How accurate is ssdlife?

It says I've written 644 gb in 467 hrs which seems insane since this notebook is just used for work, browsing, music, etc.

 

I'm more interested in the fact how a ssd can actually stop working since there are no moving parts.
All these software programs that indicate your ssd life seem like a sign of planned obsolescence ...

 

iirc, ssdlife doesn't do jack but read info from SSD. Crystaldiskinfo do the same thing I believe.

 

If I recall correctly, for 34nm MLC NAND was rated to be good for up to 5,000-10,000 write cycles. Some 25nm NAND for just 3,000-5,000 cycles. In practice, some chips can go much higher than that though.

 

So is there a list of technologies and/or drives which supposedly will last longer?

 

Intresting, though those numbers dont give me any clue xD If I recall correctly ssd's should perform longer than hdd's?
I'm sure much depends on the brand, and that for example one cant go wrong with intel's current ssd's, and that ocz drives with the sandforce controller should be avoided (but now intel is going to use that controller too, so by the time I get a ssd, the ssd world will have changed already, but i'll start a new thread by then, or try to find an up to date one).

 

It doesn't matter how many P/E cycles any particular NAND cell is capable of, because the controller/firmware counts the number of P/E cycles which have been made so far and artificially retires the NAND after it hits a hardcoded limit. Usually, the limit is set to whatever number IMFT or Samsung/Toshiba gives out.

A firmware maker could easily raise the hardcoded limit beyond the NAND is rated for, but that would be the SSD equivalent of a factory overclock.

Perhaps one day controller technology will be sophisticated enough to perform integrity testing on the drive's NAND and retire it dynamically, only when it reaches its actual end of life...

In this context, the SSD controller/brand is irrelevant. Assuming an identical (and incompressible) workload, an Intel 320 will reach its end of life at the exact same time as a Vertex 3.

The failures you're thinking about have nothing to do with this. Those failures are drive-level failures, which can happen at any time and might not happen at all, whereas these failures are chip-level failures, and are guaranteed to happen after a certain number of writes.

I guess one analogy would be the difference between a car running out of gas vs getting a flat tire. Both have the same end result in that the car can no longer be driven, but the causes are as different as night and day. What we're talking about is the SSD "running out of gas" (inevitable, but predictable) whereas what you're thinking of is the SSD "getting a flat tire" (which could happen randomly, or not at all).

 

Don't forget that Intel uses 10% minimum over-provisioning in the 320 vs. 7% min in the X25-M to account for this as well.

 

Does SSDlife really matter?



Even after hundreds of TB of writes, SSDs still going strong. Worst case in 2-3 years when your SSD wears out, SSDs will be dirt cheap...voila problem solves itself.

 

As far as I know, bad blocks in SSDs are retired dynamically. The disk is rendered read-only when there aren't any spare blocks left. Or at least, I recall a few small (64GB) Indilinx Barefoot SSDs that people posted about. After the SSD's health meter went to zero (as reported by S.M.A.R.T.), it just started again from 100%.

 

well, not all of that is "yours", 830 loves to move data around with its gc and does 3-4 times original writes.

If we say notebook was used for 12 hrs a day every day, that would be 39 days of 16gb a day, or about 4 gb users and host writes to it and rest 830s gcs, which seems highly plausible.

 

SSDLife says I have worn out the 128gb SSD in my desktop by 2% in the past 9 months I've owned it. Expected lifetime based on MWI is until 2050, and it should last well beyond that according to the graph above. It will be obsolete in a fraction of that time.

I would not worry one bit about keeping everything on the SSD unless you find yourself running out of space.

 

26176 GB written in 6342 hours = 4.1GB written every hour. Probably more than most users.

 

And I'll bet you won't be using that SSD in 5 years, just because it'll be really old tech by then and you'll want new stuff.