SSD Write Endurance 25nm Vs 34nm

[Anvil]

Kingston SSDNow 40GB (X25-V)

412.07TB Host writes
Reallocated sectors : 12

MD5 OK

33.30MiB/s on avg (~71 hours)

--

Corsair Force 3 120GB

01 82/50 (Raw read error rate)
05 2 (Retired Block count)
B1 76 (Wear range delta)
E6 100 (Life curve status)
E7 44 (SSD Life left)
E9 215415 (Raw writes)
F1 286886 (Host writes)

106.19MiB/s on avg (~71 hours)

power on hours : 845

It's been 3 days without issues, I'm almost convinced that the 1.3.2 fw is what made the drive stop failing.

[B.A.T]

If the firmware update from sandforce fixed the bsod/hang/disconnect problem, that is great news
Then you guys finally can get some writing done.

[Anvil]

I'm not sure that it is the new fw on the Corsair and they won't tell, I guess we'll know in a few days.

Anyways, no disconnects since the 4th of October, that's 18 days and about 140TB of host writes ago.

[B.A.T]

You'll pass 1 PiB within 3 mnd then

[Christopher]

I can try the standard SF firmware equivalent of 1.3.2 which is 3.3.0. I've been reluctant to upgrade, but I'll give it a shot tonight.

I did get a freeze and blue screen with the Mushkin as the system drive, but I think it was me trying to use the system and install some drivers while endurance testing.


B.A.T.,

The great thing about SB processors is that they all idle around the same. A 2600K and the lowest-end dual core Celeron 2.4 G530 pretty much use the same power at idle. For a while, my 2500K was running at it's lowest idle frequency for about 15 hrs (on accident, through some Asus power saving software) and I didn't even notice. Both of my systems now are virtually silent (tiny apartment + loud system = sleepless nights), and so I'm thinking about future plans for another drive to test.

[Anvil]

The Force 3 is set to idle for the night, will resume tomorrow morning.

Current status
05 2
B1 77
E6 100
E7 44
E9 216182
F1 287909
Power on hours 848

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 30

05 2
Retired Block Count

B1 25
Wear Range Delta

F1 284182
Host Writes

E9 219916
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 128.85MB/s Avg
Intel RST drivers, Biostar TH67+ Celeron G530

688 Hours Work (24hrs since the last update)
Time 28 days 16 hours

10 GiB Minimum Free Space


To date, the Mushkin has used the same 3.2.0 FW. I'm going to upgrade tomorrow I think.

[Computurd]

That is AS-SSD access time, which is measured with 512B IOs, not 4KiB.

Yup, but i feel that it is telling of latency regardless. Main reason I linked the 4K from the other site is that they were actually using true 4K results, which is rare with those drives unfortunately. The V3 writes are a bit better though, nice point.

You have to be careful with Tom's IOMeter data. Some of the data in their reviews is labeled as QD1, but it really is not (for example, they give 100+ MB/s 4KB QD1 reads, which is absurd). I think the problem is that they set IOMeter for QD1 but also start up 4 or 8 or more worker threads. Each worker thread may only let one IO queue up, but since the worker threads are submitting IOs in parallel, the queue fills up higher than 1 anyway. I think the only thing that should be labeled QD1 is when there is only one worker submitting IOs, and it limits its queue depth to 1.

i noticed that with his link. Again, hard to find Low QD numbers for SF. weird.

I don't think the have realized how bad/wrong those iometer test are.
Iirc it was the workstation, web and server test that were the worst.

They should update those test to show the actual number of outstanding IO's. (workers * outstanding IO's)

yes horrible. i was thoroughly confused when reading some of those results.

[Anvil]

Kingston SSDNow 40GB (X25-V)

413.51TB Host writes
Reallocated sectors : 12

MD5 OK

33.16MiB/s on avg (~83 hours)

System up-time 83.5 hours

--

No movement on the Force 3 Wear Range Delta, still at 77.

About to restart the test.

[One_Hertz]

523.26TB. 6977 reallocated sectors. Max speed down to 41mb/s.

It is interesting that once the reserve space got to 1, the OS stopped complaining about the imminent failure of the SSD...

[johnw]

523.26TB. 6977 reallocated sectors. Max speed down to 41mb/s.

It is interesting that once the reserve space got to 1, the OS stopped complaining about the imminent failure of the SSD...

Based on what happened with the Samsung, I wonder if it would be worth trying a power-off intermission at this point. Say, power the SSD off for a week and see if it still works and MD5 checks after that time, before continuing the writes.

One thing that would have been nice to know about the Samsung is how much I could have written to it before its unpowered data retention went down to some amount (say, a week or a month). I'm guessing I wrote so much to it that the unpowered data retention was less than a week, maybe even only a couple days. For practical purposes, I'd think the minimum data retention anyone would want to risk would be a week or so. In that case, the amount that can be written to the SSD is probably going to be less than the amount that we measure when we write until we start getting write errors....

[sergiu]

523.26TB. 6977 reallocated sectors. Max speed down to 41mb/s.

It is interesting that once the reserve space got to 1, the OS stopped complaining about the imminent failure of the SSD...

I got some of the data from last posts and I made some simple statistics:

TB Reallocated sectors Average reallocated sectors/TB
491,4 4904
500,71 5478 61,65413534
502,35 5588 67,07317073
511 6207 71,56069364
516 6556 69,8
523,26 6977 57,98898072

Seems reallocations are slowing... It might be yet another stabilization around some certain value

[B.A.T]

Todays update:
m4
730.6351 TiB
2618 hours
Avg speed 89.23 MiB/s.
AD gone from 194 to 190.
P/E 12664.
Value 01 (raw read error rate) has changed from 10 to 11.
MD5 OK.
Still no reallocated sectors


Kingston V+100:
Will be online again from monday.

[Anvil]

Kingston SSDNow 40GB (X25-V)

414.92B Host writes
Reallocated sectors : 12

MD5 OK

36.35MiB/s on avg (~11 hours)

--

Corsair Force 3 120GB

01 92/50 (Raw read error rate)
05 2 (Retired Block count)
B1 79 (Wear range delta)
E6 100 (Life curve status)
E7 43 (SSD Life left)
E9 219382 (Raw writes)
F1 292169 (Host writes)

106.89MiB/s on avg (~11 hours)

power on hours : 880

Idling for 10 hours made no difference at all, Wear Range Delta is still increasing.
I'll move all files to another drive and then back again just to see what happens. (no cleaning or formatting)

[Christopher]

Anvil,

I think that deleting everything off the Force 3 and then putting it back won't make much of a difference in WRDelta.

Incidentally, the Mushkin peaked at 26 a couple days ago, and now is on it's way back down (it's currently at 21). If history is any indication, it should drop back down to 10 or 11, 12 eventually.

I haven't done moved the files back and forth, and I've not secure erased the drive. I have deleted the Mushkin and then performed fresh Windows installs on three separate occasions. That didn't seem to change anything either.

[Anvil]

It remained at 79 throughout the move (from and to) but during the first loop it dropped to 78

Just out of curiosity
There are 94460 files on the drive, it took ~40 minutes to move the data to an HDD, it took 18 minutes to move it back.

Also, the first loop was pretty slow, looks like avg speed is almost back to normal.

[Christopher]

WRD will often go up or down frequently. I guessed that the Mushkin would peak, then drop again, and so far that looks to be the case, but nothing would really surprise me. Lots of the drive's behavior is inexplicable, but I will say that Win7 installs from a USB key are insanely fast on the Chronos D.

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 31

05 2
Retired Block Count

B1 20 Down from 25
Wear Range Delta

F1 294589
Host Writes

E9 227218
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 128.48MB/s Avg
MSAHCI drivers, Asus M4g-z

712 Hours Work
Time 29 days 16 hours

11 GiB Minimum Free Space



I could't be happier with the new Celeron/H67 setup. I have the case in my TV stand/component rack and connected to the TV via DVI->HDMI, so I can just switch inputs on the TV so see what's going on. But then I discovered this awesome remote desktop application for WebOS and my HP TouchPad (Splashtop Remote for WebOS/iOS/Android). It's works great, and I can do almost anything from it. I bought a TouchPad when they were $99 in the HP firesale and didn't have too much to do with it until now. I just keep the TP over by my main system so I can either see the system itself or have the TP to check in on the endurance rig. Of course, it's not as big of a deal if you're using a SSD that doesn't crash when you look at it wrong.

[Ao1]

It remained at 79 throughout the move (from and to) but during the first loop it dropped to 78

Just out of curiosity
There are 94460 files on the drive, it took ~40 minutes to move the data to an HDD, it took 18 minutes to move it back.

Also, the first loop was pretty slow, looks like avg speed is almost back to normal.

Interesting. So based on observations from your drive and SymbiosVyse’s drive it seems that SF drives do not rotate static data, rather they rely on the user to change the static data, at which point they then try to limit wear on the most worn blocks.

Christopher, your drive seems to be an exception. Have you at any stage removed the static data?

[B.A.T]

Todays update:
m4
737.2672 TiB
2637 hours
Avg speed 89.22 MiB/s.
AD gone from 190 to 187.
P/E 12773.
Value 01 (raw read error rate) has changed from 11 to 15.
MD5 OK.
Still no reallocated sectors


Kingston V+100:
Will be online again from monday.

Another thing I've been aware of during the last couple of updates are the strange increase of value 195. According to Micron that value is not described. Wikipedia tells me this:

195 0xC3 Hardware ECC Recovered N/A (Vendor specific raw value.) The raw value has different structure for different vendors and is often not meaningful as a decimal number.

Link

I'm not able to find anymore useful data on this so any help from you would be great

[Christopher]

Ao1,

Yes, I've removed the static data 3 times by way of deleting the volume, then reinstalling windows.



I had another TRIM/File delete related crash just a few minutes ago. I should have left it the way it was when I had the problem eliminated.

Using the Chronos as the boot drive leaves it vulnerable to more problems in my estimation.

Luckily I watched it happen so I got the system restarted immediately and back up and running.

Two days ago WRD was at 26, now it's down to 17 again. If you were graphing Wear Range Delta (and Ao1 is) you'd see two peaks, then valleys down at 10,11,12 or so.


The Mushkin just hit 303TiBHW a short time ago.

[Anvil]

Kingston SSDNow 40GB (X25-V)

417.61B Host writes
Reallocated sectors : 12

MD5 OK

33.91MiB/s on avg (~35 hours)

--

Corsair Force 3 120GB

01 90/50 (Raw read error rate)
05 2 (Retired Block count)
B1 82 (Wear range delta)
E6 100 (Life curve status)
E7 41 (SSD Life left)
E9 225669 (Raw writes)
F1 300510 (Host writes)

MD5 OK (MD5 checking was enabled for this session)

102.68MiB/s on avg (~22 hours)

power on hours : 905

Yesterday B1 started increasing within a few loops so the decrease was only temporary.

--

Drive is back up running on fw 1.3.3, avg speed looks to be a bit up, will know more next update.

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 32

05 2
Retired Block Count

B1 16
Wear Range Delta

F1 304933
Host Writes

E9 235194
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 127.92MB/s Avg
Intel RST drivers, Celeron G530 Biostar TH67+

736 Hours Work (24hrs since the last update)
Time 1 month 16 hours

11 GiB Minimum Free Space




Wear Range Delta continues to drop, but I'm a little surprised that the increase and decrease has happened so quickly.

It was 25 when I posted the update on the 29th day, but perhaps the difference between B1 = 11 and B1 = 21 is not so much when the Force 3 and F40-A are both given to high WRDs.

Besides the capacity difference, NAND/PE differences, there is also the matter of speed. Shouldn't faster drives be able to rotate static data more quickly than slower drives? The Chronos D is roughly 25% faster than the Force 3, and maybe over 2x the speed of the F40-A (I believe it was about 55MBs @ 46% [Application]).

[Computurd]

I had another TRIM/File delete related crash just a few minutes ago. I should have left it the way it was when I had the problem eliminated.

hmm. is this with the new FW?

[Computurd]

I had another TRIM/File delete related crash just a few minutes ago. I should have left it the way it was when I had the problem eliminated.

hmm. is this with the new FW?