"This is a common industry practice used for just about every piece of hardware and software on the market."
Not necessarily. When AMD moved to their new speed rating system for the Athlon XPs, they *usually* performed equivalently to a P4 clocked at the "speed rating" on average. It did NOT reflect the peak performance of the CPU compared to a P4, but the average instead.
It also makes it a hell of a lot harder to expand the array in the future. Chances are far more likely that someone will want to add storage at a later date than have to rebuild a drive.
Even if they DO have to rebuild a drive, it isn't that much more difficult to rebuild the partition table and then rebuild the arrays with this approach than to use the maximum amount of common space for each partition.
If you have equal sized drives and no plans for upgrading, then RAIDing the entire array together with a hardware controller is a better solution. Most people can't assume availability of equal sized drives though, and they also cannot assume that they'll never need to expand the size of their storage system.
No, each md device is fully redundant, since it contains one partition from each drive.
Splitting the drives into smaller partitions and then RAIDing the partitions provides more flexibility. As I mentioned above, one thing it allows for is the ability to RAID drives of unequal sizes. Also, it provides HUGE benefits when you decide to upgrade your system by adding a drive or replacing a drive with a larger one. LVM has an extremely solid track record, so I wouldn't really consider it a point of failure. (At the very least it's a HIGHLY unlikely point of failure unlike the hardware.
Another benefit is that you can use different RAID levels on each of the partition sets. For example, one partition set of 25G partitions could be RAID 0 for stuff that you don't care about at all but for which you want lots of I/O speed. (Temp directories for compiling, DVD rips/recodes, etc.) Another set could be RAID 1 for the stuff you want to be able to survive a multi-drive failure. The rest could be RAID 5 for decent redundancy and performance.
Numerous engineering tools (such as Xilinx WebPack) do not run on MacOS. There is a Linux version of the Xilinx tools, but I believe only in the most expensive versions ($1000-2000+ a pop). Cheaper versions are Windows-only. Also won't work in emulation/virtualization since it needs to access the FPGA programmer hardware.
The list of Windows-only apps that may be enough to hold someone back from switching to a Mac is endless. No, except in a few situations I think it's stupid to fully nuke MacOS and install only Windows, but there are a lot of people who have seriously considered switching to a Mac but haven't because of "that one Windows app" they need. Even if they may spend 95%+ of their time booted into MacOS eventually, that remaining 5% is so critical that they NEED Windows around. As to buying a second machine - that costs more money than dual-booting, and lugging two laptops around you on a trip is a pain. Given the choice of only bringing one laptop, almost anyone will choose the one that provides functionality they absolutely need, even if it's not normally their preference.
There are some great posts on this topic in a past Slashdot discussion (Taco should've done his Googling ffs, it was only 2-3 months ago that the discussion in question was on Ask/.)
Split drives into small partitions, say 20-25 GB each. Since most drives available now are a multiple of 50GB, I suggest going with 25GB or 50GB per partition. Make software RAID devices out of sets of these partitions, one on each drive. e.g. md5 = sda5 + sdb5 + sdc5 + sdd5. Take all of those smaller RAID drives, and then LVM them together.
I just set up such a system on my dad's fileserver back at home, and will be doing the same with a machine I'm building within the next week or two. So far my opinion is that this approach ROCKS.
There are more details on neat tricks you can do with such a RAID + LVM setup in the discussion I posted the link to. Among other things, if you have a 150GB drive and three 250GB drives, you can have four-drive RAID for the first 150GB of the drive set, then 3-drive RAID with the remaining 100GB.
When you're talking about BIOS (which is essentially what the compatibility module is, a legacy BIOS), the "platform" goes way beyond the base chipset. It deals with every single peripheral on the mainboard, and often implementation-specific timing features. As an example, try flashing the BIOS from one nForce4 motherboard onto another nForce4 board - If your system remains bootable at all, it's likely going to have severely reduced functionality/reliability.
Same goes for the EFI compatibility module - It doesn't just have to be written for the 945 in general, it has to be written for the exact 945 implementation Apple uses. (i.e. specific to the MacBook or iMac only.)
DAoC most definately does not run under WINE, except possibly the SI client which has for all practical purposes been obsoleted by Mythic. Transgaming didn't even support the SI client the last time I checked, they only supported the Classic client, which has been totally obsoleted by Mythic since the release of New Frontiers, their free (and essentially required) RvR expansion.
None of the more recent DAoC clients run under WINE or Cedega - I've tried with every Cedega release and failed. Also, even if they did get it working, something about WINE/Cedega's filesystem handling code makes the patcher's file check extremely slow, on the order of 5-10 minutes under WINE or Cedega vs. 10 seconds under Windows.
My other two examples may have been bad, but there are still hundreds of games out there that have no Mac support and no ports planned.
"It seems unlikely that Apple would have left this out. It has already said it isn't doing anything to prevent Windows from booting on a Mac."
They have said they won't actively take any measures to prevent Windows from booting on an Intel-based Mac, but they've also made it clear they have no intention of actively supporting Windows on a Mac.
It's not like removing the CSM would require any additional work, considering that unless it's written for the platform in question, the CSM doesn't exist in the first place! From the documentation I've seen, the compatibility module is not a generic off-the-shelf component that you can just compile in - It has to be custom-built for the platform, just like legacy BIOS is always specific to a particular platform (usually specific to only one single motherboard design.) Adding legacy compatibility to their Intel products would require a LOT of development work on Apple's part.
In short, Apple will take the easiest and cheapest route. If it were harder to release an EFI system without legacy compatibility, Apple would just leave the compatibility module in. Unfortunately, it's almost guaranteed that it will be the other way around - putting in the optional compatibility module will require significant effort.
I've been thinking of the same thing myself, except I've decided against it (I'll go into why later).
While I've had OK results with X-10 equipment (they were a good company until they started their popunder/spam/camera obsession. rather than updating their products for modern times, they went on an annoying advertising spree. The end result is that Smarthome's Insteon is going to kick X-10's ass in its original market.), I would go with something more robust/flexible than X-10 now. Smarthome (www.smarhome.com), mentioned in a few posts here, has an RS-485 thermostat available. The documentation seems to indicate that their $100ish thermostat requires a $200+ controller, but I can't be sure of that. You may be able to get away with a PC and a $30-50 RS-232 to RS-485 converter. (The $200 controller appeared to be meant for a PC-less environment.)
That said, a few other people have suggested simply working on upgrading your insulation. Within a week of moving to this apartment, I realized the value insulation has. I can turn my heater completely off and the temp will never go below 65 degrees Fahrenheit (heated by leakage from the neighbors). Huge difference from my parents' house, which costs $100-200/month to heat even though they have the temperature at insanely cold levels. The house in question isnt' even that old - it's only 15ish year old.
As nice as a computer-controlled thermostat is, good insulation is even better.
Except for direct wired Ethernet, not a single one of those features can't be obtained on (even faster) Palm devices. And unlike the Newton, 802.11b doesn't need to be hacked in, some of those Palms come with it built in.
The only recent Palm device that requires any hacks to get 802.11b running are the Treos, and the service providers (Verizon, etc) have to be blamed for that.
That said, even my dad's old Palm Professional blew away the Newton we had before it. The Newton was a monstrous anemic brick with utterly crappy handwriting recognition. The Palm Pro *just worked*. Every Palm we've used since that Pro has been an improvement on that, and light years ahead of the Newton that's been collecting dust on my dresser for a decade. Currently my dad has an (ancient by today's standards) Kyocera 6035, and I've got a Treo 650, and we both love them.
At least in this case, the HSD system is so dead-on similar to what is described in the patent that I wouldn't be surprised if at some point, some Toyota engineer had seen this other company's technology.
Toyota most likely did not willfully rip off this company's technology, but if even a single engineer on the project saw even some details of the patent or this marine company's systems, that "seed" idea is enough to give the original company entitlement to some royalties. It's going to be damn near impossible for Toyota to prove that not a single engineer on the project didn't come across this technology while doing a trade survey of what's out there. Hell, I'm positive that's what happened - some engineer saw this technology at some point, and even if they didn't remember where, it put a seed of an idea into their head from which HSD grew.
I believe they have to claim infringement within a certain amount of time of knowing that an infringement situation exists. If they can show that they didn't know about it, it's fine. If they knew about it and did nothing, that's a different story.
Such clauses are there because sometimes it is extremely difficult to determine whether infringement actually exists. (An example, patents on integrated circuit technologies require advanced and expensive reverse engineering techniques to identify infringement.)
Yeah, two of the base technologies have been around for ages, but this particular combination of the two is a novel idea. The patent in question is a lot narrower than the Slashdot summary makes it seem. It basically almost exactly describes Toyota's HSD system, which you must admit, IS a novel and original method for creating a hybrid vehicle. The patent will not be enforceable against most other hybrid manufacturers, as they use different hybrid systems.
In this situation, unlike some of the past abusers of the patent system, the company asserting the patent has a damned solid case from what I've seen so far, and a right to at least reasonable royalties from Toyota.
Maybe the low-level APIs are, but cross-platform toolkits are rather abundant these days. As a result, an intelligent developer who chooses their toolkits wisely (iD Software for example) can port to multiple targets with ease.
Developers are more likely to support a platform if they have a machine supporting that platform. Now it may be possible for a developer to target Windows and MacOS with a single machine, only needing to reboot for testing. If no graphics acceleration is required, they can likely target both *without rebooting* (assuming that Apple doesn't disable the Core Duo's VT extensions in firmware.).
When he says "performance intensive", does he mean "CPU intensive" or "CPU and graphics intensive"?
Read the Xen mailing lists. It's designed for servers. People apparently HAVE gotten it to run vanilla Windows XP on CPUs that support VT.
There's one big problem - Only the host OS gets accelerated video. From what I've seen, the guest OSes essentially display via VNC. Other graphics options may be available, but it's pretty clear that accelerated graphics (esp. 3D accelerated) aren't in the picture or even close to it.
As much as I like the MacBook hardware, unless it can reliably run 3D accelerated Windows apps (specifically games) with solid drivers (this means an NVidia option instead of ATI), I won't be buying. If I can run Windows apps reliably with 3D acceleration, there's a good chance I'll buy it. If somehow 3D acceleration with virtualized Windows becomes possible, I'd order it immediately.
It has a maximum life, even with the utmost care, of 5-10 years (if you're lucky) before the battery fails. By that time, the device is obsolete and you can't buy a new battery.
Less if you actually use the device routinely - note that many people consider themselves incredibly lucky if they get 3-4 years out of a lithium-ion device before it fails. (laptop, iPod, etc.)
Meanwhile, the bookshelf at my family's home has paperbacks that are older than I am. Many of those were obtained from garage sales out of boxes with piles of unsorted books - you greatly underestimate the durability of paperback books.
It means that if you buy 1000 processors at once, you get that price.
If you buy a single unit, it will cost (usually significantly) more. I'm not sure if AMD or Intel will sell quantities less than 1000 directly to anyone. If you buy from someone who bought 1k units from AMD or Intel, you're definately going to pay more than that price.
Expect these to be available to consumers at a 10% markup or more.
If you buy more than 1000 units, you may get a lower price.
I've seen some ICs go for $5+ each in quantities below 100, drop to less than half that for order quantities over 100, and then go sub-$1 for quantities over 1000.
Books (even paperback) have MUCH longer lifetimes than any piece of consumer electronics.
Thus, the lifetime of your DRMed content is going to be 5-10 years at most before the thing breaks and your DRMed ebooks are now unreadable. (Unless someone cracks the DRM)
Even if the original poster lives near an Apple store or other iPod-carrying retailer, do you realize how idiotic your advice is?
Hint: Think about the kind of listening environment that a busy Apple store presents. Especially since Apple stores are usually located in shopping malls.
Same goes for many other brick and mortar stores such as Best Buy. Circuit City is an exception, as most CC stores I've seen do actually have sound-isolated listening booths for demonstrating some of what they sell. That said, I don't think any CC is set up so that you can demo iPods in those booths, just the higher end home theater and car audio stuff.
Have you looked at the specs of the headphones you are using?
If they're low impedance (16 ohms, sometimes even 32 ohms), try switching to headphones with a higher impedance, or use an external headphone amp.
There's a good chance the clipping will disappear even if you listen at higher volume levels - from what I've heard the biggest problem with iPod sound quality is that the amp has insufficient current handling capability.
Actually, given the cause of most of the sound quality problems I've heard about, you would not notice most of them when hooking it up to a $10,000 Hi-Fi system. The problem is that the iPods apparently have output amps with current handling capability that is way too low. As a result, the amps start becoming nonlinear as hell when connected to lower-impedance headphones. (I've heard 16-ohm headphones sound like shit in iPods at any reasonable listening level, and 32-ohm headphones are so-so, while anything 64 ohms and above has no problems.)
If you were connecting the iPod to a hifi system, its inputs would likely have (at the lowest) a 75 ohm impedance. Thus, the worst of the sound quality problems would disappear.
Hmm, I wasn't thinking of using the flash as a writeback cache, I assumed you were talking about using it as a read cache. (Since overall, more reading is done than writing with most usage patterns.) The problem with using flash as a read cache from the hard drive is that you have to write something to the flash in order to read from it, and the flash memory's low write speed will be a huge bottleneck here. It really doesn't work that well if you use something as a cache that has a lower write speed than the read speed of whatever you're caching.:)
Volume-priced PC100 SDRAM please! IIRC, PC100 had a 100 MHz clock and was 32 bits wide, giving around 400 MB/sec of transfer rate, 8-10 times the sustained read speed of most hard drives. 4 gigs of that as a drive buffer would be great.
Hmm, I wonder if low-speed SDRAM chips are available at reasonable prices... The sticks are obscenely expensive per megabyte, but that's probably because of price-gouging the few people who need PC100 sticks. The ICs themselves not in PC100 format might be much cheaper.
Slashdot Mirror
"This is a common industry practice used for just about every piece of hardware and software on the market."
Not necessarily. When AMD moved to their new speed rating system for the Athlon XPs, they *usually* performed equivalently to a P4 clocked at the "speed rating" on average. It did NOT reflect the peak performance of the CPU compared to a P4, but the average instead.
It also makes it a hell of a lot harder to expand the array in the future. Chances are far more likely that someone will want to add storage at a later date than have to rebuild a drive.
Even if they DO have to rebuild a drive, it isn't that much more difficult to rebuild the partition table and then rebuild the arrays with this approach than to use the maximum amount of common space for each partition.
If you have equal sized drives and no plans for upgrading, then RAIDing the entire array together with a hardware controller is a better solution. Most people can't assume availability of equal sized drives though, and they also cannot assume that they'll never need to expand the size of their storage system.
No, each md device is fully redundant, since it contains one partition from each drive.
r eshold=-1&commentsort=0&mode=thread&cid=14117414 is the exact link to Swillden's post about his setup (which I am now using) and the nifty tricks you can do with LVM.
Splitting the drives into smaller partitions and then RAIDing the partitions provides more flexibility. As I mentioned above, one thing it allows for is the ability to RAID drives of unequal sizes. Also, it provides HUGE benefits when you decide to upgrade your system by adding a drive or replacing a drive with a larger one. LVM has an extremely solid track record, so I wouldn't really consider it a point of failure. (At the very least it's a HIGHLY unlikely point of failure unlike the hardware.
Another benefit is that you can use different RAID levels on each of the partition sets. For example, one partition set of 25G partitions could be RAID 0 for stuff that you don't care about at all but for which you want lots of I/O speed. (Temp directories for compiling, DVD rips/recodes, etc.) Another set could be RAID 1 for the stuff you want to be able to survive a multi-drive failure. The rest could be RAID 5 for decent redundancy and performance.
http://ask.slashdot.org/comments.pl?sid=169386&th
I forgot a few other examples:
Numerous engineering tools (such as Xilinx WebPack) do not run on MacOS. There is a Linux version of the Xilinx tools, but I believe only in the most expensive versions ($1000-2000+ a pop). Cheaper versions are Windows-only. Also won't work in emulation/virtualization since it needs to access the FPGA programmer hardware.
The list of Windows-only apps that may be enough to hold someone back from switching to a Mac is endless. No, except in a few situations I think it's stupid to fully nuke MacOS and install only Windows, but there are a lot of people who have seriously considered switching to a Mac but haven't because of "that one Windows app" they need. Even if they may spend 95%+ of their time booted into MacOS eventually, that remaining 5% is so critical that they NEED Windows around. As to buying a second machine - that costs more money than dual-booting, and lugging two laptops around you on a trip is a pain. Given the choice of only bringing one laptop, almost anyone will choose the one that provides functionality they absolutely need, even if it's not normally their preference.
There are some great posts on this topic in a past Slashdot discussion (Taco should've done his Googling ffs, it was only 2-3 months ago that the discussion in question was on Ask /.)
3 37226
The discussion in question http://ask.slashdot.org/article.pl?sid=05/11/26/0
The basic idea:
Split drives into small partitions, say 20-25 GB each. Since most drives available now are a multiple of 50GB, I suggest going with 25GB or 50GB per partition. Make software RAID devices out of sets of these partitions, one on each drive. e.g. md5 = sda5 + sdb5 + sdc5 + sdd5. Take all of those smaller RAID drives, and then LVM them together.
I just set up such a system on my dad's fileserver back at home, and will be doing the same with a machine I'm building within the next week or two. So far my opinion is that this approach ROCKS.
There are more details on neat tricks you can do with such a RAID + LVM setup in the discussion I posted the link to. Among other things, if you have a 150GB drive and three 250GB drives, you can have four-drive RAID for the first 150GB of the drive set, then 3-drive RAID with the remaining 100GB.
When you're talking about BIOS (which is essentially what the compatibility module is, a legacy BIOS), the "platform" goes way beyond the base chipset. It deals with every single peripheral on the mainboard, and often implementation-specific timing features. As an example, try flashing the BIOS from one nForce4 motherboard onto another nForce4 board - If your system remains bootable at all, it's likely going to have severely reduced functionality/reliability.
Same goes for the EFI compatibility module - It doesn't just have to be written for the 945 in general, it has to be written for the exact 945 implementation Apple uses. (i.e. specific to the MacBook or iMac only.)
DAoC most definately does not run under WINE, except possibly the SI client which has for all practical purposes been obsoleted by Mythic. Transgaming didn't even support the SI client the last time I checked, they only supported the Classic client, which has been totally obsoleted by Mythic since the release of New Frontiers, their free (and essentially required) RvR expansion.
None of the more recent DAoC clients run under WINE or Cedega - I've tried with every Cedega release and failed. Also, even if they did get it working, something about WINE/Cedega's filesystem handling code makes the patcher's file check extremely slow, on the order of 5-10 minutes under WINE or Cedega vs. 10 seconds under Windows.
My other two examples may have been bad, but there are still hundreds of games out there that have no Mac support and no ports planned.
"It seems unlikely that Apple would have left this out. It has already said it isn't doing anything to prevent Windows from booting on a Mac."
They have said they won't actively take any measures to prevent Windows from booting on an Intel-based Mac, but they've also made it clear they have no intention of actively supporting Windows on a Mac.
It's not like removing the CSM would require any additional work, considering that unless it's written for the platform in question, the CSM doesn't exist in the first place! From the documentation I've seen, the compatibility module is not a generic off-the-shelf component that you can just compile in - It has to be custom-built for the platform, just like legacy BIOS is always specific to a particular platform (usually specific to only one single motherboard design.) Adding legacy compatibility to their Intel products would require a LOT of development work on Apple's part.
In short, Apple will take the easiest and cheapest route. If it were harder to release an EFI system without legacy compatibility, Apple would just leave the compatibility module in. Unfortunately, it's almost guaranteed that it will be the other way around - putting in the optional compatibility module will require significant effort.
Hmm, let me list a few applications that virtualization/emulation won't work for, which also aren't available for OSX.
World of Warcraft
Dark Age of Camelot
Battlefield 2 (possibly available under OSX, but I don't think so...)
Do you sense a recurring theme yet?
I've been thinking of the same thing myself, except I've decided against it (I'll go into why later).
While I've had OK results with X-10 equipment (they were a good company until they started their popunder/spam/camera obsession. rather than updating their products for modern times, they went on an annoying advertising spree. The end result is that Smarthome's Insteon is going to kick X-10's ass in its original market.), I would go with something more robust/flexible than X-10 now. Smarthome (www.smarhome.com), mentioned in a few posts here, has an RS-485 thermostat available. The documentation seems to indicate that their $100ish thermostat requires a $200+ controller, but I can't be sure of that. You may be able to get away with a PC and a $30-50 RS-232 to RS-485 converter. (The $200 controller appeared to be meant for a PC-less environment.)
That said, a few other people have suggested simply working on upgrading your insulation. Within a week of moving to this apartment, I realized the value insulation has. I can turn my heater completely off and the temp will never go below 65 degrees Fahrenheit (heated by leakage from the neighbors). Huge difference from my parents' house, which costs $100-200/month to heat even though they have the temperature at insanely cold levels. The house in question isnt' even that old - it's only 15ish year old.
As nice as a computer-controlled thermostat is, good insulation is even better.
Except for direct wired Ethernet, not a single one of those features can't be obtained on (even faster) Palm devices. And unlike the Newton, 802.11b doesn't need to be hacked in, some of those Palms come with it built in.
The only recent Palm device that requires any hacks to get 802.11b running are the Treos, and the service providers (Verizon, etc) have to be blamed for that.
That said, even my dad's old Palm Professional blew away the Newton we had before it. The Newton was a monstrous anemic brick with utterly crappy handwriting recognition. The Palm Pro *just worked*. Every Palm we've used since that Pro has been an improvement on that, and light years ahead of the Newton that's been collecting dust on my dresser for a decade. Currently my dad has an (ancient by today's standards) Kyocera 6035, and I've got a Treo 650, and we both love them.
Most likely a password-protected BT tracker or DC hub.
At least in this case, the HSD system is so dead-on similar to what is described in the patent that I wouldn't be surprised if at some point, some Toyota engineer had seen this other company's technology.
Toyota most likely did not willfully rip off this company's technology, but if even a single engineer on the project saw even some details of the patent or this marine company's systems, that "seed" idea is enough to give the original company entitlement to some royalties. It's going to be damn near impossible for Toyota to prove that not a single engineer on the project didn't come across this technology while doing a trade survey of what's out there. Hell, I'm positive that's what happened - some engineer saw this technology at some point, and even if they didn't remember where, it put a seed of an idea into their head from which HSD grew.
I believe they have to claim infringement within a certain amount of time of knowing that an infringement situation exists. If they can show that they didn't know about it, it's fine. If they knew about it and did nothing, that's a different story.
Such clauses are there because sometimes it is extremely difficult to determine whether infringement actually exists. (An example, patents on integrated circuit technologies require advanced and expensive reverse engineering techniques to identify infringement.)
Yeah, two of the base technologies have been around for ages, but this particular combination of the two is a novel idea. The patent in question is a lot narrower than the Slashdot summary makes it seem. It basically almost exactly describes Toyota's HSD system, which you must admit, IS a novel and original method for creating a hybrid vehicle. The patent will not be enforceable against most other hybrid manufacturers, as they use different hybrid systems.
In this situation, unlike some of the past abusers of the patent system, the company asserting the patent has a damned solid case from what I've seen so far, and a right to at least reasonable royalties from Toyota.
Maybe the low-level APIs are, but cross-platform toolkits are rather abundant these days. As a result, an intelligent developer who chooses their toolkits wisely (iD Software for example) can port to multiple targets with ease.
Developers are more likely to support a platform if they have a machine supporting that platform. Now it may be possible for a developer to target Windows and MacOS with a single machine, only needing to reboot for testing. If no graphics acceleration is required, they can likely target both *without rebooting* (assuming that Apple doesn't disable the Core Duo's VT extensions in firmware.).
Garmin MapSource won't run under WINE.
Trust me, I've tried. MapSource and Dark Age of Camelot are pretty much the only reasons I keep a Windows partition around these days.
When he says "performance intensive", does he mean "CPU intensive" or "CPU and graphics intensive"?
Read the Xen mailing lists. It's designed for servers. People apparently HAVE gotten it to run vanilla Windows XP on CPUs that support VT.
There's one big problem - Only the host OS gets accelerated video. From what I've seen, the guest OSes essentially display via VNC. Other graphics options may be available, but it's pretty clear that accelerated graphics (esp. 3D accelerated) aren't in the picture or even close to it.
As much as I like the MacBook hardware, unless it can reliably run 3D accelerated Windows apps (specifically games) with solid drivers (this means an NVidia option instead of ATI), I won't be buying. If I can run Windows apps reliably with 3D acceleration, there's a good chance I'll buy it. If somehow 3D acceleration with virtualized Windows becomes possible, I'd order it immediately.
This is a piece of consumer electronics.
More importantly, it has a battery.
It has a maximum life, even with the utmost care, of 5-10 years (if you're lucky) before the battery fails. By that time, the device is obsolete and you can't buy a new battery.
Less if you actually use the device routinely - note that many people consider themselves incredibly lucky if they get 3-4 years out of a lithium-ion device before it fails. (laptop, iPod, etc.)
Meanwhile, the bookshelf at my family's home has paperbacks that are older than I am. Many of those were obtained from garage sales out of boxes with piles of unsorted books - you greatly underestimate the durability of paperback books.
It means that if you buy 1000 processors at once, you get that price.
If you buy a single unit, it will cost (usually significantly) more. I'm not sure if AMD or Intel will sell quantities less than 1000 directly to anyone. If you buy from someone who bought 1k units from AMD or Intel, you're definately going to pay more than that price.
Expect these to be available to consumers at a 10% markup or more.
If you buy more than 1000 units, you may get a lower price.
I've seen some ICs go for $5+ each in quantities below 100, drop to less than half that for order quantities over 100, and then go sub-$1 for quantities over 1000.
Simple.
Books (even paperback) have MUCH longer lifetimes than any piece of consumer electronics.
Thus, the lifetime of your DRMed content is going to be 5-10 years at most before the thing breaks and your DRMed ebooks are now unreadable. (Unless someone cracks the DRM)
Even if the original poster lives near an Apple store or other iPod-carrying retailer, do you realize how idiotic your advice is?
Hint: Think about the kind of listening environment that a busy Apple store presents. Especially since Apple stores are usually located in shopping malls.
Same goes for many other brick and mortar stores such as Best Buy. Circuit City is an exception, as most CC stores I've seen do actually have sound-isolated listening booths for demonstrating some of what they sell. That said, I don't think any CC is set up so that you can demo iPods in those booths, just the higher end home theater and car audio stuff.
Have you looked at the specs of the headphones you are using?
If they're low impedance (16 ohms, sometimes even 32 ohms), try switching to headphones with a higher impedance, or use an external headphone amp.
There's a good chance the clipping will disappear even if you listen at higher volume levels - from what I've heard the biggest problem with iPod sound quality is that the amp has insufficient current handling capability.
Actually, given the cause of most of the sound quality problems I've heard about, you would not notice most of them when hooking it up to a $10,000 Hi-Fi system. The problem is that the iPods apparently have output amps with current handling capability that is way too low. As a result, the amps start becoming nonlinear as hell when connected to lower-impedance headphones. (I've heard 16-ohm headphones sound like shit in iPods at any reasonable listening level, and 32-ohm headphones are so-so, while anything 64 ohms and above has no problems.)
If you were connecting the iPod to a hifi system, its inputs would likely have (at the lowest) a 75 ohm impedance. Thus, the worst of the sound quality problems would disappear.
Hmm, I wasn't thinking of using the flash as a writeback cache, I assumed you were talking about using it as a read cache. (Since overall, more reading is done than writing with most usage patterns.) The problem with using flash as a read cache from the hard drive is that you have to write something to the flash in order to read from it, and the flash memory's low write speed will be a huge bottleneck here. It really doesn't work that well if you use something as a cache that has a lower write speed than the read speed of whatever you're caching. :)
Volume-priced PC100 SDRAM please! IIRC, PC100 had a 100 MHz clock and was 32 bits wide, giving around 400 MB/sec of transfer rate, 8-10 times the sustained read speed of most hard drives. 4 gigs of that as a drive buffer would be great.
Hmm, I wonder if low-speed SDRAM chips are available at reasonable prices... The sticks are obscenely expensive per megabyte, but that's probably because of price-gouging the few people who need PC100 sticks. The ICs themselves not in PC100 format might be much cheaper.