i can't possibly make enough drool to last me until i can afford one of these. and an eee to stick it in. *drool*

oh my god!

Sorry, I'm looking for one that will last 163 years, not a paltry 162.76041666666666666666666666667!

Drool also.

I love where this stuff is going.

I am really liking these SSD's. My only question is what kinda of cooling do they need? If any? No moving parts, but electrons moving really fast can cause some heat to build up.

These drives can save so much space in a case. More room for the massive video cards that are coming out these days!

Oh and drool...
as well.

114.079553 years is great news since I plan on putting my brain into a robot body... maybe even an Adrian Barbobot body.

what a load of bull. solid state memory has 1/10 the read/write life span of hard drives. What are they basing this lifespan on, how long it can be plugged in, but not used? over a 100 year life span, reallly? So somehow they not only matched the life span of regular hard drives, but blew them out of the water? Or are they just using one set of numbers and promoting them as another set (hope that came across clear enough). SSD's will never be a good alternative for buisness computers because of the failure rate and corruption rate. They may also never be good for home computers for that exact same reason. But hay, they use a lot less power and are fast, but it's a compromise. Just my 2 cents.

I think the important statistic for solid state drives is how many erase cycles it will support (i.e. rewrites).

Can you say "iPhone: now with 300,000 songs in your phone"

This is going to replace the HD in my MacMini, if it is that good.
On the lifespan side: whether they claim it to be a millennia or a thousand years, we have no actual way to check it out. Unless they indeed declare what that wild speculation is based upon.

Ummm... SSD..

One thing is for sure; the kids will have their say.

Ah...but think about it for a moment.
NO MOVING PARTS!!!!
Gimme gimme! Pleeeeze gimme what's in that box!!!

me wishes I was born 50 years later

@Geeum:

And I thought I was the only one :D

@jkr: Yah, those ignorant assholes at TOSHIBA. What the hell does TOSHIBA know about electronics?

@Pope John Peeps II: [en.wikipedia.org]
specifically the write cycles. However, after reading the wiki, I didn't know about the high endurance Flash, which is comparable to hard drives. So the next question is how many read/write cycles can SSD's handle? What ever the case, 114 years is not even rational by any account, and thus must not be assumed to be correct. Unless of course it is a new tech, which then should be the main focus of the article (which I don't believe to be the case).

@jkr: Well, if you check out the Toshiba press release, it's all there.

[www.toshiba.co.jp]

The operating life is 1,000,000 hours*1.

*1: The MTTF (Mean Time to Failure) is not a guarantee or estimate of product life; it is a statistical value related to mean failure rates for a large number of products which may not accurately reflect actual operation. Actual operating life of the product may not resemble the MTTF.

@jkr:

Even if they only last 50 years MAX who uses a hard drive for more than 20 years MAX? Schools buy new computers every so often and people upgrade their own computers at least every 5 years so I don't see why you're making such a big deal about a number that could be off by 50 years and still not make a difference.

They don't mention anything about new tech to extend the life of this product far beyond anything currently available, if thats what you are referring to. Additionally, there is no mention of read/write cycle life, which is odd. Most permanent flash memory systems are rated w/ read/write cycles, like the Disk On Chip (DOC) we all know and love. Guess we'll just have to wait for the white papers.

@cellshadedninja: What I'm saying is that the number is bull to begin with, and thus doesn't tell us anything about the actual life span. 114 years is about a magnitude greater than the life span of most hard drives, especially since flash memory is not known for it's exception reliability. It's not just a moving parts vs. non moving parts debate. Just look at plasma displays vs CRT's, CRT's typically have twice the rated life (this is of course generalizing).

@jkr:
Ever heard of TrueFFS? That's your answer for how the lifespan is so long despite the limited write cycle (there's no limit to reads on Flash memory, you probably mean write/erase cycles as are commonly referred to with Flash). You can read more about it here if you don't know how to use Google, as it appears: [www.iop.org]

Yes, TrueFFS is used on DOC, it is a proprietary software layer. The general term is wear leveling. There actually is a limit on read cycles, but they are typically 10x or more than that of write cycles, and thusly not the limiting factor in the lifespan. And I definetly wouldn't call flash having a long lifespan (except perhaps the high endurance flash, but that is only comparable to that of a typical hard drive). Though wear leveling can drastically increase the overall lifespan of a flash drive, it can't overcome some of the inherent problems with nand rom, specifically that random blocks go bad much more quickly than a regular HD and thus create corruption. This argument is of course academic at this point, and only time will tell. And btw, loved debating w/ you guys :) . We mods over at the miousers are very familiar with DOC and it's limitations.

There must be an error outta there... 1 million hours are roughly 164 years. Fuck! That's a wide lifespan!

I mean... 114 years. Even my greatgrandsons will enjoy all my pr0n... Wait! No... Jesus, what a shame...

NAND blocks going bad (or are bad) aren't a consideration. All flash translation layers automatically reserve at least 2% of available blocks just for this so data is automatically relocated.

(You can't detect if a block is bad by reading it - you can get bit errors, but those happen in good blocks as well - only writes and erases detect bad blocks. Side advantage is that the relocation doesn't impact data integrity - if you're writing or erasing, you can easily reprogram the data elsewhere.)

Wear leveling really extends the life of NAND (which already has around 1 million guaranteed erase-write cycles, and generally flash has around 10x that number before it really does start to degrade). Reading wearing out the flash chip is a new one - I'd have to call BS since reads are far more common than writes, and they aren't wear leveled. Now, it's possible that in NAND, you can flip blts in the same bit line if you read a page one too many times, but that's the entire point of using ECC codes in NAND. Reads don't wear out the block, they just increase the likelihood of getting a bit-flip (but it still is a pretty rare event. Writing has the far greater chance of causing a bit-flip if you write data to it incorrectly).

And yes, I write flash drivers.

Just FYI - those considering using a CF card as a hard drive - be aware most CF cards do not support DMA or UDMA modes - so you'll be stuck with 100% CPU utilization when you do I/O to the card. SSDs do support DMA and UDMA modes, so they'll be faster.

(And SSDs rock - with zilch seek times, unless you do video editing or anything else that stripes in large blocks of data, you'll see speed improvements even if the throughput is lower - the seek times limit how much I/O you can do)

@Worf: I'm still skeptical. I once worked with a design team using flash memory for program storage and wearing out the flash was a huge concern. A live computer system can flip bits many thousands of times in a single minute. Even if you "wear level" the writes in a driver somewhere, you still risk a short lifespan before going permanently read-only. The only way these are a better choice is if they aren't being intensively used - with mostly read cycles rather than write cycles. If you're just reading email or lightly surfing the web, that's exactly where they work best. Otherwise I just don't see the benefit.

Oh and the MTBF - that was good for laugh. Do they expect us to believe they invented this 162 years ago so they could test it or would it be more accurate to say that they just pulled a number out of their butt...

nand flash has only been available in commercially relevant quantities since 1989. needless to say the performance density and longevity of the components has evolved over time. Consequently you should evaluate your bias with regards to previous generation flash components in light of the changes made since then.

Strictly speaking you should be able to write at rate of 100MB/s for 40.6 years before achieving 1 million write cycles on a 128GB of flash... It is likely that issues other than write cycle lifespan will become a significant factor in the failure of very high capacity random access flash storage.

The architecture of Flash has evolved over the years. Many would be surprised to realize just how long. If you still have an old XT computer and you are using the original BOIS you have been utilizing flash. My XT is still fine and has been running a number of hours since 1984. The EEPROM chip is a type of Flash memory. It has a grid of columns and rows with a cell that has two transistors at each intersection. That is the older technology, but the basic principles have not changed a great deal. Two transistors are separated from each other by a thin oxide layer. One of the transistors is known as a floating gate, and the other one is the control gate. The floating gate's only link to the row, or wordline, is through the control gate. As long as this link is in place, the cell has a value of 1. To change the value to a 0 requires a process called Fowler-Nordheim tunneling. That is far too much to explain here. Google it.

Heat build up is a factor. However, you can run a Core2 Quad and not burst into flames so I believe that heat can be resolved. The materials used in advanced generations of NAND Flash have allowed greater speeds using a more compact transistor relationship. Not erasing one byte at a time like an EEPROM, but erasing whole blocks that are rewritten was and is the leap. Doing it with efficiency, decreasing redundant fragmented storage (yes that still happens using Flash) will yield the desired results.

Keep in mind, Flash does not need an external power source. Flash Ram must be reloaded each time power is cycled to the unit. In most cars, take out your battery and reconnect it. How many presets do you have? Most likely your stereo will have no presets. Using NAND Flash the data is there, cool and ready without reloading. There will always be barriers, but think how long you have been using something with transistors. Oversimplified maybe, but many of those items still operate normally even after 50 years or more.