Microsoft for patent infringement and won. Microsoft counter-sued Stac for Copyright infringement, and also won (for integrating it into DOS).
I don't know of a single case where Microsoft has sued anyone initially for patent infringement, though there are several where Microsoft has counter-sued after first being sued. Come to think of it, I don't think those were patent suits either, but copyright infringement (MS v. Stac, for instance)
Seriously, Why would you even care about the difference between the kernel and the platform?
Do you really think your boss will accept the excuse "Oh yeah, our entire servers were compromised by hackers, all our customer data was stolen, and pornography was placed on our home pages, but it's Ok! The vulernabilities weren't in the kernel".
It's splitting hairs to argue that the software that runs on your platform isn't just as important as the OS itself. The vulnerability of your system as a WHOLE is what's important, not whether the flaws are in this or that part of it.