[udaman]

While magnetic storage indeed has a long term future in some segments, laptop drives are probably going to go all SSD within the next 5-10yrs, I predict.

http://www.infoworld...02571A2004B5DDE

"Seagate's patent resolves this problem by having a reservoir inside the disk casing that contains nanotube-based lubricant. Some of this is periodically pumped out as a vapor and deposited on the surface of the disk thus replenishing the evaporated lubricant. The vapor deposition process is similar to that used in the production of CDs and DVDs.

Seagate anticipates that the new technology could increase disk capacity by a factor of ten, making possible a 600GB 1.8-inch drive, a 1.46TB 2.5-inch drive, and 7.5TB Barracuda 3.5-inch drive. The lubricant reservoirs will be built to last the life of the disk."

Tom's Hardware has reviewed the 32GB Samsung, and while this is an early version (an SATA interface 64GB model use the Samsung chips, is supposed to be available this month from PQI) is effectively limited to a certain degree by the ATA-66 interface, still it looks promising...when the price comes down (currently over $1000, at retail level).

Tom's Hard(ware) review here:

http://www.tomshardw...ive_obsoletism/

There is a lovely discussion about NAND flash longivity/reliability in their forums, take a gander [cause I know we have posters here who take the same negative view, staunchly supporting HD's, claiming HD's usually don't catastrophically fail (blinders on, it happens more often that you think), but it is true that a laboring HD can usually have it's data recovered before it dies completely...what, you don't have a current backup, shame on you---uhh, like me ;0 ].

http://forumz.tomsha...pict201400.html

We can only hope the 128GB versions based on the recently announced Samsung CTP will be less expensive, and significantly faster; if and when they come out sometime next year?

[K15]

Should be interesting. Who will come out with the biggest and cheapest? That's the question. I wouldn't mind 1.5TB in my laptop

[Kennyshin]

K15, on Oct 24 2006, 09:35 PM, said:

Should be interesting. Who will come out with the biggest and cheapest? That's the question. I wouldn't mind 1.5TB in my laptop

Negroponte's US$130 laptops are based on 500MB flash, not HDD. Their Linux OS occupies just from 130MB to 250MB. That's 0.0005TB. Quanta in Taiwan makes the notebooks. AMD provides the 500MHz processors. AMD probably also provides the 500MB (or 512MB?) flash. I don't know exactly why they can't make it under US$100. Maybe the LCD costs too much yet.

Future Samsung SSD drives will get bigger than 1TB around 2010. Rumor is that Samsung employees used 32GB SSD drives in their notebooks (for OS) even a few years ago. It's only a matter of how soon until 20nm manufacturing processes will start. Intel's next-generation quad-core processors will come out of 45nm processes (the current Kentsfield's are based on 65nm processes like dual-core Conroe's.)

Both AMD and Intel are still making flash chips but their profits are low. And then there are several Japanese giants and some small Taiwanese makers.

[K15]

I don't dispute the fact that flash is an awesome technology for things that need low power, ruggedness and small size. None of which concern a typical PC. Point is, in 2010 PC hard drives will be much larger than 1TB.
But by 2010 I wouldn't be surprised at all if laptop hard drives are nearly extinct. True, too that when hard drives fail, almost all of the time, the data is still "there", the drive just can't read it.
But I'm not trying to be negative, I'm excited for flash too, it just has to become a cheaper, larger, faster technology than hard drives before it has any big chance.

[Kennyshin]

K15, on Nov 4 2006, 01:01 PM, said:

I don't dispute the fact that flash is an awesome technology for things that need low power, ruggedness and small size. None of which concern a typical PC. Point is, in 2010 PC hard drives will be much larger than 1TB.
But by 2010 I wouldn't be surprised at all if laptop hard drives are nearly extinct. True, too that when hard drives fail, almost all of the time, the data is still "there", the drive just can't read it.
But I'm not trying to be negative, I'm excited for flash too, it just has to become a cheaper, larger, faster technology than hard drives before it has any big chance.

An interesting news released three days ago in South Korea.

http://www.donga.com...?n=200611020025

Picture: http://www.donga.com...00611020025.jpg

Samsung delveoped 0.72Gb (Gigabit) NAND-based MCP in 2003. They increased the size to 3.2Gb in 2004, 11Gb in 2005, and now 128Gbps in 2006. 0.72Gb is 0.09GB. 128Gbps is 16GB. It's 90MB to 16GB in just three yeas. What is exciting is the speed of growth.

Its physical size is: 12mm x 16mm x 1.4mm. Indeed, it's small enough for some of the smallest MP3/camcorder mobile phones like the 6.9mm Samsung phone recently released.

It's much easier to have multiple "SSD" drives in a PC, nearly any PC. Space, power consumption, weight, all negligibly low. The leading HDD makers expect to have 750GB HDD for notebooks. It won't be easy to have four * 2.5-inch 750GB for RAID 0 in a single notebook.

Cost is one of the biggest concerns among the most prospective consumers right now. However, the only major reason for the seemingly very high cost of 32GB and larger SSD's is because there are very few manufacturers, if any, producing that many SSD drives. A few Taiwanese, a few South Korean, some more in other countries, but none of them targets the mass market. Samsung started selling SSD-based notebooks in its domestic South Korean market since at least June 2006, but Samsung never sells anything in the domestic market at a low price point because Samsung virtually monopolizes their home market so I haven't seen anyone who actually bought one. It will be Intel to make the technology first available to the mass. Maybe Intel can someday incorporate 1TB NAND right inside a 32-core processor. Meanwhile, some of the bolder manufacturers will certainly start selling 128GB 1.8-inch or 2.5-inch form factor SSD for under US$1,000, 64GB version for under US$500, 32GB version for under US$200, and so on. Currently, even a 32GB SSD costs around US$1,000. A South Korean company plans to sell a 4GB 2.5-inch SSD for US$150 and a 32GB 1.8-inch SSD for US$1,000.

[hyc]

Kennyshin, on Nov 5 2006, 12:33 AM, said:

Meanwhile, some of the bolder manufacturers will certainly start selling 128GB 1.8-inch or 2.5-inch form factor SSD for under US$1,000, 64GB version for under US$500, 32GB version for under US$200, and so on. Currently, even a 32GB SSD costs around US$1,000. A South Korean company plans to sell a 4GB 2.5-inch SSD for US$150 and a 32GB 1.8-inch SSD for US$1,000.

With 8GB CF cards at only US$140, anybody planning to sell a 4GB SSD for $150 is either a fool or just plain greedy. Anybody who buys that 4GB SSD is just a fool. A 32GB SSD should retail around $500 today, there's just nothing special about the technology or R&D to justify any higher price.

[imsabbel]

hyc, on Nov 5 2006, 11:34 PM, said:

Kennyshin, on Nov 5 2006, 12:33 AM, said:

Meanwhile, some of the bolder manufacturers will certainly start selling 128GB 1.8-inch or 2.5-inch form factor SSD for under US$1,000, 64GB version for under US$500, 32GB version for under US$200, and so on. Currently, even a 32GB SSD costs around US$1,000. A South Korean company plans to sell a 4GB 2.5-inch SSD for US$150 and a 32GB 1.8-inch SSD for US$1,000.

With 8GB CF cards at only US$140, anybody planning to sell a 4GB SSD for $150 is either a fool or just plain greedy. Anybody who buys that 4GB SSD is just a fool. A 32GB SSD should retail around $500 today, there's just nothing special about the technology or R&D to justify any higher price.

How do you come to that conclusion?
140$ per 8Gbyte is the slowest, cheapest, least realiable crap you can get on the spot market.

For a SSD, you dont want <6Mbyte/s transfer rates, nor do you want non-existing failure-rate tests.

A 100% markup is more than in line if it actually delivers as a SSD.

[hyc]

imsabbel, on Nov 6 2006, 03:39 PM, said:

How do you come to that conclusion?
140$ per 8Gbyte is the slowest, cheapest, least realiable crap you can get on the spot market.

For a SSD, you dont want <6Mbyte/s transfer rates, nor do you want non-existing failure-rate tests.

A 100% markup is more than in line if it actually delivers as a SSD.

Actually 20MB/sec, which is pretty respectable for Flash, and that's a retail price, not spot price. (See the Newegg link I already posted above.) In other words it already has at least a 100% markup. If you're happy paying twice again what the technology is worth, go ahead. Not me...

[continuum]

20MB/sec is pretty good, but if Tom's tests are representative of a typical sample, then 50MB/sec is on the high end of flash performance today, and is very, very good.

Albiet pricey.

[hyc]

continuum, on Nov 6 2006, 09:44 PM, said:

20MB/sec is pretty good, but if Tom's tests are representative of a typical sample, then 50MB/sec is on the high end of flash performance today, and is very, very good.

Albiet pricey.

Flash memory speed is easily scalable, just gang more cells together in parallel. Using standard 512 byte sectors, 8 in parallel would let you write a 4096 byte block in the same time as a single sector; so 20MB/sec becomes 160MB/sec. Since most filesystems today write in 4K or larger pages anyway, this is already very practical. There's nothing difficult here, all it takes is for one vendor to decide to do it, and all of the other vendors will be forced to follow suit or be left in the dust.