SSD Write Endurance 25nm Vs 34nm

[johnw]

What I was trying to say was that MWI is based on the theoretical P/E cycles for NAND. What this thread is showing is that actual P/E capability is significantly more. The Samsung looked like it kept writing until the P/E capability was physically deleted. The 1 year data retention duration however is most likely based on expiry of the theoretical P/E cycles, rather than physical depletion.

Ah, okay, that makes more sense than what you wrote before!

So, you think that if I had stopped writing to the SSD when MWI hit the failure threshold (which was probably around 5000 erase cycles), then we could have expected one year unpowered data retention. But since I wrote well past MWI threshold, all bets are off and it could last much less than a year. I think that may be the correct way to look at it.

It will be interesting to see what One_Hertz finds.

[Hopalong X]

johnw

That hammer probably didn't help either.

[Christopher]

johnw

That hammer probably didn't help either.

If a hammer won't fix it, it's clearly defective.

[johnw]

That hammer probably didn't help either.

That was not a hammer , that was my forehead

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 26
Once you get to MWI 10 with a SandForce, you enter the doldrums.
My sailboat won't be going anywhere for a while.

05 2
Retired Block Count

B1 17
Wear Range Delta

F1 243978
Host Writes

E9 188168
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 130.20MB/s
RST drivers, Asus M4g-z

595 Hours Work (20 hrs since the last update)
Time 24 days 19 hours

6 GiB Minimum Free Space
to get me over the 10K files a loop

[Christopher]

So SandForce has officially identified the BSOD bug, and released some firmware. OCZ is getting it first.

http://www.anandtech.com/show/4973/s...vailable-today

And I had just fixed mine...

We'll see how it works out.

EDIT
I can test the TRIM problems, and I can test the other BSOD issues I guess, so I'll update the FW if it ever comes out for the Mushkin and revert my system to the "SF unfriendly" config.

[Anvil]

Kingston SSDNow 40GB (X25-V)

402.90TB Host writes
Reallocated sectors : 12
MD5 OK

33.95MiB/s on avg (~34 hours)

--

Corsair Force 3 120GB

01 92/50 (Raw read error rate)
05 2 (Retired Block count)
B1 62 (Wear range delta)
E6 100 (Life curve status)
E7 50 (SSD Life left)
E9 192881 (Raw writes)
F1 256859 (Host writes)

107.07MiB/s on avg (~34 hours)

power on hours : 759

B1 is still climbing.

Looks like the 1.3.2 (3.3.2?) made the difference, no disconnects or BSOD's for more than a week!
(and I'm back to normal settings)

[Ao1]

It looks like SF drives struggle with static wear levelling when the drives are in constant use. Still I don’t think this tells the whole story, as it only records the max difference between highest and least worn block, not the average wear across all blocks.

Vapor are you still with us? Is your SF drive still chugging along?

[bulanula]

Nice news about SF. Should have been fixed a long time ago.

[One_Hertz]

511TB. 6207 reallocated sectors. The speed is down by roughly 3MB/s from the original.

[bluestang]

M225->Vertex Turbo 64GB Update:

528.02 TiB (580.57 TB) total
1371.28 hours
10725 Raw Wear
117.52 MB/s avg for the last 16.62 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count (Realloc Sectors) at 9.
(1=Bank 6/Block 2406; 2=Bank 3/Block 3925; 3=Bank 0/Block 1766; 4=Bank 0/Block 829; 5=Bank 4/Block 3191; 6=Bank 7/Block 937; 7=Bank 7/Block 1980; 8=Bank 7/Block 442; 9=Bank 7/Block 700)

[B.A.T]

New update, and today I bring good news. A new milestone is passed
m4
700.4629 TiB
2519 hours
Avg speed 90.49 MiB/s.
AD gone from 211 to 207.
P/E 12152.
MD5 OK.
Still no reallocated sectors
And a new development. Value 01 (raw read error rate) has changed from 0 to 4.


Kingston V+100
My rig is at work and I'm home sick rest of the week so I'll have to wait until monday. I'm monitoring it with teamviewer.

[Christopher]

After about 40hrs of running with only 6GB of free space on the drive, Wear Range Delta has jumped up to 20. I would expect it to be a little higher by tonight when I post an update. Unfortunately, I need the those extra 6GB to get the file count over 10K, or it will crash during file deletes after a few hours.

With less static data and more free space the wear range delta had been mostly holding steady. Now with the increase in static data and half as much free space, there's not much room left for wear leveling. I think I'll run it like this for a little while and see where it ends up.




Anvil,

Corsair's 1.3.2 is 3.3.0 I believe.
OCZ and Corsair rename their FW to make FW comparisons difficult -- which is silly. In OCZs case though, they throw some other stuff on top for more performance in some areas.

[Anvil]

I'll send you a preview of the next beta, there are some tweaks you could try.
My Force 3 will have to run another 4-6 hours before I'm willing to make any changes.

Also I'm going to reinstall the ASRock Z68 tonight, just to make sure that I'm running a non tweaked installation.
(will be upgrading the old Vertex 1 as well, it's on fw 1.5 and 1.6/1.7 looks to have fixed a lot of issues + better Endurance)

[bulanula]

Nothing seems to be able to stop the M4. Not even 1 reallocated sector !?

[bluestang]

Nothing seems to be able to stop the M4. Not even 1 reallocated sector !?

It's still at 0 reallocated sectors genius

[Christopher]

I'll send you a preview of the next beta, there are some tweaks you could try.
My Force 3 will have to run another 4-6 hours before I'm willing to make any changes.

Also I'm going to reinstall the ASRock Z68 tonight, just to make sure that I'm running a non tweaked installation.
(will be upgrading the old Vertex 1 as well, it's on fw 1.5 and 1.6/1.7 looks to have fixed a lot of issues + better Endurance)

1.7fw for the Vertex looks to have a few problems. Originally it was yanked due to what OCZ calls an installer problem. Subsequently it was released, but some are still having installation issues. Once you get it installed, some have had issues with the drive exhibiting strange behavior. 1.7 is supposed to manage some aspects of drive endurance and wear leveling differently, but I'm using 1.6 on three drives and find it to be pretty good over 1.5's good-but-not-great write amplification. I may try upgrading one of my Agilities to 1.7 to see what the fuss is about. I'd like to try the new Arrowana FW if they ever get their stuff together and get it out, but it doesn't seem to be doing the new VertexPlus drives any favors -- they're far more problematic than any SF2200 drive floating around out there. The Arrowana FW for old school Indilinx drives was supposed to boost random performance by 5x according to OCZ (back in May, in a press release), but I doubt it's ever going to come out now.

I think I'll upgrade the older Agility now -- what's the worst that could happen??

EDIT
There is an AgilityTrb1.7 file in the 1.7 updates for Indy drives... What's up with that?

[Computurd]

wasnt aware there was a 1.7 for the older vertex. interesting. on 1.6 for my drives.

is 1.7 stable?

[Computurd]

wasnt aware there was a 1.7 for the older vertex. interesting. on 1.6 for my drives.

is 1.7 stable?

[Christopher]

wasnt aware there was a 1.7 for the older vertex. interesting. on 1.6 for my drives.

is 1.7 stable?

They had it out a few months ago... but removed it. Back in September they brought it back out. I just flashed my Agility from 1.6 to 1.7 -- I think you should probably stay where you're at for the moment. My Agility seems a fair bit slower, but it could be an optical illusion... or I have a brain tumor. Not sure.

Now I feel extra stupid because I noticed that there was an AgilityTurbo 1.7 in the update .zip files, so I posted about it on the forums, and I got the "we never released it answer"... I was hoping they had developed some kind of faster Agility FW somewhere along the line and were going to hook it up. The 4K randoms on my Agilities are pathetic... like 8GB PATA SSD pathetic, SD or CF pathetic... no where near as good as Bluestang's Franken-tex or my 120GB Vertex Turbo.

[DooRules]

got a buddy that has been using this FW on 4 30 gb vertex turbos in R0 on x58 mobo, has been stable for about a week now, previously he could only go a day or two and drives would start to drop out...we shall see, got 4 of them turbos myself somewhere

[RyderOCZ]

Arowana will only run on 34nm Nand Vertex/Agility/Solid2 drives, so unless you have a drive with 34nm Nand on it, Arowana will not work.
Vertex Plus is not "far worse than SF-2000" is.. already released a firmware update and have another one ready. Should resolve the 2 issues we have seen.

Vertex/Agility have ALWAYS had anemic 4K's in AS-SSD, no Cognac build will change that. Also, the smaller the drive, the slower the 4k with that controller.

[Christopher]

Arowana will only run on 34nm Nand Vertex/Agility/Solid2 drives, so unless you have a drive with 34nm Nand on it, Arowana will not work.
Vertex Plus is not "far worse than SF-2000" is.. already released a firmware update and have another one ready. Should resolve the 2 issues we have seen.

Vertex/Agility have ALWAYS had anemic 4K's in AS-SSD, no Cognac build will change that. Also, the smaller the drive, the slower the 4k with that controller.

In my experience the 2281 issue's gravity really depends on what the hell you're doing with the drive; So I wasn't crestfallen once I spent a couple weeks figuring out what was going on, though some might be more sensitive. Once I figured out that there was only one thing I needed to do for my 2281 endurance drive, I've not had any problems since. It just took a couple weeks of throwing ideas against the wall, trying this and that, to come to a solution that works for me and my system. I guess you could say that I was lucky in that my usage and system resulted in crazy consistent crashes, which makes it easier to work around.

Have you seen how awesome Bluestang's drive is in 4K's? Fantastic for a 64GB... it's a thing of beauty. It's seriously Hall of Fame material.

My one Agility is hamstrung by other factors as well. I knew it wasn't so hot when I bought it, and if I didn't, I sure as hell figured it out when I bought another one last month. I might piss and moan about them, but I sure do have an unreasonable fondness for my Indilinx units -- all of them.

[Anvil]

The drives have been offline while reinstalling and upgrading the boot drive.
(no issues except for PE count spec was halved for some reason, used to be 10K and is now 5K, health 92%->83%)

Kingston SSDNow 40GB (X25-V)

403.95TB Host writes
Reallocated sectors : 12

--

Corsair Force 3 120GB

01 120/50 (Raw read error rate)
05 2 (Retired Block count)
B1 64 (Wear range delta)
E6 100 (Life curve status)
E7 49 (SSD Life left)
E9 195534 (Raw writes)
F1 260392 (Host writes)

power on hours : 770

Both drives restarted the test just minutes ago.

[Christopher]

Anvil,
from 1.6 on, V1/Agility/Core2s only have 5k PEs now. Even if it had 50nm Samsung @10K PEs, it will only be 5K now.

Some people got upset, but if the NAND doesn't fit, you must acquit.