Precisely, which is why I'm trying to disprove the rumour of them being stupid enough to solder flash memory to the mainboard. Soldering NAND flash to the logic board would be like having tyres on your car attached so you have to replace the whole chassis whenever they wear out. Hard disks have always been the part of the computer most likely to wear out, and so can be replaced relatively inexpensively. If they're soldered to the mainboard, you'd have to fork out $1000+ just to replace something that's known to have a finite life.
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
The term "wear out" is being used in different ways between solid state and GMR (Giant MagnetoResistance) (aka HDD) drives.
To put it simply, solid state drives are limited to the number of times they can be written to and HDDs are limited to the rotations the motor can do.
Here is an AnandTech article that should answer all your questions and concerns in greater detail.
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
NAND cells wear out because they have to be zapped with about 10 volts to erase them. Each zap damages them slightly, until eventually, they can no longer be erased or written to reliably. If you only ever read from a SSD, it would last forever.
CPUs and RAM don't suffer this problem because they don't hold the value stored in them permanently. DRAM resets to zero unless it's refreshed around 20 times a second, and CPUs don't have permanent storage, just temporary cache and even more temporary registers. There's no zapping involved, just 1v or less changing the transistor state temporarily. Hope that helps
In addition to SB's Turbo Boost that it will get, the Air is looking more and more compelling as a main laptop for me. The current one is good, but I can feel it hit its processors limits quite often. With Turbo Boost for single threaded apps as well as SB's faster IPC, processor-wise it will outdo most midrange Core 2 Duo laptops in the majority of workloads, and with an SSD this fast it will feel much faster day to day (well, the current one already does that).
A NAND capacity doubling at the same prices would be icing on the cake, I can make due with 256 easily. And as always, time to make 4GB RAM baseline, I say.
NAND cells wear out because they have to be zapped with about 10 volts to erase them. Each zap damages them slightly, until eventually, they can no longer be erased or written to reliably. If you only ever read from a SSD, it would last forever.
CPUs and RAM don't suffer this problem because they don't hold the value stored in them permanently. DRAM resets to zero unless it's refreshed around 20 times a second, and CPUs don't have permanent storage, just temporary cache and even more temporary registers. There's no zapping involved, just 1v or less changing the transistor state temporarily. Hope that helps
Quote:
Originally Posted by solipsism
The term "wear out" is being used in different ways between solid state and GMR (Giant MagnetoResistance) (aka HDD) drives.
To put it simply, solid state drives are limited to the number of times they can be written to and HDDs are limited to the rotations the motor can do.
Here is an AnandTech article that should answer all your questions and concerns in greater detail.
Given this, and, if the soldering-NAND-to-the-motherboard rumor is true, then my guess is that it will only be done with the MBA and the MB (assuming that's not discontinued), but the MBPs will still have the SSDs on a card—at least I hope so. About a year ago, I swapped out the 160 GB HD on my MacBook for a 500GB drive, and I'm glad I had that capability—and certainly, anyone living in the "Pro" universe would want some flexibility with their investment as well. That said, I can understand if Apple would make compromises for the lower end—sacrifice upgradability for portability and to keep the prices (relatively) low.
Intel's controller understands NTFS and FAT, so it knows what areas the filesystem has tagged as deleted. No controllers (that I know of) understand HFS, so they don't know what parts of the SSD aren't in use. That's the idea of TRIM, the OS that understands the filesystem can tell the SSD the areas that can be erased.
I think you can be pretty sure the SSD's Apple uses are aware of the filesystem so they can do garbage collection when idle in exactly the way you describe. It's not a controller feature, but a firmware feature, and most likely the main reason the SSD's Apple puts in their machines run custom firmware. OCZ drives have had garbage collecting firmwares that work on NTFS, FAT and ext2/3 (and probably also ext4) for quite a while by the way, and elaborate tests you can find on their support forums show that it's actually faster than TRIM, because you get higher IOPS with the GC firmwares. All in all it's nothing specific to Intel drives, and there is no reason to assume Apple left out TRIM for so long *and* run SSDs without GC firmware. HPFS is very similar to NTFS by the way, they have the same origin (OS/2) and are almost each others evil twins, so modifying a GC firmware with NTFS support so it works with HPFS filesystems should be relatively easy.
As an aside: TRIM support is really one of these examples of overhyped acronyms that almost no-one actually understands but still everyone thinks is essential for SSD performance. It's not, TRIM serves only 1 purpose: prevent performance degradation. Even though TRIM is a much more elegant solution for this, GC firmwares do just as well.
In addition to SB's Turbo Boost that it will get, the Air is looking more and more compelling as a main laptop for me. The current one is good, but I can feel it hit its processors limits quite often. With Turbo Boost for single threaded apps as well as SB's faster IPC, processor-wise it will outdo most midrange Core 2 Duo laptops in the majority of workloads, and with an SSD this fast it will feel much faster day to day (well, the current one already does that).
Where did you get that from? Flash is reliable yes, but the lifetime isn't so good. Thats completely false if the drive is written to a lot, SSDs have a limited number of erase/write cycles. Mine is just a year old, and already has 25% of its erase/write cycles consumed.
How do you know how many write cycles have been consumed? Is there a utility, or does it show in the System Profiler?
Where did you get that from?.... Mine is just a year old, and already has 25% of its erase/write cycles consumed....
Where could one find that write/erase cycles info if you don't mind telling. Mind though, my SSD is only on Windows but Macs only info is just fine too.
Not sure for this specific technology but SSD's in general are very reliable and have a lifespan of 10+ years
Lifespan depends upon a number of things including feature size and the specific technology used. So while 10 years may be true for most of the SSDs currently on the market it is a given for the devices built on much smaller feature sizes. In general the smaller the flash memory cell the less reliable.
Quote:
Bottom line, a normal hard drive will fail sooner then an SSD
For the most part that is true today but may not hold at all for denser chips.
Quote:
Wonder if this means there will be higher capacity options available? Like a 512GB air...
Honestly I hope so. That is higher capacity at a lower cost. Behind screen size it is my next issue with the AIRs. I simply need more than 256GB of boot drive storage.
Doesn't matter. The major fact that has to be dealt with here is that flash becomes less reliable as feature size shrinks.
Quote:
Besides, there are lot's of strategies for writing, especially regarding virtual memory, and it is very likely that Apple took advantage of these. If the rumor that Apple is holding off the shipment of the new hardware for Lion release is true, there might be a technical reason for that.
Doubtful!
However I have heard through the grapevine that it is taking Apple a lot longer to validate the newer flash chips for production. This is a very bad sign if true.
Personally I have to agree with the many others here expressing concern about SSD drive wear. The life span of an SSD depends to much upon how it is used. In some applications they will wear out faster than a magnetic drive. So we could have one AIR user get his ten years out of the drive and the guy next to him wearing it out in ten months.
I predict that the storage sizes won't shift much at all. In fact the upgrade here will probably be speed not quantity. Being somewhat realistic about this, we should expect a MBA that is basically 2x as fast as the one it replaces (in terms of both cpu performance and storage performance, perhaps faster) at the same prices.
External I/O performance will be magnitudes faster, obviously, with thunderbolt.
The Macworld benchmarks prove that the current top end 13" MBA (2.13ghz C2D, 4GB ram) is sometimes slower, sometimes faster, but on balance is as fast or faster than a 2.4ghz Corei5 MBP. This is no small feat. Apple may be able to argue that the new MBA with faster cpu, disk and I/O is the fastest laptop they've ever shipped. And if the current numbers hold true will also be the thinnest and lightest, and one of the least expensive. So really who cares if the storage sizes are the same, this is a very compelling argument to buy the new MBA and begin to move away from traditional laptops with their outmoded disk drives and heavy, bulky form factors.
I think that this will be more than good enough for a cycle upgrade less than 1 year from the last release.
I see, you are much smarter than all Apple engineers combined. They decided to put the FLASH on the main board to make Steve Jobs happy, and none of them was aware that the FLASH memory can wear off so they did nothing about it. Oh, no! Apple is known to sell products that fell apart after 1 year of usage so you are forced to buy a new one!
Let me remind you, we are commenting on rumor here. It may turn out that Apple was just experimenting with such approach, or the rumor is made up just to get more clicks. Apple does make mistakes, but the engineers there are not stupid. If they decided they can put the FLASH on the motherboard, they know what they are doing. Or know better that most (all?) of the folks who commented above.
There are real issues with flash based SSDs. Apple engineers putting such tech on the motherboard does not invalidate those issues.
The fundamental problem with flash is that wear is dependent on usage and feature size. No engineer at Apple can make those issues go completely away. So yeah you have the potential of a machine failing within a year depending upon who uses it.
It really has nothing to do with the engineers being stupid but rather it has everything to do with a technology that works in a dramatically different way than a magnetic drive. This difference means that users impact lifetimes in a way that is not seen on magnetic drives. People just need to realize that SSDs work out really well for many users.
I think you can be pretty sure the SSD's Apple uses are aware of the filesystem so they can do garbage collection when idle in exactly the way you describe. It's not a controller feature, but a firmware feature, and most likely the main reason the SSD's Apple puts in their machines run custom firmware.
I didn't think the Air had a custom firmware on their SSDs? Not sure though. I read reports of the early (SATA) Air ones slowing down after a few months of use, so any custom firmware on those wasn't doing its job or didn't have garbage collection. The only custom firmware I know of that Apple's used on their HDDs is in the Thunderbolt iMac.
Quote:
Originally Posted by JeffDM
How do you know how many write cycles have been consumed? Is there a utility, or does it show in the System Profiler?
Quote:
Originally Posted by Splash-reverse
Where could one find that write/erase cycles info if you don't mind telling. Mind though, my SSD is only on Windows but Macs only info is just fine too.
It's not in System Profiler, you need to use smartctl to get the info. It's not that easy to build yourself, so just get something like SmartReporter which has a precompiled version in it. Just mount the disk image, and the commands below should work.
Run this command and it'll show the SMART attributes, you're looking for ID 209, Remaining_Lifetime_Percent, and maybe the couple above that in the list too.
Code:
/Volumes/SMARTReporter/SMARTReporter/SMARTReporter.app/Contents/Resources/smartctl -a disk0
If you get an error about SMART being disabled, just run:
Code:
/Volumes/SMARTReporter/SMARTReporter/SMARTReporter.app/Contents/Resources/smartctl -s on disk0
Both assume you are using the SSD as your startup drive, if not, replace "disk0" with the BSD name of your drive, such as disk1, disk2, etc.
Here's my output:
And yeah, that is 10 billion reads. Phew. You can see the disparity between Error_Bits_Flash_Tot_Ct and Corr_Read_Errors_Tot_Ct, the uncorrectable errors are sent to the host which then logs it as an I/O error, so it shows SSDs aren't quite perfect.
I'd be interested in your smartctl output too if you'd care to post?
I understand that. When I had my company, I used to replace my own, at home, PowerMac, every two years, and upgrade between. But when we sold the company in 2004, and I retired, I'm replacing my Mac Pro less often, maybe four years. I'm fortunate to be able to afford it, but I understand the more tightly controlled buyers of others.
SSD's are still a new technology. Right now, their reliability is, on average, no better than a HDD. intel's are much better, and OCZs' are much worse. It's hard to understand why, but as flash memory on iPods seem to be lasting for along time, with admittedly many more reads than writes, it's thought that the reliability of these drives are due to some other factor, since many are using the same chips and controllers.
The average PC OS puts a lot of stress on a solid state drive that one does not see in a iPod. It will be very interesting to see how iPad and iPod Touch hold up in the long run as usage changes. Some of the new software initiatives in iOS5 should cause more wear than has been seen in the past.
Upgrading a drive is the #1 thing that I do to breathe new life into a computer.
Mac laptops have almost always shipped with low performance drives as the standard option. My Pismo was 4200 RPM, my Blackbook was 5400 RPM. The first thing I'd do whenever I got a new Mac is to swap out the drive and put in a larger and faster drive. Even the brand new 2011 Macbook Pro's only come with a 5400 drive as standard.
However, flash drives are a totally different ballgame, and if the machine already comes with a super fast drive built into it, then I can't really complain about that.
In addition to SB's Turbo Boost that it will get, the Air is looking more and more compelling as a main laptop for me.
Turbo boost is a sword that cuts both ways. The boost results in significant thermal loading thus impacting battery life.
Quote:
The current one is good, but I can feel it hit its processors limits quite often. With Turbo Boost for single threaded apps as well as SB's faster IPC, processor-wise it will outdo most midrange Core 2 Duo laptops in the majority of workloads, and with an SSD this fast it will feel much faster day to day (well, the current one already does that).
Again don't get to excited about turbo boost until hardware is out in the wild. Thermal throttling could be an issue. You need to know that the boost can be maintained for your problem set.
Quote:
A NAND capacity doubling at the same prices would be icing on the cake, I can make due with 256 easily. And as always, time to make 4GB RAM baseline, I say.
Cost is a big issue with SSD storage. 256GB is a lower limit for me and frankly I'd like a bit more. Also I have to agree about RAM. The good point here is that Apple should be able to address these two issues easily in the next rev.
Comments
Precisely, which is why I'm trying to disprove the rumour of them being stupid enough to solder flash memory to the mainboard. Soldering NAND flash to the logic board would be like having tyres on your car attached so you have to replace the whole chassis whenever they wear out. Hard disks have always been the part of the computer most likely to wear out, and so can be replaced relatively inexpensively. If they're soldered to the mainboard, you'd have to fork out $1000+ just to replace something that's known to have a finite life.
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
The term "wear out" is being used in different ways between solid state and GMR (Giant MagnetoResistance) (aka HDD) drives.
To put it simply, solid state drives are limited to the number of times they can be written to and HDDs are limited to the rotations the motor can do.
Here is an AnandTech article that should answer all your questions and concerns in greater detail.
Can someone who's more tech-savvy than I explain how it is that memory chips (i.e. NAND flash) would "wear out"?? I thought the whole point of going solid state was that there are no moving parts?>more durability, more speed & less wear. I get that HDDs wear out, because they have lots of little moving parts, but solid state "wearing out"? I don't get it.
Also, if the concern is NAND chips wearing out from usage, well, wouldn't the main processor be vulnerable to the same problems? It certainly gets A LOT more use, i.e. all the time!!
NAND cells wear out because they have to be zapped with about 10 volts to erase them. Each zap damages them slightly, until eventually, they can no longer be erased or written to reliably. If you only ever read from a SSD, it would last forever.
CPUs and RAM don't suffer this problem because they don't hold the value stored in them permanently. DRAM resets to zero unless it's refreshed around 20 times a second, and CPUs don't have permanent storage, just temporary cache and even more temporary registers. There's no zapping involved, just 1v or less changing the transistor state temporarily. Hope that helps
A NAND capacity doubling at the same prices would be icing on the cake, I can make due with 256 easily. And as always, time to make 4GB RAM baseline, I say.
NAND cells wear out because they have to be zapped with about 10 volts to erase them. Each zap damages them slightly, until eventually, they can no longer be erased or written to reliably. If you only ever read from a SSD, it would last forever.
CPUs and RAM don't suffer this problem because they don't hold the value stored in them permanently. DRAM resets to zero unless it's refreshed around 20 times a second, and CPUs don't have permanent storage, just temporary cache and even more temporary registers. There's no zapping involved, just 1v or less changing the transistor state temporarily. Hope that helps
The term "wear out" is being used in different ways between solid state and GMR (Giant MagnetoResistance) (aka HDD) drives.
To put it simply, solid state drives are limited to the number of times they can be written to and HDDs are limited to the rotations the motor can do.
Here is an AnandTech article that should answer all your questions and concerns in greater detail.
Thanks! I feel all edumacated now!
Given this, and, if the soldering-NAND-to-the-motherboard rumor is true, then my guess is that it will only be done with the MBA and the MB (assuming that's not discontinued), but the MBPs will still have the SSDs on a card—at least I hope so. About a year ago, I swapped out the 160 GB HD on my MacBook for a 500GB drive, and I'm glad I had that capability—and certainly, anyone living in the "Pro" universe would want some flexibility with their investment as well. That said, I can understand if Apple would make compromises for the lower end—sacrifice upgradability for portability and to keep the prices (relatively) low.
Isn't soldering RAM and storage directly to the motherboard what they currently do anyway?
Only for the MBA, and then only the RAM. The MBA's SSD storage is on a plug-in card connected to the motherboard, not soldered directly to it.
Isn't ~/Applications (under the Home folder) new to Lion?
No. It's been around, IIRC, at least since Tiger, probably before. It just isn't used much.
Intel's controller understands NTFS and FAT, so it knows what areas the filesystem has tagged as deleted. No controllers (that I know of) understand HFS, so they don't know what parts of the SSD aren't in use. That's the idea of TRIM, the OS that understands the filesystem can tell the SSD the areas that can be erased.
I think you can be pretty sure the SSD's Apple uses are aware of the filesystem so they can do garbage collection when idle in exactly the way you describe. It's not a controller feature, but a firmware feature, and most likely the main reason the SSD's Apple puts in their machines run custom firmware. OCZ drives have had garbage collecting firmwares that work on NTFS, FAT and ext2/3 (and probably also ext4) for quite a while by the way, and elaborate tests you can find on their support forums show that it's actually faster than TRIM, because you get higher IOPS with the GC firmwares. All in all it's nothing specific to Intel drives, and there is no reason to assume Apple left out TRIM for so long *and* run SSDs without GC firmware. HPFS is very similar to NTFS by the way, they have the same origin (OS/2) and are almost each others evil twins, so modifying a GC firmware with NTFS support so it works with HPFS filesystems should be relatively easy.
As an aside: TRIM support is really one of these examples of overhyped acronyms that almost no-one actually understands but still everyone thinks is essential for SSD performance. It's not, TRIM serves only 1 purpose: prevent performance degradation. Even though TRIM is a much more elegant solution for this, GC firmwares do just as well.
In addition to SB's Turbo Boost that it will get, the Air is looking more and more compelling as a main laptop for me. The current one is good, but I can feel it hit its processors limits quite often. With Turbo Boost for single threaded apps as well as SB's faster IPC, processor-wise it will outdo most midrange Core 2 Duo laptops in the majority of workloads, and with an SSD this fast it will feel much faster day to day (well, the current one already does that).
I agree. My next Mac portable may be an MBA.
Bottom line, a $1,000 piece of electronic equipment better not have a projected lifespan of three years. Not acceptable.
So, "renting" a piece of equipment for $27+ per month is too cheap? Seems a bit of an odd idea, somehow.
Where did you get that from? Flash is reliable yes, but the lifetime isn't so good. Thats completely false if the drive is written to a lot, SSDs have a limited number of erase/write cycles. Mine is just a year old, and already has 25% of its erase/write cycles consumed.
How do you know how many write cycles have been consumed? Is there a utility, or does it show in the System Profiler?
Where did you get that from?.... Mine is just a year old, and already has 25% of its erase/write cycles consumed....
Where could one find that write/erase cycles info if you don't mind telling. Mind though, my SSD is only on Windows but Macs only info is just fine too.
Not sure for this specific technology but SSD's in general are very reliable and have a lifespan of 10+ years
Lifespan depends upon a number of things including feature size and the specific technology used. So while 10 years may be true for most of the SSDs currently on the market it is a given for the devices built on much smaller feature sizes. In general the smaller the flash memory cell the less reliable.
Bottom line, a normal hard drive will fail sooner then an SSD
For the most part that is true today but may not hold at all for denser chips.
Wonder if this means there will be higher capacity options available? Like a 512GB air...
Honestly I hope so. That is higher capacity at a lower cost. Behind screen size it is my next issue with the AIRs. I simply need more than 256GB of boot drive storage.
There is a lot of progress on this lately, see the "Memory wear" section here:
Wikipedia - Flash memory.
Doesn't matter. The major fact that has to be dealt with here is that flash becomes less reliable as feature size shrinks.
Besides, there are lot's of strategies for writing, especially regarding virtual memory, and it is very likely that Apple took advantage of these. If the rumor that Apple is holding off the shipment of the new hardware for Lion release is true, there might be a technical reason for that.
Doubtful!
However I have heard through the grapevine that it is taking Apple a lot longer to validate the newer flash chips for production. This is a very bad sign if true.
Personally I have to agree with the many others here expressing concern about SSD drive wear. The life span of an SSD depends to much upon how it is used. In some applications they will wear out faster than a magnetic drive. So we could have one AIR user get his ten years out of the drive and the guy next to him wearing it out in ten months.
External I/O performance will be magnitudes faster, obviously, with thunderbolt.
The Macworld benchmarks prove that the current top end 13" MBA (2.13ghz C2D, 4GB ram) is sometimes slower, sometimes faster, but on balance is as fast or faster than a 2.4ghz Corei5 MBP. This is no small feat. Apple may be able to argue that the new MBA with faster cpu, disk and I/O is the fastest laptop they've ever shipped. And if the current numbers hold true will also be the thinnest and lightest, and one of the least expensive. So really who cares if the storage sizes are the same, this is a very compelling argument to buy the new MBA and begin to move away from traditional laptops with their outmoded disk drives and heavy, bulky form factors.
I think that this will be more than good enough for a cycle upgrade less than 1 year from the last release.
I see, you are much smarter than all Apple engineers combined. They decided to put the FLASH on the main board to make Steve Jobs happy, and none of them was aware that the FLASH memory can wear off so they did nothing about it. Oh, no! Apple is known to sell products that fell apart after 1 year of usage so you are forced to buy a new one!
Let me remind you, we are commenting on rumor here. It may turn out that Apple was just experimenting with such approach, or the rumor is made up just to get more clicks. Apple does make mistakes, but the engineers there are not stupid. If they decided they can put the FLASH on the motherboard, they know what they are doing. Or know better that most (all?) of the folks who commented above.
There are real issues with flash based SSDs. Apple engineers putting such tech on the motherboard does not invalidate those issues.
The fundamental problem with flash is that wear is dependent on usage and feature size. No engineer at Apple can make those issues go completely away. So yeah you have the potential of a machine failing within a year depending upon who uses it.
It really has nothing to do with the engineers being stupid but rather it has everything to do with a technology that works in a dramatically different way than a magnetic drive. This difference means that users impact lifetimes in a way that is not seen on magnetic drives. People just need to realize that SSDs work out really well for many users.
I think you can be pretty sure the SSD's Apple uses are aware of the filesystem so they can do garbage collection when idle in exactly the way you describe. It's not a controller feature, but a firmware feature, and most likely the main reason the SSD's Apple puts in their machines run custom firmware.
I didn't think the Air had a custom firmware on their SSDs? Not sure though. I read reports of the early (SATA) Air ones slowing down after a few months of use, so any custom firmware on those wasn't doing its job or didn't have garbage collection. The only custom firmware I know of that Apple's used on their HDDs is in the Thunderbolt iMac.
How do you know how many write cycles have been consumed? Is there a utility, or does it show in the System Profiler?
Where could one find that write/erase cycles info if you don't mind telling. Mind though, my SSD is only on Windows but Macs only info is just fine too.
It's not in System Profiler, you need to use smartctl to get the info. It's not that easy to build yourself, so just get something like SmartReporter which has a precompiled version in it. Just mount the disk image, and the commands below should work.
Run this command and it'll show the SMART attributes, you're looking for ID 209, Remaining_Lifetime_Percent, and maybe the couple above that in the list too.
/Volumes/SMARTReporter/SMARTReporter/SMARTReporter.app/Contents/Resources/smartctl -a disk0
If you get an error about SMART being disabled, just run:
/Volumes/SMARTReporter/SMARTReporter/SMARTReporter.app/Contents/Resources/smartctl -s on disk0
Both assume you are using the SSD as your startup drive, if not, replace "disk0" with the BSD name of your drive, such as disk1, disk2, etc.
Here's my output:
And yeah, that is 10 billion reads. Phew. You can see the disparity between Error_Bits_Flash_Tot_Ct and Corr_Read_Errors_Tot_Ct, the uncorrectable errors are sent to the host which then logs it as an I/O error, so it shows SSDs aren't quite perfect.
I'd be interested in your smartctl output too if you'd care to post?
I understand that. When I had my company, I used to replace my own, at home, PowerMac, every two years, and upgrade between. But when we sold the company in 2004, and I retired, I'm replacing my Mac Pro less often, maybe four years. I'm fortunate to be able to afford it, but I understand the more tightly controlled buyers of others.
SSD's are still a new technology. Right now, their reliability is, on average, no better than a HDD. intel's are much better, and OCZs' are much worse. It's hard to understand why, but as flash memory on iPods seem to be lasting for along time, with admittedly many more reads than writes, it's thought that the reliability of these drives are due to some other factor, since many are using the same chips and controllers.
The average PC OS puts a lot of stress on a solid state drive that one does not see in a iPod. It will be very interesting to see how iPad and iPod Touch hold up in the long run as usage changes. Some of the new software initiatives in iOS5 should cause more wear than has been seen in the past.
Upgrading a drive is the #1 thing that I do to breathe new life into a computer.
Mac laptops have almost always shipped with low performance drives as the standard option. My Pismo was 4200 RPM, my Blackbook was 5400 RPM. The first thing I'd do whenever I got a new Mac is to swap out the drive and put in a larger and faster drive. Even the brand new 2011 Macbook Pro's only come with a 5400 drive as standard.
However, flash drives are a totally different ballgame, and if the machine already comes with a super fast drive built into it, then I can't really complain about that.
In addition to SB's Turbo Boost that it will get, the Air is looking more and more compelling as a main laptop for me.
Turbo boost is a sword that cuts both ways. The boost results in significant thermal loading thus impacting battery life.
The current one is good, but I can feel it hit its processors limits quite often. With Turbo Boost for single threaded apps as well as SB's faster IPC, processor-wise it will outdo most midrange Core 2 Duo laptops in the majority of workloads, and with an SSD this fast it will feel much faster day to day (well, the current one already does that).
Again don't get to excited about turbo boost until hardware is out in the wild. Thermal throttling could be an issue. You need to know that the boost can be maintained for your problem set.
A NAND capacity doubling at the same prices would be icing on the cake, I can make due with 256 easily. And as always, time to make 4GB RAM baseline, I say.
Cost is a big issue with SSD storage. 256GB is a lower limit for me and frankly I'd like a bit more. Also I have to agree about RAM. The good point here is that Apple should be able to address these two issues easily in the next rev.