SSD Write Endurance 25nm Vs 34nm

[Anvil]

Yes, Secure Erasing looks to stop the throttling, however, it's not what one wants, especially for a system/boot drive.

There may be restrictions along that path though, my drives aren't that much used for benchmarking so I just don't know, it's been working the few times I've felt I had to clean them.

[Micutzu]

Interesting, some hours of considerations made you to change your mind to no way to maybe. The way things are going even 150TB is believable.

...

Flash cell wearing might not be - and probably isn't - linear with the amount of data written on the drive. From a certain point, the wear indicator will start to accelerate.

[Ao1]

You mean the paranoia cliff in this slide? It should not happen until the wear out indictor gets to 1, but we will soon see

[Anvil]

Time for a small update, I found an old Intel SSD Toolbox that works with the Kingston. (haven't had the time to modify the latest one yet)

7.32TBW

7_32_tbw.PNG

The only thing that has changed is E9 (Wear out), -1

7_33_tbw_cdi.PNG

I changed the Power Plan to "Balanced" and completely forgot to disable hibernation so I ended up with a hibernated computer, not sure for how long, could be 1-2 hours.
(won't do that again )

The TaskManager showing number of "bytes" written
7_37_TB_TaskMan.PNG

[ImportantAwareness89]

How do I get the Host Writes for my SSD? I have a Crucial m4 128gb

[Computurd]

@ anvil ROFL! that bytes write counter is insane!

very interesting guys, great to see everyone testing, just like the good ol days cant wait to see the results of this!

[groberts101]

Very nice commitment here, guys. Will be linking to this thread every time I see someone worrying about every little write to their SSD.

May I suggest that you guys use the 10 [DEC] readout in CDI instead? Helps us to see the numbers without a calculator or being told where they're at now.

using.. Function>Advanced Feature>Raw Values> 10 [DEC].. will show up like so.


Thanks for the effort here. Much appreciated.

[bulanula]

WOW !

Just cannot thank you enough, guys. I really wanted this sort of test done for a while now before I jump on the unproven SSD bandwagon. Time will prove what REALLY happens when the NAND cycles are all used up !

Thanks again !

[Anvil]

May I suggest that you guys use the 10 [DEC] readout in CDI instead? Helps us to see the numbers without a calculator or being told where they'r at now.
...
Thanks for the effort here. Much appreciated.

The hex values are handy as I'm preparing to include SMART in my own application.

I'll see if I can change to decimal when taking screenshots
(the most interesting values are still easily readable though)

..

A milestone of 10TB Host Writes was set some time ago.
Media Wearout changed to 93, nothing else of interest has changed, speed hasn't changed either. (each loop takes 495 seconds +/- a few seconds)

More than 3.3 million files have been created so far.

10_05_tbw.PNG

[Roadhog]

Subscribed for turnout.

[bulanula]

VERY interesting !!!

The decline in the E9 statistic is not so important. What I am most interested is the end result. How many host writes were needed to make the SSD read only ( if that happens or not is debatable etc. ) !

Just cannot wait. Thanks every so much Anvil and One_Hertz !!! You are doing this for science

Just to make sure I have read this right. Does your program accomplish all of these writes through large sequential files or small random files ? If I understood OK it is the latter which would be very nice indeed !

Also, is this the G1 X25-V with 50nm MLC NAND or the G2 X25-V with 34nm MLC NAND ? How filled is it when it comes to static data ( eg windows ) and the rest taken up by your program ? What is the empty space ? Thanks !

[One_Hertz]

We are doing both smaller and larger files with the majority being smaller files. Anvil has a G2 X25-V and I have a G3 320. Both 40GB. We both have around 11GB of static data on it and we are both keeping about 12GB free with the rest of the space being used to write to. Here is what my toolbox looks like so far.

[Computurd]

on a side note...Intel has expanded their warranty on the 320 from three to five years! thats a good warranty, five years. it wqs just announced so they must be pretty confident in the durability of this device. might be harder to kill than we thought!!

http://www.xtremesystems.org/forums/...d.php?t=271196

[railmeat]

subscribed.

[nn_step]

Here is some already collected endurance information on several drives
http://www.codinghorror.com/blog/201...ive-scale.html

[johnw]

Here is some already collected endurance information on several drives
http://www.codinghorror.com/blog/201...ive-scale.html

That is just anecdotal evidence of failures. And it does not even include any details about the symptoms of the failures, possible reasons for failures, whether any data could be recovered, and whether the drives were replaced under warranty. Completely useless. Worse than useless, since the systematic data I have seen on failure rates of SSDs shows them to have lower failure rates than HDDs, and this article tends to imply the opposite, without having a credible methodology.

[Ao1]

What's more impressive is that Anvils app generates un-buffered file sizes randomly, so it's a realistic work load. Also no system cache.
My retailer let me down on the V2 (out of stock, despite being shown as in stock). Thanks SCAN. I've ordered another one elsewhere. Should be delivered Saturday.

[railmeat]

That is just anecdotal evidence of failures. And it does not even include any details about the symptoms of the failures, possible reasons for failures, whether any data could be recovered, and whether the drives were replaced under warranty. Completely useless. Worse than useless, since the systematic data I have seen on failure rates of SSDs shows them to have lower failure rates than HDDs, and this article tends to imply the opposite, without having a credible methodology.

"Completely useless. Worse than useless, since the systematic data I have seen on failure rates of SSDs shows them to have lower failure rates than HDDs"

its not useless,its a eye opener.

"since the systematic data I have seen on failure rates of SSDs shows them to have lower failure rates than HDDs"

how is this even possible since ssd,s are still to new(2+yrs) to compare to tried and true veteran (6+ year old) western digital raptors...lol? and sorry raptors are thee only drives worth comparing against any ssd.





im simplying replying from facts from my life.i have had raptors since they were released (36 gigers) in raid o at $250 a pop and ran FLAWLESS for 3-4 years then the 74gigers,then i grabbed 2 raptor-x (clear top)150,s in raid,again flawless.they just last period.plug and play literally.

now as far as ssd,s go they simply have NOT been out Long enough for me to give them any real life credibility or stabilty being i bought 1 crucical c-300 and x58a gig ud3r mobo(tried both ports,all firmware on mobo and ssd done)but it died quickly,like 2 weeks or less with 0 signs of problems,wierd.but man was it fast.fine so i got a ocz vertrex limited editon 100 gig,died 1 week,got another ocz lidetical drive lasted 1 month and i was like wow its ok now,then bam stuttering,slowing down,non-responsive ALOT,had enough and returned that 1 as well.heres my link.are they FAST?...omgosh yes but i simply still dont trust them period for good reasons i witnessed myself 3x.i still think its the intels chipset and not the drives problem reading the ssd instruction set on the ssd controller.i will let time pass,more tests done,let more bugs be worked out.im a gamer,not a guina pig for hardware.need stabilty.

my post
http://www.xtremesystems.org/forums/...ight=bootracer






this is NOT a isolated instance and needs not to be down played as one,just the facts being diff ssd,s died period and shows me i was not doing something wrong with my 3 ssd,s i gave a chance.

http://www.codinghorror.com/blog/201...ive-scale.html


"Portman Wills, friend of the company and generally awesome guy, has a far scarier tale to tell. He got infected with the SSD religion based on my original 2009 blog post, and he went all in. He purchased eight SSDs over the last two years … and all of them failed. The tale of the tape is frankly a little terrifying:"


•Super Talent 32 GB SSD, failed after 137 days
•OCZ Vertex 1 250 GB SSD, failed after 512 days
•G.Skill 64 GB SSD, failed after 251 days
•G.Skill 64 GB SSD, failed after 276 days
•Crucial 64 GB SSD, failed after 350 days
•OCZ Agility 60 GB SSD, failed after 72 days
•Intel X25-M 80 GB SSD, failed after 15 days
•Intel X25-M 80 GB SSD, failed after 206 days
You might think after this I'd be swearing off SSDs as unstable, unreliable technology. Particularly since I am the world's foremost expert on backups."


i like this thread alot and watching it closely and slowly watching 99-98-97-96-95 on this 1 drive being tested from the op.u dont think i wanna EASILY run out and grab a ocz vertex3 right now.im waiting and letting time prove itself with these drives...

lack of communication is the problem here.intel/amd and all of the major ssd controller manufactuers need to spend ALOT more time together in the labs then a quick "hotfix"


@NN "It's hardware that makes a machine fast. It's software that makes a fast machine slow." agee %100

[johnw]

how is this even possible since ssd,s are still to new(2+yrs) to compare to tried and true veteran (6+ year old) western digital raptors...lol? and sorry raptors are thee only drives worth comparing against any ssd.

Of course we cannot have good statistics on 5 year reliability of SSDs (although we can make good guesses based on tests and data we do have). But that is not the point. The article referenced is pure propaganda since it purports to say that SSDs have a high one or two year failure rate. There is some reasonable data already on one or two year failure rates of SSDs, so it is absurd to give any credibility to the kind of anecdotal evidence the article mentions.

this is NOT a isolated instance and needs not to be down played as one,just the facts being diff ssd,s died period and shows me i was not doing something wrong with my 3 ssd,s i gave a chance.

It absolutely should NOT be given credence. Anecdotal evidence is useless, or worse than useless, when looking at reliability. A careful statistical study with good methodology is required.

The problem with anecdotal evidence is that there are millions of SSDs in use. Even if the annual failure rate is only 0.1%, there should be thousands of SSDs failing each year. All anecdotal evidence tells us is that there are indeed failures -- it gives us no useful information on the failure rate. The useful information that anecdotal evidence might provide is failure modes and how companies deal with warranties. But the referenced article provides nothing like that.

[bluestang]

storagereview.com is wrong. Intel is not the 1st mfg to give a 5 yr warranty on consumer drive....Crucial had a 5 yr on the M225.

[Vapor]

Let's keep the thread on subject

[railmeat]

That is just anecdotal evidence of failures. And it does not even include any details about the symptoms of the failures, possible reasons for failures, whether any data could be recovered, and whether the drives were replaced under warranty. Completely useless. Worse than useless, since the systematic data I have seen on failure rates of SSDs shows them to have lower failure rates than HDDs, and this article tends to imply the opposite, without having a credible methodology.

Let's keep the thread on subject

yes sir

[alfaunits]

This will be nice - I'll know whether my X25-Es died for being an X25-V in disguise Thanks for the work guys, truely appreciated.

[railmeat]

This will be nice - I'll know whether my X25-Es died for being an X25-V in disguise Thanks for the work guys, truely appreciated.

"Thanks for the work guys, truely appreciated."

agree %100,it is very much so appreciated,thank you.


and to mr.vapor.---full respect for the forums

[nn_step]

Of course we cannot have good statistics on 5 year reliability of SSDs (although we can make good guesses based on tests and data we do have). But that is not the point. The article referenced is pure propaganda since it purports to say that SSDs have a high one or two year failure rate. There is some reasonable data already on one or two year failure rates of SSDs, so it is absurd to give any credibility to the kind of anecdotal evidence the article mentions.



It absolutely should NOT be given credence. Anecdotal evidence is useless, or worse than useless, when looking at reliability. A careful statistical study with good methodology is required.

The problem with anecdotal evidence is that there are millions of SSDs in use. Even if the annual failure rate is only 0.1%, there should be thousands of SSDs failing each year. All anecdotal evidence tells us is that there are indeed failures -- it gives us no useful information on the failure rate. The useful information that anecdotal evidence might provide is failure modes and how companies deal with warranties. But the referenced article provides nothing like that.

If you prefer a more careful study. You will have to wait about 1-2 years before the following study will finish:

I have bought 2 of each of the following:

OCZ Agility 3 AGT3-25SAT3-120G
Corsair Force CSSD-F120GB2-BRKT
OCZ Vertex 2 OCZSSD2-2VTXE60G
Corsair Performance 3 Series CSSD-P3128GB2-BRKT
Crucial RealSSD C300 CTFDDAC064MAG-1G1
SAMSUNG 470 Series MZ-5PA128/US
Intel 510 Series (Elm Crest) SSDSC2MH120A2K5
Intel X25-M SSDSA2MH160G2K5
Kingston SSDNow V+ Series SNVP325-S2B/128GB


and am going to subject them to the following actions:

Using a Debian 6.0 Linux box running off of a single 8GB MicroSD card, an undervolted Phenom2, 2GB of DDR2, and a twin set of areca ARC-1231ML-2G PCI Express SATA II (3.0Gb/s) Controller Cards, I will perform a continuous series of reads and writes to the drives until they fail and record exactly how many writes it can actually survive. [The other 3 cores will be dedicated to WCG]

Any questions or comments before I begin the test on May 24, 002011 @ 12noon EST?