It certainly could happen soon but there are still reliability issues to over come, especially when using a Unix derived Operating System. This is not impossible of course as Linux is capable of mounting volumns that do not generate excessive writes to the "disk", just that certain properties of Flash have to be taken into account.
Further Flash is slow so the types of devices it might be implemented in first may not be the laptops we think of today. An ideal implementation platform for flash would be a tablet type device with built in WiFi. WiFi to allow data storage to a more traditional computer. What would really be neat is a tablet that is both a functional computing appliance and a video display device.
Flash offers all sorts of opportunities but I don't see it catching up with disk drive technology anytime soon. At this point it offers a low power advantage andthat is about it.
Thanks
dave
Maybe the instant on stuff is about having enough flash memory to cover the spin-up time for a HD. OK it is slow, but the HD does not start reading for at least 2 or 3 seconds when it is first starting, because it has to get up to speed. As was pointed out here.
Quote:
Originally posted by kim kap sol
We're not going to see Flash replace HDs...but what we might see are hybrid that keep files or parts of the OS that need rapid access...the HD would mostly be used for data that rarely needs attention.
Flash can be fast when read writing to it is another thing. Of course thechnology moves forward and speed improves either way.
Along your line of thinking it is rumored that Intel is working on using Flash as a Cache for disk drive access. One would still have and need the disk drive but frequently referenced code would sit in a Flash cache. This could help with instant on, but could be a huge win for power usage.
I suspect that Apple was shown a number of interesting Intel technologies related to portables and this is where the big draw for i86 came from. Sure performance is an issue but I see Intel having a much large selection of technologies to implement into a portable. Everything from integrated video to the stuff that isn't talked about in public. Freescale simply wasn't willing to even consider that market.
Dave
Quote:
Originally posted by Brendon
Maybe the instant on stuff is about having enough flash memory to cover the spin-up time for a HD. OK it is slow, but the HD does not start reading for at least 2 or 3 seconds when it is first starting, because it has to get up to speed. As was pointed out here.
The problem with solid state components is that once they go bad for whatever reason your screwed. You have to buy a new one, and I'm pretty sure file recovery is not an option. Everything is lost if a solid state component goes bad, and/ or burned out - which is what solid state does. It burns out. A hard drive can be recovered, and transfered after a total drive failure. The platters are still there, and you can take it to anyone that specializes in data retrieval, and have it all back in most cases. But I think with solid state you have 0% chance of getting anything back in once it fries out.
I think you're exaggerating. Flash memory is just as persistent as magnetic. One cell dying isn't going to corrupt all the other cells any more than a head crash would trash everything on a hard drive. You'll just have an unrecoverable file. In fact, recovery services do exist for flash memory devices. One such company that services all media is Ontrack. It'll be expensive, of course, but data recovery from dead hard drives isn't exactly cheap, either.
This is absolutely a lie. Modern flash memory has more than 11million rewrite cycles. Including NAND. It would take you over 19 years of straight reading/writing over and over again nonstop to "possible" destory the flash.
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily (but not as heavily as it would have been used had the swap file been on it!).
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily (but not as heavily as it would have been used had the swap file been on it!).
The question is: what constitutes a rewrite cycle?
webmail's calculations was based on a modest 1 write per second (assuming the 11M rewrites is accurate). If you've worn a new flash drive in 3 months...we're looking at FORTY writes per second (constant writing for 3 months). To me this sounds pretty insane and considering you probably weren't writing to it 24/7...we're probably looking at over a hundred writes per second. This to me makes no sense...there's some kind of inefficiency somewhere or your flash drive was a lemon (or the 11M writes is inaccurate info...I'll check whenever I've got time.)
The question is: what constitutes a rewrite cycle?
webmail's calculations was based on a modest 1 write per second (assuming the 11M rewrites is accurate). If you've worn a new flash drive in 3 months...we're looking at FORTY writes per second (constant writing for 3 months). To me this sounds pretty insane and considering you probably weren't writing to it 24/7...we're probably looking at over a hundred writes per second. This to me makes no sense...there's some kind of inefficiency somewhere or your flash drive was a lemon (or the 11M writes is inaccurate info...I'll check whenever I've got time.)
How do you get 40/second? My calculations require only 1.4/second to exhaust the 11 million write cycle budget in three months.
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.
You are talking about using flash drive like a floppy. The hard drive in a computer running a Unix, Unix-like, or NT-based OS is a very different animal.
There are a couple fo things to consider here. First there are a number of flash technologies / proccesses available and they all have differrent durability issues. Last I knew no one was to 11 million writes but then again I haven't really been interested in the technology lately. Even if that is a real number it doesn't really amount to much.
As to a persons usage of a drive and how they can manage to wear out a flash drive I'd have to point out that a persons usage is only part of the equation. If not configured properly some OS'es will write to the stoarage area everytime a file is accessed even for a read. This is a big killer on Linux systems and I would suspect most other unix systems. Mounting that hardware in the correct manner can help as long as the last access time isn't important to you. That is system level and doesn't take into account user land software that is noted for working a disk hard.
Don't get me wrong I'm the first to sing the praises of flash. The problem is that it does have issues that don't exist on harddrive based hardware. None of these issues would prevent me from buying a 16gig nano when they come out though. It is just a question of buying the technology that has been correctly applied.
(by the way the current nano is very tempting, just wish it had a bit more memory)
Dave
Quote:
Originally posted by Kolchak
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.
Comments
Originally posted by wizard69
It certainly could happen soon but there are still reliability issues to over come, especially when using a Unix derived Operating System. This is not impossible of course as Linux is capable of mounting volumns that do not generate excessive writes to the "disk", just that certain properties of Flash have to be taken into account.
Further Flash is slow so the types of devices it might be implemented in first may not be the laptops we think of today. An ideal implementation platform for flash would be a tablet type device with built in WiFi. WiFi to allow data storage to a more traditional computer. What would really be neat is a tablet that is both a functional computing appliance and a video display device.
Flash offers all sorts of opportunities but I don't see it catching up with disk drive technology anytime soon. At this point it offers a low power advantage andthat is about it.
Thanks
dave
Maybe the instant on stuff is about having enough flash memory to cover the spin-up time for a HD. OK it is slow, but the HD does not start reading for at least 2 or 3 seconds when it is first starting, because it has to get up to speed. As was pointed out here.
Originally posted by kim kap sol
We're not going to see Flash replace HDs...but what we might see are hybrid that keep files or parts of the OS that need rapid access...the HD would mostly be used for data that rarely needs attention.
Flash can be fast when read writing to it is another thing. Of course thechnology moves forward and speed improves either way.
Along your line of thinking it is rumored that Intel is working on using Flash as a Cache for disk drive access. One would still have and need the disk drive but frequently referenced code would sit in a Flash cache. This could help with instant on, but could be a huge win for power usage.
I suspect that Apple was shown a number of interesting Intel technologies related to portables and this is where the big draw for i86 came from. Sure performance is an issue but I see Intel having a much large selection of technologies to implement into a portable. Everything from integrated video to the stuff that isn't talked about in public. Freescale simply wasn't willing to even consider that market.
Dave
Originally posted by Brendon
Maybe the instant on stuff is about having enough flash memory to cover the spin-up time for a HD. OK it is slow, but the HD does not start reading for at least 2 or 3 seconds when it is first starting, because it has to get up to speed. As was pointed out here.
Originally posted by onlooker
The problem with solid state components is that once they go bad for whatever reason your screwed. You have to buy a new one, and I'm pretty sure file recovery is not an option. Everything is lost if a solid state component goes bad, and/ or burned out - which is what solid state does. It burns out. A hard drive can be recovered, and transfered after a total drive failure. The platters are still there, and you can take it to anyone that specializes in data retrieval, and have it all back in most cases. But I think with solid state you have 0% chance of getting anything back in once it fries out.
I think you're exaggerating. Flash memory is just as persistent as magnetic. One cell dying isn't going to corrupt all the other cells any more than a head crash would trash everything on a hard drive. You'll just have an unrecoverable file. In fact, recovery services do exist for flash memory devices. One such company that services all media is Ontrack. It'll be expensive, of course, but data recovery from dead hard drives isn't exactly cheap, either.
The best option is, as always, a good backup.
Originally posted by webmail
This is absolutely a lie. Modern flash memory has more than 11million rewrite cycles. Including NAND. It would take you over 19 years of straight reading/writing over and over again nonstop to "possible" destory the flash.
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily (but not as heavily as it would have been used had the swap file been on it!).
Originally posted by CharlesS
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily (but not as heavily as it would have been used had the swap file been on it!).
The question is: what constitutes a rewrite cycle?
webmail's calculations was based on a modest 1 write per second (assuming the 11M rewrites is accurate). If you've worn a new flash drive in 3 months...we're looking at FORTY writes per second (constant writing for 3 months). To me this sounds pretty insane and considering you probably weren't writing to it 24/7...we're probably looking at over a hundred writes per second. This to me makes no sense...there's some kind of inefficiency somewhere or your flash drive was a lemon (or the 11M writes is inaccurate info...I'll check whenever I've got time.)
Originally posted by kim kap sol
The question is: what constitutes a rewrite cycle?
webmail's calculations was based on a modest 1 write per second (assuming the 11M rewrites is accurate). If you've worn a new flash drive in 3 months...we're looking at FORTY writes per second (constant writing for 3 months). To me this sounds pretty insane and considering you probably weren't writing to it 24/7...we're probably looking at over a hundred writes per second. This to me makes no sense...there's some kind of inefficiency somewhere or your flash drive was a lemon (or the 11M writes is inaccurate info...I'll check whenever I've got time.)
How do you get 40/second? My calculations require only 1.4/second to exhaust the 11 million write cycle budget in three months.
Originally posted by CharlesS
Last year I managed to completely wear out a brand-new flash drive after only about 3 months by using it heavily
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.
Originally posted by Kolchak
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.
You are talking about using flash drive like a floppy. The hard drive in a computer running a Unix, Unix-like, or NT-based OS is a very different animal.
As to a persons usage of a drive and how they can manage to wear out a flash drive I'd have to point out that a persons usage is only part of the equation. If not configured properly some OS'es will write to the stoarage area everytime a file is accessed even for a read. This is a big killer on Linux systems and I would suspect most other unix systems. Mounting that hardware in the correct manner can help as long as the last access time isn't important to you. That is system level and doesn't take into account user land software that is noted for working a disk hard.
Don't get me wrong I'm the first to sing the praises of flash. The problem is that it does have issues that don't exist on harddrive based hardware. None of these issues would prevent me from buying a 16gig nano when they come out though. It is just a question of buying the technology that has been correctly applied.
(by the way the current nano is very tempting, just wish it had a bit more memory)
Dave
Originally posted by Kolchak
I take it you're a semiconductor engineer with the proper test equipment to tell us exactly what the mode of failure was? If not, then you're just making an assumption about why it failed. It could have been caused by any number of things, from a manufacturing flaw to electrostatic damage. One failure proves absolutely nothing. It's an anecdote, nothing else. I'm sure others can point out pen drives they've "heavily used" (hard to compare considering you never told us what "heavy" was) for much longer are still working fine.
Even accepting the lower value of 100,000 writes to failure, that's over 1,000 writes per day. Unless you're a test technician, I can't see how you could do that.