SSD Write Endurance 25nm Vs 34nm

[Anvil]

Kingston SSDNow 40GB (X25-V)

1517.82TB Host writes 49735977*32)
Reallocated sectors : 05 769 +1
Available Reserved Space : E8 86 (86)
POH 14074
MD5 OK

31.76MiB/s on avg (~184 hours)

--

[rkagerer]

This is a fantastic thread! It's about time someone did this sort of testing. AFAIK not even Anand tests their drives to destruction.

This is troubling, very troubling. None of the SSDs so far have gone into read-only mode upon failure. Not even the latest incarnation as of 2013. This was touted as one of the main features of SSD.

In fact, the failure mode is hard. There is no way to get the data back and programs hang hard. Nostalgically looking, HDDs were way better in this area. I would gladly buy even a slower and slightly more expensive SSD drive which always fails into a read-only mode, leaving the data accessible.

I agree wholeheartedly. I want to chime in, though, because I have a relevant story from the early days of SSD. Back in 2008 I bought a 128GB Memoright SSD. This was back when SSD's were pricey and very rough around the edges - the thing cost ~$2000. But the upgrade was the single biggest "you can feel it" improvement to my laptop's performance (not to mention solved my heat problem). After about two months it failed. I'm not sure what was written or what type of failure occurred. But SSD technology was new, and I was curious, so I did some non-scientific investigating. After roughly 10 minutes of use, the system would start freezing up and Windows would start reporting errors. If I rebooted immediately, the BIOS would hang for a while and not detect it. If I left the drive off for a period of time, I found I could plug it in and get about 10 minutes of use, although corruptions would be reported. Writing to the drive caused it to freeze up a lot quicker than reading. Two different computers gave identical results. Eventually I plugged it into a machine set up for recovery, using software (Acronis) that would only read from the drive. Although my read rate was incredibly slow, after many hours it did manage to read all my data off.

I binary compared large files to what I had in backups, checked on CRC's / PAR files, and it turned out the data that had been dormant all looked good. It only seemed the very recent data had been affected by corruptions.

Anyway I figured this seemed interesting because, although the failure mode wasn't ideal (in that the drive threw up lots of corrupted data), it did keep most of my data safe and despite being in the throws of death allowed me to read it all off. As far as I'm concerned, it fulfilled a promise here that it seems newer SSD's have all abandoned.

Granted it could just be that I was lucky and this was a fluke occurrence of this mode of failure.

[Kain665]

Oh where oh where has everybody gone?

[rkagerer]

Not sure.. but this thread may start getting more attention after it was mentioned in SlashDot comments:
http://tech.slashdot.org/story/13/02...rite-endurance

Personally, I find the real-world work folks are doing here much more compelling than the theoretical calculations that got Slashdotted. So often other factors (poor engineering, inferior firmware) have a much greater influence on failure than nand spec.

[Anvil]

Kingston SSDNow 40GB (X25-V)

1572.29TB Host writes 51520860*32)
Reallocated sectors : 05 774 +5 (769)
Available Reserved Space : E8 86 (86)
POH 14799
MD5 OK

31.72MiB/s on avg (~334 hours)

(It idled for about a week)

--

[HTWingNut]

So is there a dummy's guide for doing these SSD writes? What software do you use, size files, etc? I'm going to get one of the Samsung 840 (not Pro) 120GB drives and torture test it. Although one thing I'd like to do is every 500 P/E cycles is do a secure erase and check performance. Thanks.

edit: Nevermind I found the AnvilBenchmark program.

[Anvil]

Kingston SSDNow 40GB (X25-V)

1609.47TB Host writes 52738985*32)
Reallocated sectors : 05 775 +1 (774)
Available Reserved Space : E8 86 (86)
POH 15138
MD5 OK

31.83MiB/s on avg (~338 hours)

--

[Anvil]

Kingston SSDNow 40GB (X25-V)

1638.34TB Host writes 53685046*32)
Reallocated sectors : 05 777 +2 (775)
Available Reserved Space : E8 86 (86)
POH 15401
MD5 OK

31.83MiB/s on avg (~600 hours)

--

[sergiu]

Sandforce came and went away, so it did Samsung and Marvell, yet this poor Intel drive seems it will outlive all of them.

[Zaxx]

The very definition of 'slowly but surely'...lol

[3rase]

A test on the 256GB OCZ Vector would be interesting. Unfortunately I have neither the money nor any PC to test it 24/7. Is anyone on the Vector right now?

[canthearu]

A test on the 256GB OCZ Vector would be interesting. Unfortunately I have neither the money nor any PC to test it 24/7. Is anyone on the Vector right now?

Look at how my vertex 4 tests went ... a vector would behave quite similarly (but not as strongly as the 120gig version of course)

[Anvil]

Kingston SSDNow 40GB (X25-V)

1696.02TB Host writes 55575099*32)
Reallocated sectors : 05 788 +11 (777)
Available Reserved Space : E8 86 (86)
POH 15927
MD5 OK

31.69MiB/s on avg (~527 hours)

--

[Conquistador SW]

Thank you for keeping updating Anvil!

I am still amazed this old little Intel SSD keeps on going.

[Unit Igor]

Hey people i register just to say some things.
I follow this thread about one year now and this is something amazing that you do here.You crash your SSDs so SSD manufacturers cant lie to us,consumers,about their endurance.
Man,a lot of people must thank you.
THANK YOU
I want to say one more important thing.It seems that this thread is cooling down and this is something we can not allow!
I believe,a lot of people will like to know Vector,840 PRO and Corsair Neutron endurance.
So i suggest that you open thread(or in this) where you(moderator)will select 5 latest drives and everyone will be able to participate with at least 1$ to buying process,of SSD they want to see in endurance test.
I would be the first person that will participate with 10$ for testing Samsung 840 PRO 128GB.

[OldChap]

Wrong place to ask I know..... but given a worst case scenario running 24/7 of 6144GB of writes daily, let's assume non-compressible, choose a drive:

Samsung 840pro or Intel 520? Here, they cost a similar amount for a 128/120

Much as I have followed this thread for ages I feel I still want some of you specialists to comment on this

Back to topic: I too would chip in for purchase of drives to be endurance tested

[Zaxx]

non-compressible is 520's (SandForce's) worst enemy. Sammy 840 Pro is the winner easy.

[pH(x)]

When will we be seeing petabytes in this count? (1 billion gigabytes)

[figuretti]

When will we be seeing petabytes in this count? (1 billion gigabytes)

Just one page behind is the chart with the petabytes..

[Anvil]

Kingston SSDNow 40GB (X25-V)

1759.1TB Host writes 57642046*32)
Reallocated sectors : 05 799 +11 (788)
Available Reserved Space : E8 86 (86)
POH 16528
MD5 OK

31.83MiB/s on avg (~515 hours)

--

8 more days, and it's been 2 years!

[Zaxx]

2 years indeed....chuggin' on....slowly but very surly, lol

[sergiu]

What happened with this thread? As far as I see, with some small exceptions, it's almost dead from the end of January 2013.

[danwat1234]

Bump, keep the challenge going!

[turbox997]

We should start doing bounties to test drives, lol. I'm wondering how the new Samsung 840 EVO's will perform.

[lagittaja]

Fantastic thread! Thank you for all the hard work you've done. Really, thank you.
I'd love to chip in with some $$ for a 840 EVO SSD purchase for it to be tested. Going to get one of them myself as well.