Why Java? Because I can concentrate more on the programming logic side rather than worry so much about memory resources, it has a lot of library support, and it runs on most real operating systems as well as those marketed by Microsoft.
In your case it probably is not valid, but the problem of new SW is that they seem to forget everything about resources. This is especially true to those MS IDE/tools users that have never seen/studied assembly. And with java and C# (never used it actually), it seems that pointer operations (which is the hardest for new programmers) seem to disappear.
Is it a T61? Anyway, the idea is that a card reader is very (very very) cheap to manufacture for USB. So even if it is a card-express or pcmcia card reader, it's actualy an USB card reader chip, and a USB host controller on PCI (pcmcia) or PCIe (cardexpress). And as I remember, card-express can use either PCIe or USB. So in the latter case the USB host controller is the one in your laptop chipset. So I really doubt you have your SD card reader on it's own bus. Maybe on it's own USB bus. But I'm 99% sure it is USB. Any other connection is 80-90% more expensive because the design (R&D) costs are spread on a much smaller number of chips.
Please stop comparing THEORETICAL speeds. A car needs 400+HP to reach 300km/h, but a bike can do it with 150HP. But in which one would you be brave to try? A HDD speed is given by platter rotation and density. The only consumer HDD's I can think of reaching SATA150 speeds are the WD raptors, but even those I remember going to 120MB/s. You can get 300MB/s from a SATA300 HDD only by reading/writing exclusively to it's cache, but this data would be cached in RAM already by the OS. The fact is that only this year the USB sticks (and cards) started sutaining more than 10MB/s. So the point is that flash vs. HDD comparison is very dependent on usage. On random reads, flash clearly wins, but on writes it can hold a small lead only on random data, preferably in the 4-16kb range. On sequential transfers HDDs are still leaders and it has nothing to do with the interface. Also, USB has a high CPU usage because of the multi-device architecture and the protocol to the USB host controller.
Actually I would call Virtual Memory whatever the application(s) can allocate. So once you run out of virtual memory, you used up RAM and paging file. But since you will not use all the pages at the same time, the OS will save the least used ones in swap and restore the needed ones. So (Virtual memory)=(physical memory)+(sum of page files). Since the DOS extender time, this was the term usage. Virtual address space is the memory 1 application sees. Although the application knows that it's continuous and not shared with other applications (unless specifically requested), the reality is that it can be spread between physical memory and swap. But this is how the OS+CPU present it to the app.
I am not an expert at systems, but seems to me that he is saying that Windows tries to write disk buffers ("file cache"?) to the pagefile in order to make more space.
Well, you did get this idea wrong. Disk cache is treated differently and IS NEVER swapped out (MS is not THAT stupid). However, the kernel may swap an application which you just minimized in order to increase the cache size. My main frustration with NT-based kernels is that you cannot limit the cache size. At my work a virus scanner runs weekly. Working during scanning is murder, because windows grows the cache size to the whole physical memory and swaps out everything, except maybe the current application, but I always use 3-4 applications which require real-time response. I heard they changed this behavior in Vista but I have not tested it. At least in Win98 I could (and always did) limit the cache to 1/4 of physical RAM and it worked perfectly. As a comparison, in linux you barely get any swapping to disk if you use up to 80% (estimation, no hard evidence) of your physical RAM. I always have 784MB of swap partition regardless of RAM (512MB-4GB). The swap size is because of partition alignment. I always align to multiple of 1000 cyclinders, but I give 100 cylinders from the linux partition to swap space. This is because if a HDD failiure occurs (SW of HW), I know where to look for partitions.
Their goal is not to stop piracy, but to control it. They know this, but will never admit it. I started with Windows 3.1 in 6th grade and the 1st OS I bought was XP64. But I had many MS OSes installed: Win9x/ME (all editions), NT4, 2000, XP Vista. After trying Vista and end of production for XP arriving made me buy XP64 so I could still catch it (me=Vista hater). The point of this explanation is that I gained a lot of SW/HW knowledge and there are few that can solve an issue which I'm unable. If I would have been limited to an OS bough by parents, I'm sure I would not have been in the IT business. Also, why do you think there are so many 3rd country people (majority from India; if it's not 3rd country, please don't kill me) that work in the US in IT? Make them take a polygraph test and I'm sure none of them started with legal SW. And let's not forget that piracy numbers/losses are IMPOSSIBLE to obtain. And if anti-piracy measures would be effective, not many would actually buy the product (they would not use it anymore). Come to think of it, yes, apply anti-piracy measures so linux could spread more and improve faster and game companies to start an interest in it. Because gaming is 80% of why I still use Windows (and 15% a good file manager like total commander). My router is gentoo, my laptop is ubuntu (although I do have the OEM XP on it I barely use it).
Your logic needs more information: when a game/publisher has low revenues, it's investors will retreat, meaning more money lost (to the company). That in turn means people have to settle for lower pays or numbers. That is indirectly related to quality of work. So piracy of a game does not affect that game (maybe it's support), but it does affect it's sequel (or other game on which work is done). And that is the biggest problem. So a low-sale game will most likely affect badly the quality of the next game. This is why in many.nfo files it's written "support the developers, buy the game".
That could explain the localized versions, but the English versions cost the same (as the localized ones). Actually, I saw in my country (Romania) on some websites the same version (OEM) was almost 50% of the price of the English version (I, personally, am against localized versions, but that's because I grew up with english).
I don't try to be negativist, but you WILL get lost is someone does not respect the else branch ident with it's coresponding if in case you have many if/else conditions. One ident is wrong, and who knows what you read. I'm always for open-after-statement and close on ident, with only "else" being accepted after a closing. A brace opening on new line seems to take too much vertical space, especially since the desired transition of the manufacturers to widescreens.
I don't understand it either. What you described is done something like this:
-a+b+c+d:
1.copy a,b,c,d to a new string
2.print copies everything to buffer while looking for "%".
-%s%s%s%s:
1.copy a,b,c,d to a new list (same as previous 1)
2.interpret %s%s%s%s - found "%", followed by "s"
3.copy element from list to buffer
4.continue with 2 until string is complete
now, memory allocation is not an issue, because these are all the variables you use. But the print parsing a+b+c+d is faster compared to "%s%s%s%s", because when "%" is found, more checks are done (i.e. initialize all possible parameters because you still don't know if it's a string, number, pointer address, etc., either of which could be limited). So even if "%s%s%s%s" is only 9 characters (including \0), it can be costlier than a+b+c+d, which is 13 characters.
Remember that the "printf"-type functions are very general.
I'm also used to green on black. And green is better than yellow or white on CRT monitors which have convergence problems, because you don't have red and blue that need to converge to green. It's probably better even on LCD monitors when you need small fonts. Also, out of the three primary colors, green appears the brightest (human eye perceiving).
Anyway, I seem to be very confortable with black on white used by web browsers if no convergence problems exist (no old CRT).
If you do the comparison, do it right: With your ISP you have a direct medium (usually cable) capable of high-speeds (in this case, even 1mbps is high speed). And data overhead is less than 50% (IP header compared to 140 characters of data) on a pre-established link. With your cell, you have a shared medium (air) with a limited number of frequency bands. The overhead is not only the extra data transfered, but also (like a phone conversation) it has a separate line negociated to transfer. If you would have smaller prices on SMS (let's say 10 times smaller), more and more users would use it. This would increase the providers load, and even if they could handle it, some cells could be limited by their bandwidth which is regulated by the FCC. This would increase the transmission times and even affect regular communications, which means more angry calls to tech support.
So providers probably justify it as a "crowd control" (something like use it only if you really have to).
Im Romania at least one of the ISPs had a 1st 3 seconds not charged. Needless to say, the consumers started making 1-word calls (call, say 1 word and hang up, then do the same for each other word). I've heard about 1000-page detailed phone bills which were less than 10$. After the 1st year, they cancelled it on ALL contracts, not just the new ones. I don't have to say how it was during phone "rush hour" when you wanted to make a regular call.
I think it's the "paperless" idea/solution (maybe a push towards paperless office for next generation). The less paper is used, the more trees are left standing. Although I think the wood burning still uses more wood than papers. Oh wait, in ecuatorian countries, they do not need to generate heat (at least much less than northern/southern countries).
Not true...it has spares and exchanges those with existing blocks ONLY when they are written. But if you never write to "LBA block x", it will never participate in wear leveling. That's because it has to retain the data in the physical sector, and that's done by copying, which slow down the drive. While it is possible to do it, it's not feasible for consumer products as it's too complex. I put a post with a link to Corsairs FAQ.
All these cycles are given as one cell write cycles. But flash has more than the advertised cells (spare cells). And these are written every time a block is wrote. So a newly written spare block becomes the new block while the old block is moved to the spare blocks. So in theory, you can exceed those write cycles, even by writing data to the same sector. After all, that's why it's called "wear leveling". I think DVD-RAM (and DVD+-RW, but in SW if used in packet writing mode) uses the same algorithm. My only concern is how/where the mapping is stored. Besides, I would be satisfied if I would get 3-5 years with no errors. I had to deal with corrupted cards, but usually it was with low-cost cards used in low-cost cameras. And I also would like to try an XP installation on my 4GB Corsair Voyaget GT.
Actually, I don't think the jail will make a person "better". Remember, he's surrounded only by "his kind". The only rehabilitation is for him to realize that there is a penalty ("beaten by te others" or "wasted time in jail") and be afraid of that penalty. I think the point to jail (historic times) was just to put away the bad people and give examples to the others. The "adaptation" tried in these times is kind of flawed, because some people (parole officers or what they're called) take a chance and listen to the lies (or truths) of some and let them out of prison. Sure, some of them really stay away from trouble, but I think the jailtime si actually worsening them.
You want to switch Windows users to Linux. Guess what: they ALREADY have data on NTFS. So the windows ext2 driver is low priority. Also, there was a comment here about the linux distribuitors being hesitant to producing windows code for users that already use linux.
"Microsoft's defragmentation tools are the some of the best at handling defragmentation."
It's not even microsoft's tool. What comes in XP is Dikeeper "super-light" version. Try the full version. And ever since Win2k, I cannot do a fast defragging on any FS, especially in FAT. I miss the old day with Norton Speedisk under Windows98. That was the best I ever saw. Until Speeddisk for NT (2k actually) became dependant on MS's implementation of file move. It seems to me that they read-write-read-verify on FAT32, because the same file is moved 3 times slower on FAT32 than NTFS. It's a conspiracy to force you to NTFS (along with not FAT32 format option on 30Gb+ partitions).
This has nothing to do with the filesystem. I bet you do administration for "normal" users, which install any "greeting card" their friend sends. Also, do you reformat and install Windows from scratch? My idea is that any filesystem put to such a workload would be slowed down. And also the ext2/3fs is badly interpreted as non-fragmenting. It is "designed to reduce fragmenting", but there are situation it cannot prevent it, especially if the drive is over 80% full. But the fragmenting is not the big issue, it's how consecutive fragments are read that is the issue. If the metadata and fragments are near, you will not notice big delays from HDD, and this is the advantage of extfs over FAT. I really don't know how the file chain is kept in NTFS, so I cannot make a valid comparison. But do not blame the FS for bad user workload.
"is it true" is not the problem, because I am sure OS has patent violations. But the problem is how many can be solved/worked around. I think MS plays for quantity, not quality. It does not target the IT specialists which can/know how to verify the claims, it targets the management (who controls the money) that does not know details. They can obtain more money by saying "your software has 230 patent violations" instead of "your software has 30 violations". When a manager hears of 30, he may delegate some employees/company to check how important those 30 are for them and maybe settle for much less, but 230 could be a scary number for smaller companies and they may just settle for what MS wants, because it may be costlier to investigate those 230 violations than to pay MS royalties. They want to milk the end-users of anything they can, even if it's their competitors work, because they cannot touch the competitor. Or maybe they can with FUD?
And related to GPL3, if MS would put such restrictions then customers would complain a lot, and either GPL3 would not be used or MS would loose it's customers. My company (10k+ employees) uses perl in 90% of Windows scripts. Think about what would happen if perl would switch to GPL3 and MS would impose these restrictions.
I think they could be sued for slander. They say of violations but not disclose the entire truth, thus hurting the open-source reputation (trademark???). If some newspaper would say that a politician cheated on his wife without actually showing any pictures (proof), they would be sued for slander. So why should this not work for OpenSource? Money for lawyers seems to be the only blocker for me.
It's pretty easy to configure an X server to distinguish between a USB mouse+keyboard, and PS2 mouse+keyboard and create 2 sessions. I think you can even do it using 1 Nvidia (or ATI) card with 2 outputs. Since USB is dinamicaly allocating resources (IDs), I don't think you can put the same type of devices on one computer and expect to be mapped always to the same session, so going with a different configuration like I said before is risky.
I forgot to say that aligning is not an issue with modern processors because of big caches. It counted for older processors because they accessed the memory directly, so they needed 16/32bit aligning. But today, the cache reads like 8-bytes (aligned) at a time, and from L1 cache you can retrieve any byte/word, with no worries about alignment. Where alignment counts is big structures+arrays, where the compiler can do nothing and you have to optimize your code/design yourself.
Speedups from switching all come from compiler. Even 32-bit Intel (I don't know about AMD) use register-renaming (and support many registers, not just what compiler assigns), but they are used only for out-of-order processing. It still uses cpu-memory load/stores, which are slower even if it's in L1 cache. With x86-64, the compiler can optimize the code by using more (and bigger) registers, so cache (or worst, memory) access is reduced. So, like from the 386/486, the compiler is a major speed factor.
Of course it can. It's a direct consequence of the deals Microsoft makes with vendors. If the PC manufacturers want discounts on Windows - which they need in order to keep their prices down to match the competition - they have to pay for it on every computer they ship. But they do have the choice...they could ally with other vendors and oppose MS...so the vendors can be considered responsible because they enforce MS to offer those deals. So I still put at least 1% of the blame on vendors.
PS: this is exactly as the Intel deals. But that is starting to slip. So if Apple would change it's marketing, in a few years we could see the exact thing in OSes: Apple sueing MS for anti-competitive practices. Linux/Unix does not have financial support to do this.
Slashdot Mirror
Why Java? Because I can concentrate more on the programming logic side rather than worry so much about memory resources, it has a lot of library support, and it runs on most real operating systems as well as those marketed by Microsoft.
In your case it probably is not valid, but the problem of new SW is that they seem to forget everything about resources. This is especially true to those MS IDE/tools users that have never seen/studied assembly. And with java and C# (never used it actually), it seems that pointer operations (which is the hardest for new programmers) seem to disappear.
Is it a T61?
Anyway, the idea is that a card reader is very (very very) cheap to manufacture for USB. So even if it is a card-express or pcmcia card reader, it's actualy an USB card reader chip, and a USB host controller on PCI (pcmcia) or PCIe (cardexpress). And as I remember, card-express can use either PCIe or USB. So in the latter case the USB host controller is the one in your laptop chipset.
So I really doubt you have your SD card reader on it's own bus. Maybe on it's own USB bus. But I'm 99% sure it is USB. Any other connection is 80-90% more expensive because the design (R&D) costs are spread on a much smaller number of chips.
Please stop comparing THEORETICAL speeds. A car needs 400+HP to reach 300km/h, but a bike can do it with 150HP. But in which one would you be brave to try?
A HDD speed is given by platter rotation and density. The only consumer HDD's I can think of reaching SATA150 speeds are the WD raptors, but even those I remember going to 120MB/s. You can get 300MB/s from a SATA300 HDD only by reading/writing exclusively to it's cache, but this data would be cached in RAM already by the OS.
The fact is that only this year the USB sticks (and cards) started sutaining more than 10MB/s.
So the point is that flash vs. HDD comparison is very dependent on usage. On random reads, flash clearly wins, but on writes it can hold a small lead only on random data, preferably in the 4-16kb range. On sequential transfers HDDs are still leaders and it has nothing to do with the interface.
Also, USB has a high CPU usage because of the multi-device architecture and the protocol to the USB host controller.
Actually I would call Virtual Memory whatever the application(s) can allocate. So once you run out of virtual memory, you used up RAM and paging file. But since you will not use all the pages at the same time, the OS will save the least used ones in swap and restore the needed ones.
So (Virtual memory)=(physical memory)+(sum of page files).
Since the DOS extender time, this was the term usage. Virtual address space is the memory 1 application sees. Although the application knows that it's continuous and not shared with other applications (unless specifically requested), the reality is that it can be spread between physical memory and swap. But this is how the OS+CPU present it to the app.
I am not an expert at systems, but seems to me that he is saying that Windows tries to write disk buffers ("file cache"?) to the pagefile in order to make more space.
Well, you did get this idea wrong. Disk cache is treated differently and IS NEVER swapped out (MS is not THAT stupid). However, the kernel may swap an application which you just minimized in order to increase the cache size.
My main frustration with NT-based kernels is that you cannot limit the cache size. At my work a virus scanner runs weekly. Working during scanning is murder, because windows grows the cache size to the whole physical memory and swaps out everything, except maybe the current application, but I always use 3-4 applications which require real-time response. I heard they changed this behavior in Vista but I have not tested it. At least in Win98 I could (and always did) limit the cache to 1/4 of physical RAM and it worked perfectly.
As a comparison, in linux you barely get any swapping to disk if you use up to 80% (estimation, no hard evidence) of your physical RAM. I always have 784MB of swap partition regardless of RAM (512MB-4GB). The swap size is because of partition alignment. I always align to multiple of 1000 cyclinders, but I give 100 cylinders from the linux partition to swap space. This is because if a HDD failiure occurs (SW of HW), I know where to look for partitions.
Their goal is not to stop piracy, but to control it. They know this, but will never admit it. I started with Windows 3.1 in 6th grade and the 1st OS I bought was XP64. But I had many MS OSes installed: Win9x/ME (all editions), NT4, 2000, XP Vista. After trying Vista and end of production for XP arriving made me buy XP64 so I could still catch it (me=Vista hater).
The point of this explanation is that I gained a lot of SW/HW knowledge and there are few that can solve an issue which I'm unable. If I would have been limited to an OS bough by parents, I'm sure I would not have been in the IT business.
Also, why do you think there are so many 3rd country people (majority from India; if it's not 3rd country, please don't kill me) that work in the US in IT? Make them take a polygraph test and I'm sure none of them started with legal SW.
And let's not forget that piracy numbers/losses are IMPOSSIBLE to obtain. And if anti-piracy measures would be effective, not many would actually buy the product (they would not use it anymore). Come to think of it, yes, apply anti-piracy measures so linux could spread more and improve faster and game companies to start an interest in it. Because gaming is 80% of why I still use Windows (and 15% a good file manager like total commander). My router is gentoo, my laptop is ubuntu (although I do have the OEM XP on it I barely use it).
Your logic needs more information: .nfo files it's written "support the developers, buy the game".
when a game/publisher has low revenues, it's investors will retreat, meaning more money lost (to the company). That in turn means people have to settle for lower pays or numbers. That is indirectly related to quality of work.
So piracy of a game does not affect that game (maybe it's support), but it does affect it's sequel (or other game on which work is done).
And that is the biggest problem. So a low-sale game will most likely affect badly the quality of the next game.
This is why in many
That could explain the localized versions, but the English versions cost the same (as the localized ones). Actually, I saw in my country (Romania) on some websites the same version (OEM) was almost 50% of the price of the English version (I, personally, am against localized versions, but that's because I grew up with english).
I don't try to be negativist, but you WILL get lost is someone does not respect the else branch ident with it's coresponding if in case you have many if/else conditions. One ident is wrong, and who knows what you read. I'm always for open-after-statement and close on ident, with only "else" being accepted after a closing. A brace opening on new line seems to take too much vertical space, especially since the desired transition of the manufacturers to widescreens.
-a+b+c+d:
1.copy a,b,c,d to a new string
2.print copies everything to buffer while looking for "%".
-%s%s%s%s:
1.copy a,b,c,d to a new list (same as previous 1)
2.interpret %s%s%s%s - found "%", followed by "s"
3.copy element from list to buffer
4.continue with 2 until string is complete
now, memory allocation is not an issue, because these are all the variables you use. But the print parsing a+b+c+d is faster compared to "%s%s%s%s", because when "%" is found, more checks are done (i.e. initialize all possible parameters because you still don't know if it's a string, number, pointer address, etc., either of which could be limited). So even if "%s%s%s%s" is only 9 characters (including \0), it can be costlier than a+b+c+d, which is 13 characters. Remember that the "printf"-type functions are very general.
I'm also used to green on black. And green is better than yellow or white on CRT monitors which have convergence problems, because you don't have red and blue that need to converge to green. It's probably better even on LCD monitors when you need small fonts. Also, out of the three primary colors, green appears the brightest (human eye perceiving).
Anyway, I seem to be very confortable with black on white used by web browsers if no convergence problems exist (no old CRT).
If you do the comparison, do it right:
With your ISP you have a direct medium (usually cable) capable of high-speeds (in this case, even 1mbps is high speed). And data overhead is less than 50% (IP header compared to 140 characters of data) on a pre-established link.
With your cell, you have a shared medium (air) with a limited number of frequency bands. The overhead is not only the extra data transfered, but also (like a phone conversation) it has a separate line negociated to transfer.
If you would have smaller prices on SMS (let's say 10 times smaller), more and more users would use it. This would increase the providers load, and even if they could handle it, some cells could be limited by their bandwidth which is regulated by the FCC. This would increase the transmission times and even affect regular communications, which means more angry calls to tech support.
So providers probably justify it as a "crowd control" (something like use it only if you really have to).
Im Romania at least one of the ISPs had a 1st 3 seconds not charged. Needless to say, the consumers started making 1-word calls (call, say 1 word and hang up, then do the same for each other word). I've heard about 1000-page detailed phone bills which were less than 10$. After the 1st year, they cancelled it on ALL contracts, not just the new ones. I don't have to say how it was during phone "rush hour" when you wanted to make a regular call.
I think it's the "paperless" idea/solution (maybe a push towards paperless office for next generation). The less paper is used, the more trees are left standing. Although I think the wood burning still uses more wood than papers. Oh wait, in ecuatorian countries, they do not need to generate heat (at least much less than northern/southern countries).
Not true...it has spares and exchanges those with existing blocks ONLY when they are written. But if you never write to "LBA block x", it will never participate in wear leveling. That's because it has to retain the data in the physical sector, and that's done by copying, which slow down the drive. While it is possible to do it, it's not feasible for consumer products as it's too complex. I put a post with a link to Corsairs FAQ.
All these cycles are given as one cell write cycles. But flash has more than the advertised cells (spare cells). And these are written every time a block is wrote. So a newly written spare block becomes the new block while the old block is moved to the spare blocks. So in theory, you can exceed those write cycles, even by writing data to the same sector. After all, that's why it's called "wear leveling". I think DVD-RAM (and DVD+-RW, but in SW if used in packet writing mode) uses the same algorithm.
My only concern is how/where the mapping is stored.
Besides, I would be satisfied if I would get 3-5 years with no errors. I had to deal with corrupted cards, but usually it was with low-cost cards used in low-cost cameras. And I also would like to try an XP installation on my 4GB Corsair Voyaget GT.
PS: wear leveling explained by Corsair (pdf)
Actually, I don't think the jail will make a person "better". Remember, he's surrounded only by "his kind". The only rehabilitation is for him to realize that there is a penalty ("beaten by te others" or "wasted time in jail") and be afraid of that penalty. I think the point to jail (historic times) was just to put away the bad people and give examples to the others. The "adaptation" tried in these times is kind of flawed, because some people (parole officers or what they're called) take a chance and listen to the lies (or truths) of some and let them out of prison. Sure, some of them really stay away from trouble, but I think the jailtime si actually worsening them.
You want to switch Windows users to Linux. Guess what: they ALREADY have data on NTFS. So the windows ext2 driver is low priority.
Also, there was a comment here about the linux distribuitors being hesitant to producing windows code for users that already use linux.
"Microsoft's defragmentation tools are the some of the best at handling defragmentation."
It's not even microsoft's tool. What comes in XP is Dikeeper "super-light" version. Try the full version. And ever since Win2k, I cannot do a fast defragging on any FS, especially in FAT. I miss the old day with Norton Speedisk under Windows98. That was the best I ever saw. Until Speeddisk for NT (2k actually) became dependant on MS's implementation of file move. It seems to me that they read-write-read-verify on FAT32, because the same file is moved 3 times slower on FAT32 than NTFS. It's a conspiracy to force you to NTFS (along with not FAT32 format option on 30Gb+ partitions).
This has nothing to do with the filesystem. I bet you do administration for "normal" users, which install any "greeting card" their friend sends. Also, do you reformat and install Windows from scratch?
My idea is that any filesystem put to such a workload would be slowed down. And also the ext2/3fs is badly interpreted as non-fragmenting. It is "designed to reduce fragmenting", but there are situation it cannot prevent it, especially if the drive is over 80% full.
But the fragmenting is not the big issue, it's how consecutive fragments are read that is the issue. If the metadata and fragments are near, you will not notice big delays from HDD, and this is the advantage of extfs over FAT. I really don't know how the file chain is kept in NTFS, so I cannot make a valid comparison.
But do not blame the FS for bad user workload.
"is it true" is not the problem, because I am sure OS has patent violations. But the problem is how many can be solved/worked around. I think MS plays for quantity, not quality. It does not target the IT specialists which can/know how to verify the claims, it targets the management (who controls the money) that does not know details. They can obtain more money by saying "your software has 230 patent violations" instead of "your software has 30 violations". When a manager hears of 30, he may delegate some employees/company to check how important those 30 are for them and maybe settle for much less, but 230 could be a scary number for smaller companies and they may just settle for what MS wants, because it may be costlier to investigate those 230 violations than to pay MS royalties. They want to milk the end-users of anything they can, even if it's their competitors work, because they cannot touch the competitor. Or maybe they can with FUD?
And related to GPL3, if MS would put such restrictions then customers would complain a lot, and either GPL3 would not be used or MS would loose it's customers. My company (10k+ employees) uses perl in 90% of Windows scripts. Think about what would happen if perl would switch to GPL3 and MS would impose these restrictions.
I think they could be sued for slander. They say of violations but not disclose the entire truth, thus hurting the open-source reputation (trademark???). If some newspaper would say that a politician cheated on his wife without actually showing any pictures (proof), they would be sued for slander. So why should this not work for OpenSource? Money for lawyers seems to be the only blocker for me.
It's pretty easy to configure an X server to distinguish between a USB mouse+keyboard, and PS2 mouse+keyboard and create 2 sessions. I think you can even do it using 1 Nvidia (or ATI) card with 2 outputs.
Since USB is dinamicaly allocating resources (IDs), I don't think you can put the same type of devices on one computer and expect to be mapped always to the same session, so going with a different configuration like I said before is risky.
I forgot to say that aligning is not an issue with modern processors because of big caches. It counted for older processors because they accessed the memory directly, so they needed 16/32bit aligning. But today, the cache reads like 8-bytes (aligned) at a time, and from L1 cache you can retrieve any byte/word, with no worries about alignment. Where alignment counts is big structures+arrays, where the compiler can do nothing and you have to optimize your code/design yourself.
Speedups from switching all come from compiler. Even 32-bit Intel (I don't know about AMD) use register-renaming (and support many registers, not just what compiler assigns), but they are used only for out-of-order processing. It still uses cpu-memory load/stores, which are slower even if it's in L1 cache. With x86-64, the compiler can optimize the code by using more (and bigger) registers, so cache (or worst, memory) access is reduced. So, like from the 386/486, the compiler is a major speed factor.