" $1000 next year.
$500 in two years.
$250 in three years.
$125 in four years.. getting VERY interesting here...
$63 in five years."
This assumes that 32 Gig wont be considered some trivial amount of storage in 5 years from now.
It assumes also that flash will continue to get cheaper at the rate of 100% per year. Which it might.
You must have a different experience witht he stock market to me.
When ever I see companies remove themselves from the red tape of governement their stock price goes up.
Re:Scoffing Posts Are From Those With Sort/No Memo
on
Hard Drive Memory Lane
·
· Score: 1
Ram has worsened, relative to hard disks in the last 14 years. Thats because hard disks have improved about 105% each year in the megs per dollar figure but RAM has only improved about 70% each year. This has meant RAM has fallen way behind.
You use of the "human shredder" example is commical.
It is very likely that it never happened and the story is actually part of the propaganda that you think is not a proplem. The story itself emerged just a few days before the invasion, the day of the Iraq debate in the House of Commons. http://www.google.com.au/search?hl=en&q=saddam++ industrial+shredders+false&btnG=Google+Search&meta =
You are part of the problem that the story is talking about.
What? A mention of the French on the front page of Slashdot and there isn't 10 posts all making the same "cheese eating surrender monkeys" joke. Half of which have been moderated up to +5 funny.
Maybe those people are too busy out eating their Freedom Fries.
What you say is perfectly correct but come next year this clown will claim he was right because "Google continued to roll out new products and services".
Thanks for the correction. I'll be fixing that up shortly. The change doesnt alter the fact that if this rate of improvement continues then flash will actually pass disk in performance way before 11 years. Which is what I was trying to show.
You are entirely correct that flash will eat away the disk market starting at the areas where disk offers the worst megs per $.
The reason I chose 3.5" IDE drives for my survey was that they offer the very best $ per meg you can get. If flash surpases these disks then all disks are pretty much gone. I was really trying to find out roughly when we would be saying bye to disk drives.
I realise that you were making a joke but trends holding true for decades is not the unusual for the computer industry. If that wasnt the case then I wouldn't have bothered with all the research. That said, new developments could vary the timeframe dramatically either way or even render flash obsolete.
Thanks for the additional information on process size, I'll use to guide my thoughts when I make further comments on my site. I will point out that shrinking the process may not be the only way to improve $ per meg figure for flash. The 10,000 fold increase in demand (thats just an guess of how extra megs need to be produced to replace hard disks) may help drive costs down signigicantly. The simple repetive layout of flash may also lead to improvements that are not avalable other types of silicon based equipement.
"Meanwhile, all of this assumes that the hard disk people do nothing surprising in 11 years."
I think a suprising thing is happening right now. Something that could make my prediction look conservative. Kryder's law seems to be broken. In a big way. Time will tell.
"doesn't flash burn out much faster if it's being constantly written to? "
This is the first time I have ever heard this idea. If you have a source, I am keen to learn from it. It will modify my opinion.
I think that people have a poor perspective on just how long a flash drive can last with wear leveling.
Taking a the example of a theoretical hard disk replacement flash drive of 200GB with a write speed of 40 Megabytes per second and doing some basic calculations shows that it could be written to continuously for just over 15 years before every block passed the 100 000 write mark. The equivalent of todays 200GB drive some 15 years ago was the 210MB disk. There are not many machines running today with 210MB hard drives, let alone dong the kind of work that requires continuous writing to the disk. And 100 000 writes is often considered a minimum. Average failure figures are often quoted as 1 million writes. Worrying about it wearing out just isnt worth the time.
" I doubt I'll be an early adopter for the first wave of flash notebooks"
Wise advice for any technological advancement.
I have looked into the performance trends of flash vs hard disk and I suspect, in a just few years, that flash will catch up to the point that there is no meaningful difference in the sustained read and write speeds between the two. This means when the $ per meg is the same, there will be no logical reason to chose disk over flash, including in a hybrid setup. Could be a few years yet for desktops, 11 years for 3.5" IDE drives according to my study. Probably a lot sooner for 2.5" drives.
Nice rant but you totally mised the point.
A 300 Gig IDE drive doesnt fit in a laptop.
A 300 Gig IDE drive uses loads of power.
A 300 Gig IDE drive has faster sustained transfer speed but much a longer access times than flash. Horses for courses.
Wear leveling algorithms can make the write limit of flash irrelevant.
That the interface (eg, ATA) for accessing storage media usually goes out of date before the media wears out is true for both disks and flash.
The real story here is that flash is trouncing disk in improvements in Megs per $ and will one day catch up to and overtake disk. And it will be sooner than mmost people expect.
Today, you can buy a 256MB USB Watch without much trouble at all. I dont see any issue with these novelty USB drives being available in reasonable sizes at the time of pruchase.
The real problem is that the sweet spot for USB drives have acheived an improvment about 80% each year. The means your USB watch seems out of date in only 2 years.
Betting on Moores law is a smart idea. MicroSoft has done it for more than a decade and won every time.
The Megabytes per $ for flash has averaged about 130% improvment each year over the last 5 years. There is no reason to think this year will be any different.
If you are saying that you can only afford a 128 MB flash drive right now, then it will only be about 10 months before you can afford a 256 MB one.
I agree, "they" would have to witness you leave without voting and then be motivated enough to organise getting you charged. In practice, it never happens.
Actually, by the letter of the law, voting is compulsary.
Interestingly, advising people to vote "informally" is also illegal in Australia.
That isnt saying much as 99.9% of iPod use in Autralia is also illegal and no-one seems to have been arrested for that either.
Voting in Australia is compulsary. Thats why they have barbecues, to keep people (and their kids) happy and entertained while they perform their civic duty.
"There are 4 considerations when looking at a storage medium"
I think power usage will also be a big consideration going foreward.
"Hard drives have probably increased in capacity faster over the long term"
I know that you have said found my site http://www.mattscomputertrends.com/ but I am not sure if you read the capacity trend info. The hard disk capacity trend is about 104% each year but flash comes in at 130%. Feel free to dispute the data, though.
" I'd predict this to happen in about 2013."
I know you retracted your paragraph 4 but I am pretty impressed that some one could come up with such a close guess to mine on what I assume is just off the top your head calculations.
"One minor point I'd like to get back to is the low number of writes allowed to flash RAM devices." Wear leveling algorithms just move the high traffic data (like swap files) to lesser used areas of the flash as needed, evening out the wear across the whole flash storage device over time. These are yet to implemented in major operating systems but will probably be quickly implemented when flash as primary storage becomes viable. Wikipedia has good references about these file systems.
"However, I actually expect hard drive capacities to slow their rate of increase, so it could happen earlier." AND "Hard drives are doubling in capacity at a rate of probably more than 18 months." I think these two statements show you have noticed something that I thought noone else had discovered. Kryder's law is dead. Kryders law is moores law for hard disk capacities. It says they should double every 13 months. The other way of saying this is a 1000 fold increase in capacity every 10.5 years. The second way evens out short term bumps and dips. In March when I am due to do the 2006 figures on my site, there will be solid evidence that the trends are way behind Kryders law. I'll do a story about it and submit it here (where it will be rejected, haha).
Some points I'd like to add. I chose 32Gig SD as my example because they are expected in 2006, and the/. post is about predictions for 2006. http://www.pcworld.com/news/article/0,aid,122514,0 0.asp
I chose SD because it is a lot smaller than CF. take these two factors into account and flash starts to get a serious size and weight advantage.
The best sustained (read or write) hard disk performance of IDE drives is 60 MegaBytes per second according to a recent article on Toms Hardware. The best flash drive, as tested by Toms, had a sustained performance of 23 Megabytes per second write speed with 30MB per second read performance. That means you only need a 3 wide stripe to get the same through-put as a disk. The real advantage of flash is in seek times which are sub 1 milisecond.
I think the power requirements of an idle flash drive would be significantly less than any hard disk no matter how many "cards" in the array. They dont use power when they are not reading or writing. The power usage when reading or writing might be significantly more, as you calculated.
Obviously, we are discussing a theoretical product due to price considerations. I have pointed out in other posts that it could be 11 years before flash catches 3 1/2" IDE disk in the $ per meg area. I think there siginificant advantages to flash that will mean the transition will occur when the price is right.
I have seen another test that shows I/O operations per second, the type where SCSI drives usually trounce IDE drives. Flash blew away all types of hard disks in this type of tests purely because of the huge seek speed advantage that flash has.
Slashdot Mirror
" $1000 next year. .. getting VERY interesting here...
$500 in two years.
$250 in three years.
$125 in four years
$63 in five years."
This assumes that 32 Gig wont be considered some trivial amount of storage in 5 years from now.
It assumes also that flash will continue to get cheaper at the rate of 100% per year. Which it might.
Your overall rule seems to be this:
People should change the way they work to fit the tool that is provided instead of changing the tool to fit way people work.
Classic centralised IT Department thinking.
" Consumers don't purchase something they don't want or need."
Your whole arguement seems based around this point.
Last time I checked, you dont purchase Linux. It's Free.
The rest of your arguement may hold up but the analogies seem flawed.
The killer app is the same as it is always has been since the invention of the internet.
Pornography.
You must have a different experience witht he stock market to me.
When ever I see companies remove themselves from the red tape of governement their stock price goes up.
Ram has worsened, relative to hard disks in the last 14 years. Thats because hard disks have improved about 105% each year in the megs per dollar figure but RAM has only improved about 70% each year. This has meant RAM has fallen way behind.
Figures come from my research available on my site:
http://www.mattscomputertrends.com/
You use of the "human shredder" example is commical.+ industrial+shredders+false&btnG=Google+Search&meta =
It is very likely that it never happened and the story is actually part of the propaganda that you think is not a proplem. The story itself emerged just a few days before the invasion, the day of the Iraq debate in the House of Commons.
http://www.google.com.au/search?hl=en&q=saddam+
You are part of the problem that the story is talking about.
What? A mention of the French on the front page of Slashdot and there isn't 10 posts all making the same "cheese eating surrender monkeys" joke. Half of which have been moderated up to +5 funny.
Maybe those people are too busy out eating their Freedom Fries.
"But didn't Bill Gates once say that 640k of memory should be more than enough for anyone? :-)"
Yes, he didnt say that.
What you say is perfectly correct but come next year this clown will claim he was right because "Google continued to roll out new products and services".
Thanks for the correction. I'll be fixing that up shortly. The change doesnt alter the fact that if this rate of improvement continues then flash will actually pass disk in performance way before 11 years. Which is what I was trying to show.
You are entirely correct that flash will eat away the disk market starting at the areas where disk offers the worst megs per $.
The reason I chose 3.5" IDE drives for my survey was that they offer the very best $ per meg you can get. If flash surpases these disks then all disks are pretty much gone. I was really trying to find out roughly when we would be saying bye to disk drives.
I realise that you were making a joke but trends holding true for decades is not the unusual for the computer industry. If that wasnt the case then I wouldn't have bothered with all the research. That said, new developments could vary the timeframe dramatically either way or even render flash obsolete.
Thanks for the additional information on process size, I'll use to guide my thoughts when I make further comments on my site. I will point out that shrinking the process may not be the only way to improve $ per meg figure for flash. The 10,000 fold increase in demand (thats just an guess of how extra megs need to be produced to replace hard disks) may help drive costs down signigicantly. The simple repetive layout of flash may also lead to improvements that are not avalable other types of silicon based equipement.
"Meanwhile, all of this assumes that the hard disk people do nothing surprising in 11 years."
I think a suprising thing is happening right now. Something that could make my prediction look conservative. Kryder's law seems to be broken. In a big way. Time will tell.
"doesn't flash burn out much faster if it's being constantly written to? "
This is the first time I have ever heard this idea. If you have a source, I am keen to learn from it. It will modify my opinion.
I think that people have a poor perspective on just how long a flash drive can last with wear leveling.
Taking a the example of a theoretical hard disk replacement flash drive of 200GB with a write speed of 40 Megabytes per second and doing some basic calculations shows that it could be written to continuously for just over 15 years before every block passed the 100 000 write mark. The equivalent of todays 200GB drive some 15 years ago was the 210MB disk. There are not many machines running today with 210MB hard drives, let alone dong the kind of work that requires continuous writing to the disk. And 100 000 writes is often considered a minimum. Average failure figures are often quoted as 1 million writes. Worrying about it wearing out just isnt worth the time.
" I doubt I'll be an early adopter for the first wave of flash notebooks"
Wise advice for any technological advancement.
I have looked into the performance trends of flash vs hard disk and I suspect, in a just few years, that flash will catch up to the point that there is no meaningful difference in the sustained read and write speeds between the two. This means when the $ per meg is the same, there will be no logical reason to chose disk over flash, including in a hybrid setup. Could be a few years yet for desktops, 11 years for 3.5" IDE drives according to my study. Probably a lot sooner for 2.5" drives.
k .html
I've linked to it elsewhere here but here is the link to my study again:
http://www.mattscomputertrends.com/flashvsharddis
Nice rant but you totally mised the point.
A 300 Gig IDE drive doesnt fit in a laptop.
A 300 Gig IDE drive uses loads of power.
A 300 Gig IDE drive has faster sustained transfer speed but much a longer access times than flash. Horses for courses.
Wear leveling algorithms can make the write limit of flash irrelevant.
That the interface (eg, ATA) for accessing storage media usually goes out of date before the media wears out is true for both disks and flash.
The real story here is that flash is trouncing disk in improvements in Megs per $ and will one day catch up to and overtake disk. And it will be sooner than mmost people expect.
I did a study which estimates that flash will surpass 3.5 inch IDEs in every price by 2017.k .html
Read about it here:
http://www.mattscomputertrends.com/flashvsharddis
Today, you can buy a 256MB USB Watch without much trouble at all. I dont see any issue with these novelty USB drives being available in reasonable sizes at the time of pruchase.
The real problem is that the sweet spot for USB drives have acheived an improvment about 80% each year. The means your USB watch seems out of date in only 2 years.
Betting on Moores law is a smart idea. MicroSoft has done it for more than a decade and won every time.
The Megabytes per $ for flash has averaged about 130% improvment each year over the last 5 years. There is no reason to think this year will be any different.
If you are saying that you can only afford a 128 MB flash drive right now, then it will only be about 10 months before you can afford a 256 MB one.
I agree, "they" would have to witness you leave without voting and then be motivated enough to organise getting you charged. In practice, it never happens.
Actually, by the letter of the law, voting is compulsary.
Interestingly, advising people to vote "informally" is also illegal in Australia.
That isnt saying much as 99.9% of iPod use in Autralia is also illegal and no-one seems to have been arrested for that either.
Voting in Australia is compulsary. Thats why they have barbecues, to keep people (and their kids) happy and entertained while they perform their civic duty.
32. 'Restaurant' is the most mis-spelled word in search engines.
There is a lot of money to be made if you could get the top list of mispelled words in search engines.
"There are 4 considerations when looking at a storage medium"
I think power usage will also be a big consideration going foreward.
"Hard drives have probably increased in capacity faster over the long term"
I know that you have said found my site http://www.mattscomputertrends.com/ but I am not sure if you read the capacity trend info. The hard disk capacity trend is about 104% each year but flash comes in at 130%. Feel free to dispute the data, though.
" I'd predict this to happen in about 2013."
I know you retracted your paragraph 4 but I am pretty impressed that some one could come up with such a close guess to mine on what I assume is just off the top your head calculations.
"One minor point I'd like to get back to is the low number of writes allowed to flash RAM devices."
Wear leveling algorithms just move the high traffic data (like swap files) to lesser used areas of the flash as needed, evening out the wear across the whole flash storage device over time. These are yet to implemented in major operating systems but will probably be quickly implemented when flash as primary storage becomes viable. Wikipedia has good references about these file systems.
"However, I actually expect hard drive capacities to slow their rate of increase, so it could happen earlier." AND "Hard drives are doubling in capacity at a rate of probably more than 18 months." I think these two statements show you have noticed something that I thought noone else had discovered. Kryder's law is dead. Kryders law is moores law for hard disk capacities. It says they should double every 13 months. The other way of saying this is a 1000 fold increase in capacity every 10.5 years. The second way evens out short term bumps and dips. In March when I am due to do the 2006 figures on my site, there will be solid evidence that the trends are way behind Kryders law. I'll do a story about it and submit it here (where it will be rejected, haha).
Wow, what a well reasoned responce.
/. post is about predictions for 2006.0 0.asp
I chose SD because it is a lot smaller than CF. take these two factors into account and flash starts to get a serious size and weight advantage.
Some points I'd like to add. I chose 32Gig SD as my example because they are expected in 2006, and the
http://www.pcworld.com/news/article/0,aid,122514,
The best sustained (read or write) hard disk performance of IDE drives is 60 MegaBytes per second according to a recent article on Toms Hardware. The best flash drive, as tested by Toms, had a sustained performance of 23 Megabytes per second write speed with 30MB per second read performance. That means you only need a 3 wide stripe to get the same through-put as a disk. The real advantage of flash is in seek times which are sub 1 milisecond.
I think the power requirements of an idle flash drive would be significantly less than any hard disk no matter how many "cards" in the array. They dont use power when they are not reading or writing. The power usage when reading or writing might be significantly more, as you calculated.
Obviously, we are discussing a theoretical product due to price considerations. I have pointed out in other posts that it could be 11 years before flash catches 3 1/2" IDE disk in the $ per meg area. I think there siginificant advantages to flash that will mean the transition will occur when the price is right.
Thanks again for a well reasoned reply.
"3. No seek time (I think, not sure about this)."
n d_functional_usb_flash_drives/page9.html
shows that seeks times can come in at less than milisecond.
You can be certain. This page:
http://www.tomshardware.com/2005/08/10/two_fast_a
I have seen another test that shows I/O operations per second, the type where SCSI drives usually trounce IDE drives. Flash blew away all types of hard disks in this type of tests purely because of the huge seek speed advantage that flash has.