• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

When SSDs go bad

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Sorry I was about to say how heat seems not to affect drvies when I read further it has mech drives.

Anyway, I have not yet bought an SSD but would like to have one as my boot drive. I've wondered how you manage a failing SSD? So you clone a copy of the image to another drive? At least with the hard drive I've been able to migrate to another disk using a number of tools.
 
Same as a HDD, for backups, really. Any tool that won't copy free space is as good as any other.

Some manufacturers (Samsung included) say that Sleep should be disabled with SSD's or you end up with issues, some people don't have any issues but I guess it could be down to the individual motherboard's implementation of sleep. Do you put your computer to sleep often?
All the time. No issues (but also, no old SF drives). I think Samsung just says that because they can't get Magician to work right with sleep, and a disclaimer is an easy way out (it has a habit of preventing sleep and waking PCs up). Not working with sleep would make an SSD totally unsuitable for any notebook/tablet, and many modern home and office desktops.

IME, Forcewares have been the bane of S3 and S4, much more than SSDs (nothing like a BSOD waking up from sleep, on different Geforce gens and Intel chipset gens, but all the same updated driver version--WHQL is so useful!).
 
Same as a HDD, for backups, really. Any tool that won't copy free space is as good as any other.

All the time. No issues (but also, no old SF drives). I think Samsung just says that because they can't get Magician to work right with sleep, and a disclaimer is an easy way out (it has a habit of preventing sleep and waking PCs up). Not working with sleep would make an SSD totally unsuitable for any notebook/tablet, and many modern home and office desktops.

IME, Forcewares have been the bane of S3 and S4, much more than SSDs (nothing like a BSOD waking up from sleep, on different Geforce gens and Intel chipset gens, but all the same updated driver version--WHQL is so useful!).

Sort of makes one think twice -- thinking to upgrade a laptop with an SSD . . .

I put an Intel Elm Crest in my Mom's [nForce] 610i system around 2011. It's still going strong AFAIK. I had tweaked the system to provide reliable sleep states -- as with three of the five household machines. Recently, I noticed that the system behaves F***ey when it comes out of sleep. It's bad enough that Mom gets confused with Windows 7 and her software.

I ran system file checker "SFC /SCANNOW" on her system, and all is supposedly wonderful. But I pulled her system off its "sleep" regimen until I've had a closer look.

I waited a long time to buy an SSD, except for a 60GB unit I used in ISRT configuration. I think that unit began giving me trouble, but can't confirm if it was the SSD, the BIOS or the IRST software version. I'm testing it today to see if it would be wise to use it some more.

But I also shelled out half a kilo-buck for my Sammy 840-Pro. We were more assured of a ten-year lifespan on these puppies! Are folks reporting here that they're going south in less than 3 years?

This is rather scary.
 
My intel 80 Gb G2 is now over 4 years old and it's mSata version in my laptop about 2.5 years. No issues at all.

At work i have a Samsung 830, no issues as well and parents have an Crucial M4 running happily since about 2 years.

But then I also haven't had any issues with a hard disk lately. 2 of my green drives are also 4 years old and they have been stressed quiet a bit.
 
I have Samsung 830's all over the place.
Old beater laptops, my parents torcher chamber of a a PC, my HTPC and primary gaming rigs.

They just chug along and do their thing.
 
I've gotten into the habit of running a HDTune surface scan on my SSDs every month or so.

In my Foxconn AT-5570 NanoPC, I've had several SSDs "die" on me, due to excessive bad sector buildup. (Surface scan looks like a Christmas tree, with all of the red dots covering the green background squares.)

I finally discovered, it's the heat, silly. The NanoPC is passively cooled, and the CPU generates enough heat in the Aluminum enclosure, that the SSD heats up pretty massively. It causes the cells to fail and lose their charge.

Moreso, when the cells are actually worn down, as was the case with the refurb 240GB OCZ Vertex Plus R2 SSD I was using.

It was a major hassle to recover it the two times it died, but I finally moved it into a desktop, where it won't be subject to such heat, and it has been fine for months.

Likewise, an OCZ Agility 30GB drive, that was slightly used, that had been rock-solid, developed a bad sector once placed into the NanoPC.

Basically, the NanoPC eats SSDs. Sadly.
 
I finally discovered, it's the heat, silly. The NanoPC is passively cooled, and the CPU generates enough heat in the Aluminum enclosure, that the SSD heats up pretty massively. It causes the cells to fail and lose their charge.

Heat extends the life of NAND cells. :colbert:
http://arstechnica.com/science/2012/11/nand-flash-gets-baked-lives-longer/

@john3850, did that, as I wrote.

@npaladin-2000, the SMART data didn't change, except for the writes counter, power on count, and hours. Everything else was as it was when I got it.
Of course, when the unit was dead, then, it was impossible to pull any data from SMART so...
 
But I also shelled out half a kilo-buck for my Sammy 840-Pro. We were more assured of a ten-year lifespan on these puppies!

We were assured of no such thing. If it was supposed to last ten years, Samsung would have sold it with a 10-year warranty, not a 5-year.

There are a multitude of ways for electronics to die. The only guarantees are that it'll die, and that it'll probably be inconvenient.

Are folks reporting here that they're going south in less than 3 years?

This is rather scary.
No it's not. Stuff breaks. Some people have stuff that breaks early, and they complain about it online. Which is fair.

Read up on the bathtub curve.

OP has far too many failures with far too many different models and brands to be legit - either he's got the worst luck on the planet, or something else is eating his SSDs.
 
Sammys didn't fare too well in The Endurance Experiment. Some drives (like my Neutron GTX's) were stellar. As in most areas, you get what you pay for.

Edit: When they talk about normal wear for the average consumer, they're not talking about people who run lots of benchmarks.
 
Last edited:
Heat extends the life of NAND cells. :colbert:
http://arstechnica.com/science/2012/11/nand-flash-gets-baked-lives-longer/

@john3850, did that, as I wrote.

@npaladin-2000, the SMART data didn't change, except for the writes counter, power on count, and hours. Everything else was as it was when I got it.
Of course, when the unit was dead, then, it was impossible to pull any data from SMART so...

While heating NAND cells may help their overall lifespan of operation, it does NOT help them to retain their programming charge.

Read up on the Arrhenius equation.

http://www.sandisk.com/go/preserve
It pays to pay attention to the specifics. Especially when it is a mouse click away.

One, the article tells the reader the mechanism in which the NAND is restored: trapped electrons are freed from their trapped state after applying heat to them. A lot of heat.


You are not going to obtain the temperatures that the NAND was heated to in the article(800 degrees Celsius) in a home computer unless the computer is caught in a housefire, which would then promptly heat it over that temperature. In addition, the process is called annealing, not "just apply any bit of heat and your NAND will last longer". Even furthermore, the article says a "brief and restricted jolt" with regards to the time the NAND is heated to 800 degrees Celsius.

Then there's the matter of the melting point and the thermal conductivity of the material the NAND is wrapped up in, which is plastic. I doubt it would last that long enough at 800 degrees Celsius.

A soldering iron is operating at around 350 degrees Celsius or so if operating with lead-free solder on the circuit board. 800 degrees Celsius is more than double that.

VirtualLarry's hypothesis has not been disproven at all. Annealing, while fascinating from a scientific point of view, is irrelevant to his hypothesis of high case temps killing off SSDs.
 
Last edited:
Back
Top