Samsung 830 Any way to check its life expectancy ?

[REDBULL2K7]

Hi every one bought a samsung 830 ssd and didnt realise that i was downloading to it , i changed the download folder to another drive but didnt realise it downloaded to a temp folder to so now im wondering what ive done and how long my drive has left ? , lifessd does'nt show this .

thanks

 

[tcsenter]

Huh???

 

[OlafSicky]

It will die soon dl to an ssd is a kiss of death How could you make such a mistake

 

[REDBULL2K7]

Easy lol is there a program that works with the samsung 830 that will tell you how long its got like lifessd does ?

thanks

 

[hhhd1]

Get a program named "Crystal Disk Info" and post a screen shot of it, with all data visible in here (you can hide the serial number, its not important), and someone here will look at it.

 

[REDBULL2K7]

thanks

https://dl.dropbox.com/u/88547982/Capture.JPG

 

[REDBULL2K7]

Thanks

 

[Hellhammer]

You have absolutely nothing to worry about, 893GB of host writes is nothing.

 

[mikeymikec]

You have absolutely nothing to worry about, 893GB of host writes is nothing.

What would you regard as a lot for an SSD, just out of curiosity?

 

[REDBULL2K7]

You have absolutely nothing to worry about, 893GB of host writes is nothing.

thanks for that

 

[hhhd1]

You have used 11 erase cycles of your SSD, and its rated to have 3000 cycles, in real life tests, Samsung 830 can go for upto 6x its rated life cycles, sometimes it can reach 20k+ cycles.

since you used 11/3000 in 1261 hours,

if your usage pattern stayed the same, then the SSD should live for another 342648 hours of uptime, (14.3k days or 39 years)

assuming that you do not leave the computer on 24/7, the SSD will outlive you.

 

[REDBULL2K7]

thanks guys for taking the time .

 

[Hellhammer]

What would you regard as a lot for an SSD, just out of curiosity?

It's all up to one's workload and the age of the SSD. 10TB is a lot if you get that in a month but in a year it's nothing (around 27GB per day, which is still far more than average but nothing to worry about). If we go into hundreds of Terabytes, then that's a lot but you need a hefty workload to get anywhere near that.

 

[Phynaz]

in real life tests, Samsung 830 can go for upto 6x its rated life cycles, sometimes it can reach 20k+ cycles.

Link?

 

[tweakboy]

SSDs dont die. They wear over lifetime. You can keep a tab on things by getting

CrystalDisk info free small app , and it will says how many percentage it is and gives you wear and tear results etc.. and shows expentancy. Watch your percentage and if it says Good. gl For example my dads is at 97 percent a-data cheapo it used to transfer 375mbps lol now its down to 300mbps ,, so its wear tear lifetime is going down.

But SSDs don't suddenly die unless the controller on it sandforce maybe....

 

[hhhd1]

Link?

http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm/page209

[I][B]Samsung 830 256GB Day 189[/B][/I]

(GiB) [B]4,666,340[/B]
(TiB) [B]4,556[/B]
(PiB) [COLOR=#FF0000][B]4.48[/B][/COLOR]


(Avg) [B]297.53[/B] MB/s over the past 1600+ hours

(B1) Wear Leveling Count: [B]20,186[/B]

(B6) Erase Fail Count: [B][COLOR=#FF0000]10[/COLOR][/B]
(B6) Runtime Bad Block Count: [B][COLOR=#FF0000]11[/COLOR][/B]
(05) Reallocated Sectors: [B][COLOR=#FF0000]45056[/COLOR][/B]

(POH) 4585

 

[Wall Street]

That extreme systems thread is nuts. The samsung is now past 5 petabytes and still going. It is funny that early discussions about flash endurance worry normal users when really only server admins could really ever expect to see this many writes. The creator of this thread has yet to even reach a terrabyte in writes so there is no way flash endurance should be a factor as decades from new he would still be below a petabyte. Downloading to an SSD is not a problem.

 

[REDBULL2K7]

cheers i allways thought downloading to a ssd was a big no no .

 

[hhhd1]

... It is funny that early discussions about flash endurance worry normal users when really only server admins could really ever expect to see this many writes ...

Its also funny when people think that TLC with 1000 write cycles are too little to install OS and few games.

But to be fair, some older SSDs with horrible write amplification were too easy to go through write cycles.

(if memory serve me right) Some drives cough*ocz*cough do not always reach their rated 3000 cycles.

 

[Wall Street]

Its also funny when people think that TLC with 1000 write cycles are too little to install OS and few games.

But to be fair, some older SSDs with horrible write amplification were too easy to go through write cycles.

(if memory serve me right) Some drives cough*ocz*cough do not always reach their rated 3000 cycles.

Failures due to write cycles and failures due to firmware bugs are not the same thing. It is hard to know how many SSDs have failed due to bad Sandforce firmware or the Intel 8 MB bug vs. how many have failed due to the flash failing.

 

[ericloewe]

That extreme systems thread is nuts. The samsung is now past 5 petabytes and still going. It is funny that early discussions about flash endurance worry normal users when really only server admins could really ever expect to see this many writes. The creator of this thread has yet to even reach a terrabyte in writes so there is no way flash endurance should be a factor as decades from new he would still be below a petabyte. Downloading to an SSD is not a problem.

It's funny how people prefer "Your product may last forever (but will almost certainly randomly fail before then. you might want to avoid dropping it, too...)" to "Your product will probably fail after xxxx uses and will probably not die randomly before then (with some exceptions, like OCZ and 8MB Intels)".

 

[Hellhammer]

Failures due to write cycles and failures due to firmware bugs are not the same thing. It is hard to know how many SSDs have failed due to bad Sandforce firmware or the Intel 8 MB bug vs. how many have failed due to the flash failing.

Even if the NAND physically wears out, the SSD should still be detected by the system and the data should be readable (the data has not gone anywhere, you just can't write to the drive anymore because all user accessible blocks have been retired). Hence you can actually distinguish a failed SSD and a worn out SSD, although there are of course differences between SSDs. Either way, I seriously doubt that any consumer has been able to wear out an SSD unless specifically trying to do so.

 

[frostedflakes]

Even if the NAND physically wears out, the SSD should still be detected by the system and the data should be readable (the data has not gone anywhere, you just can't write to the drive anymore because all user accessible blocks have been retired). Hence you can actually distinguish a failed SSD and a worn out SSD, although there are of course differences between SSDs. Either way, I seriously doubt that any consumer has been able to wear out an SSD unless specifically trying to do so.

This, it's actually fairly easy to tell. Firmware bug usually bricks a drive without any warning and makes it completely unrecognizable, whereas you have a lot of warning before you run out of P/E cycles and the way the drive fails is entirely different, it will still appear to work but you'll start to get file corruption, or the drive will only work in read-only mode and you won't be able to write to it all.

 

[jwilliams4200]

Even if the NAND physically wears out, the SSD should still be detected by the system and the data should be readable (the data has not gone anywhere, you just can't write to the drive anymore because all user accessible blocks have been retired).

That is not necessarily true. It has been claimed for quite a while (I think the first place I saw this claim was from Anand in an SSD article), but actual tests have shown otherwise.

If you read through the long thread on XtremeSystems, you will see that several of the SSDs that fail from worn out flash cannot be read or often even recognized by the BIOS.

The problem is that eventually, the unpowered data retention time of the flash chips gets very short (less than a day) and if you power off the SSD and then try to power it on and read it, it will fail to even be detected by the computer.

I don't know for sure, but I think the reason is that the LBA index (aka flash translation layer = FTL) needs to be read from flash in order for the SSD to initialize properly. But if the unpowered data retention time has gotten short enough, the SSD will not be able to read the index from flash when it initializes, and so it simply fails to initialize and does not respond to normal SATA queries.

This also explains why the XS forum tests show that many SSDs can be written to for far beyond 3000 or 5000 erase cycles. The manufacturer specifications are for how many erase cycles can be applied without the unpowered data retention time dropping below 1 year (for consumer SSDs, 3 months for enterprise SSDs). But most of the XS testers are keeping their SSDs powered on continuously, so many of them are likely well past the point where unpowered data retention has fallen below 1 year. The ones that have gotten to 20,000 or 30,000 erase cycles likely have reached data retention times of less than 24 hours, and if they were to power down the SSD for a day or two, it is unlikely to be readable when they power it back up.

Unfortunately, it is much more time consuming to measure the point where an SSD has its data retention time drop below 1 year (or 3 months) since you would have to stop for 1 year every X erase cyles and test it. The manufacturers use high temperature accelerated aging (and hundreds of flash chip samples) to test such parameters, but I do not think that is practical for most home users, even Xtreme ones.

I think a more useful test than what the XS people are doing would be to do a "1 week torture / 1 week power down / 1 week torture / 1 week power down" cycle to find out how much the SSDs can write while still having a 1 week unpowered data retention time. Or even a 1 day torture / power-down cycle to look for the amount of writes before data retention drops below 24 hours.

 

[mikeymikec]

For me personally, HDDs were around before I started getting into computers, and what I've learnt (and am still learning) about them has established a kind of playing field. I think I know a fair bit about their quirks, relative strengths and weaknesses.

The rest of a computer is neither here nor there without data storage and it has longer term responsibilities. A piece of data can theoretically last forever provided it is taken care of and is transferred from one storage system to the next correctly.

SSDs represent a whole new playing field (yes, it has its base in flash memory which I'm pretty sure we've all been using for some time, but the responsibilities of a mobile piece of storage aren't the same as one with an OS on it and the longer term in mind). Only an idiot wanders on blindly without preparation.

When perpendicular data arrangement on a hard disk platter came along, a friend of mine was saying that he would tread very carefully before moving on to this bit of tech because it throws a good few principles about data retrieval from hard disks on their heads. AFAIK there wasn't anything to worry about, but that doesn't mean one should stop being cautious.

If I had the money, I would be reasonably happy about investing in a nice little SSD like the Samsung 830. I'll still have backups though

Every system has its foibles. I think anyone who thinks that SSDs won't have any and will only fail nicely (apart from "the dreaded SandForce" of course, SSDs have to have their bogey man as well) is being naive.