SSD Write Endurance 25nm Vs 34nm

[Christopher]

the VT hit mwi 0 at 281 197GiB

[Christopher]

Vertex Turbo

308373GiB
84.40 MB/s

AVG PE Count 5512


Samsung 830

544,535GiB
290.83 MB/s

PE Count 2386
MWI 33

Octane

248,260GiB
158.45MB/s

[B.A.T]

Todays update:

Intel X25-M G1 80GB
496.46 TiB
22254 hours
Reallocated sectors : 64 to 72
Available Reserved space: 96 to 95
MWI=243 to 236
MD5 =OK
43.42 MiB/s on avg


m4
680.0999 TiB
2919 hours
Avg speed 74.42 MiB/s.
AD gone from 223 to 219.
P/E 11793.
MD5 OK.
Reallocated sectors : 00

[Christopher]

OCZ Octane 128 Day 18

260,460.80 GiB
158.40 MB/s Average


Samsung 830 256 Day 25

566,875.63 GiB
290.60 MB/s Average
566 Hours

2483 Wear Leveling Count
MWI 32

0/0 Erase/Program Fail

OCZ Vertex Turbo 64 Day 43

314,914.13 GiB
84.40 MB/s Avg
1066 hours

5629 Avg PE count
100 MWI

0/0/0 Read/Program/Erase Fail

[Anvil]

Endurance_cr_20120412.png

"WMI exhaustion" took it's time on the Vertex Turbo 64.
(GiB is about on par with the C300 64GB iirc)

----

Kingston SSDNow 40GB (X25-V)

865.46TB Host writes
Reallocated sectors : 05 41
Available Reserved Space : E8 99
POH 7775
MD5 OK

34.22MiB/s on avg (~46 hours)

The mini-retention test was not an issue. (12 hours 30 minutes)

[B.A.T]

Todays update:

Intel X25-M G1 80GB
499.68 TiB
22276 hours
Reallocated sectors : 72 to 78
Available Reserved space: 95
MWI=236 to 230
MD5 =OK
43.34 MiB/s on avg


m4
685.8411 TiB
2941 hours
Avg speed 74.42 MiB/s.
AD gone from 219 to 216.
P/E 11890.
MD5 OK.
Reallocated sectors : 00

[Christopher]

"WMI exhaustion" took it's time on the Vertex Turbo 64.
(GiB is about on par with the C300 64GB iirc)

True, but MWI for this drive should really be closer to 10,000PE-levels. Basically, twice the time to MWI exhaustion. For a contrast, I could always start running the 64GB Vertex EX, but I quite like that drive -- 50nm Samsung SLC and 100KPEs -- it's perhaps the finest incarnation of a Barefoot drive.

--------------
UPDATE -- ohh. The Octane popped some blocks sometime yesterday. It's reallocated sector count jumped from 0 to 4 and as a result MWI finally dropped from 100 to 97. My theory about reallocations and MWI seems to hold water. Since the MWI is only affected by bad blocks in the Octane, it's easier to see the relationship directly. But it was plain to see in the first Turbo I tried too.

It is somewhat troubling that it blew through 4 blocks at once, but I'd have to go back and check the logs to confirm.


OCZ is "looking into" the SMART data problems with the Octane and Petrol drives. Many users are suffering from the SMART data resetting and not working in general. Which is strange since there are so few attributes. I hope reallocations don't reset when the drive is power cycled like POH.

[canthearu]

Patriot Torqx-2 64GB - Day 57

Drive hours: 1,366 (854 Hours, 512 missing)
GiB written: 89,856 GiB (87.75 TiB)
Avg MB/s: 19.69 MB/s
MD5: OK

Bad blocks: 2 grown, 83 factory
Wear cycle counter: 5616 average erase cycles, 8611 maximum erase cycles.

I think I'll do a short retention test at 100TiB for this drive.

Intel 520 60GB - Day 48

Drive hours: 1,165
Avg MB/s: 92.63 MB/s
MD5: OK

Host GB written (F1): 365,016.50 GiB (356.46 TiB, 11680528 raw)
NAND writes (F9): 258,451 GiB (252.39 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 17

[Christopher]

Okay -- the Octane looks like it has a problem. Reallocations are up to 8, MWI is now 93. This does not bode well for the future.

Update --- make that 12 reallocations -- maybe it only goes up 4 at a time?. I think it still looks sketchy either way, but there really isn't much detail.

Here is a somewhat useless chart I made. But I can confirm reallocations occur in twos. MWI is 91 with 12 reallocations.



Now that I think about it, the Octane seems to be getting slower while the other two trend upward in terms of Avg MB/s.

[B.A.T]

Todays update:

Intel X25-M G1 80GB
503.34 TiB
22300 hours
Reallocated sectors : 78 to 85
Available Reserved space: 95 to 94
MWI=230 to 224
MD5 =OK
45.71 MiB/s on avg


m4
690.8655 TiB
2966 hours
Avg speed 72.96 MiB/s.
AD gone from 216 to 213.
P/E 11981.
MD5 OK.
Reallocated sectors : 00


My powercompany did some maintenance right after I went to work yesterday so both ssd got a 10h retention test and are working fine again after a restart of the test.

[canthearu]

Patriot Torqx-2 64GB - Day 57

Drive hours: 1,389 (877 Hours, 512 missing)
GiB written: 91,456 GiB (89.31 TiB)
Avg MB/s: 19.67 MB/s
MD5: OK

Bad blocks: 2 grown, 83 factory
Wear cycle counter: 5716 average erase cycles, 8772 maximum erase cycles.

Intel 520 60GB - Day 49

Drive hours: 1,188
Avg MB/s: 92.57 MB/s
MD5: OK

Host GB written (F1): 372,691.25 GiB (363.96 TiB, 11926120 raw)
NAND writes (F9): 263,889 GiB (257.70 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 15

Hmmm, I need another drive to kill, hopefully the 330 will be avaliable soon and not awful for killing.

[Christopher]

Well, well, well. The Octane is not pleasing.

Reallocation: 28
MWI : 75
irritation index: high

On the other hand, I estimate the Octane to have used 2500 PE cycles. It is the frequency, not the number of reallocs, which makes me doubtful that Le Octane has a bright future. Its high time one of the three did something though. Its over a PiB written to the three drives and almost 11,000PE cycles between the three.

[Ao1]

So much for NDurance.

[Christopher]

So much for NDurance.

ECC and wear leveling have a new name... NDurance.

In the past 5 hours, reallocations have now reached 44 and MWI is now 62. At this rate, this drive could wake up dead before I get back home. I hope that doesn't happen... it has a nasty habit of locking up the rig.



E9 is MWI
E8 is total count of write sectors (which does not and never has worked)
05 is Reallocations (which, happen two at a time somewhat like the 830)

If you look at the timestamp in the chart above, the first reallocations happened around 50 hours ago. It's not that it's a lot in the grand scheme of it all, but it usually signals that the end is near. We'll see how it plays out.

Currently at 288,980GIB

[Christopher]

I like this chart for showing MWI and reallocations

[canthearu]

This quick degradation and death seems to be a recurring theme with ocz octane and petrol drives.

Definitely a bit early though.

More research is required.

[Christopher]

54 reallcs now with the Octane. I've been checking it incessantly with my phone, and every time I look it's higher.

[canthearu]

54 reallcs now with the Octane. I've been checking it incessantly with my phone, and every time I look it's higher.

Graph shows a exponential relationship between time and bad sectors ... that is not good for the life of an ssd.

[B.A.T]

Todays update:

Intel X25-M G1 80GB
507.33 TiB
22325 hours
Reallocated sectors : 85 to 88
Available Reserved space: 94
MWI=224 to 220
MD5 =OK
46.47 MiB/s on avg


m4
697.0389 TiB
2991 hours
Avg speed 72.98 MiB/s.
AD gone from 213 to 210.
P/E 12084.
MD5 OK.
Reallocated sectors : 00

[B.A.T]

54 reallcs now with the Octane. I've been checking it incessantly with my phone, and every time I look it's higher.

The Octane might still live long despite mwi going down towards 1. But it's exciting to follow the quick fall

[canthearu]

Patriot Torqx-2 64GB - Day 58

Drive hours: 1,415 (903 Hours, 512 missing)
GiB written: 93,200 GiB (91.02 TiB)
Avg MB/s: 19.67 MB/s
MD5: OK

Bad blocks: 2 grown, 83 factory
Wear cycle counter: 5825 average erase cycles, 8934 maximum erase cycles.

Intel 520 60GB - Day 50

Drive hours: 1,213
Avg MB/s: 92.64 MB/s
MD5: OK

Host GB written (F1): 380,924.31 GiB (372 TiB, 12189578 raw)
NAND writes (F9): 269,724 GiB (263.40 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Raw Error Rate (8B): 118 normalised
Avaliable Reserved Space (AA): 100 normalised
Media Wearout Indicator (E9): 13

I've added raw error rate and avaliable reserved space as these may come into play later in the Intel's drives life. The Intel is finally reaching the final stretch of its MWI and it is a complete mystery to me how far this drive will go or how it will die.

[Christopher]

the Octane.... what can I say?

142 reallocations
MWI has gone to 0, back to 100, and is now sitting at 91.


The other drives are model citizens.

More chart pr0n:

[Christopher]

830 256GB
641,506.63GiB

OCZ Vertex Turbo
336,754.80GiB

OCZ Octane
301,334.65GiB

[Ao1]

the Octane.... what can I say?

142 reallocations
MWI has gone to 0, back to 100, and is now sitting at 91.

Interesting that MWI is based on relocations. Have you rebooted at any stage? If so did the relocations reset? Have you tried an MD5 test?

The amount of relocations is obviously not normal behaviour for Intel or Hynix NAND. Either the controller is mistakenly calling out relocations or something is having a remarkable impact on the NAND to induce this type of behaviour.

[johnw]

The amount of relocations is obviously not normal behaviour for Intel or Hynix NAND. Either the controller is mistakenly calling out relocations or something is having a remarkable impact on the NAND to induce this type of behaviour.

More likely a firmware bug than the quasi-9174 controller being at fault. Although if OCZ is running the controller at a higher clock rate, I suppose there could be some strange interaction between the controller and the NAND flash.