SSD Write Endurance 25nm Vs 34nm

[canthearu]

I should note that in this test, the chances of a SF becoming RO are about the same as me becoming the Prime Minister of Japan. It's impossible and inconceivable. The chances of any drive becoming RO are slim at best. Maybe two or three of all drives tested in a year.

Yep, I certainly agree with this statement!

[canthearu]

Patriot Torqx-2 64GB - Day 15

Drive hours: 358
GiB written: 19486.05 (19.0295 TiB)
Avg MB/s: 16.89 MB/s

Bad blocks: 83
Wear cycle counter: 0/1262/1898

Intel 520 60GB - Day 6

Drive hours: 166
Avg MB/s: 75.18 MB/s

Host GB written (F1): 40,150.16 GiB (39.21 TiB , 1284805 raw)
NAND writes (F9): 28,396 GiB (27.73 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 94

[canthearu]

Looks like my AV might have been slowing ASU down a bit.

Getting much faster speeds after disabling it.

[Anvil]

You should disable AV scanning on the test folder, it makes a huge difference in speed. (also makes a difference to cpu utilization)

[Ao1]

Well, I didn't expect any really huge differences between the intel 520 and other sandforce drives.

What will be interesting to see is:
a) Where MWI=1 happens, which I should be able to predict soon.
b) If the device dies like other sandforce drives tested here do, at about 2x of MWI in a horrible flaming insta-death, or if it manages to hang on like the m4 does.

Haven't seen a 60gig sandforce drive hit 700TiB like we have with the m4, would be nice to see if it were possible!

If you look at the at the 320 product manual it explains how SMART metrics can be used to calculate the amount of wear incurred over a duration of time. (I assume it is the same for the 520).

To establish this you need to reset the timers, which can be done with smartmon, or more conveniently with HDSentinel. As ASU is more or less a consistent workload it should be possible to reset the timers, establish the wear percentage factor and then accurately project the amount of writes to get to MWI 0.

Anand did an article about it here:
http://www.anandtech.com/show/5518/a...f-intel-ssds/6

Intel also covered it in a podcast but I can't put my finger on it just now.

[bulanula]

Indeed. This thread has shown us a couple of things till today :

-seems like the read only SSD is a total MYTH so people should have backups

-SF SSDs are not really reliable for endurance testing

The question now is "Has Intel done something good with the 520 ( worth paying the price premium over competing drives ) or not" ?

[canthearu]

You should disable AV scanning on the test folder, it makes a huge difference in speed. (also makes a difference to cpu utilization)

Yes it does, the intel is now averaging 90 meg per sec (with peak at 96meg per sec or so)

[canthearu]

If you look at the at the 320 product manual it explains how SMART metrics can be used to calculate the amount of wear incurred over a duration of time. (I assume it is the same for the 520).

I've already used smartlog to determine:

The spacing of the fall in MWI is quite consistant.
I lose a MWI value for every 4200 GiB of host data written to the SSD.
I lose a MWI value for every 3000 GiB of NAND writes to the SSD.

This means I will exhaust MWI at about 430 TiB host data (305 TiB NAND data)

I'll post the smart log of my intel drive once MWI reaches 90, which won't be long at this rate

[B.A.T]

Todays update:
Kingston V+100
Finally home and I can't get life in the Kingston. I've tried 3 cold boots but no response. I'll try to see if I can get it up in another rig tomorrow.

Intel X25-M G1 80GB
388,9180 TiB
21333 hours
Reallocated sectors : 00
MWI=98 to 94
MD5 =OK
41.39 MiB/s on avg


m4
501.0616 TiB
1997 hours
Avg speed 63.53 MiB/s.
AD gone from 85 to 63.
P/E 8793.
MD5 OK.
Reallocated sectors : 00


I'm home for a couple of days so I'll try a retention test on the rest.

[canthearu]

Patriot Torqx-2 64GB - Day 16

Drive hours: 383
GiB written: 21,158.20 GiB (20.6624 TiB)
Avg MB/s: 19.35 MB/s

Bad blocks: 83
Wear cycle counter: 0/1367/2060

Intel 520 60GB - Day 7

Drive hours: 191
Avg MB/s: 90.67 MB/s

Host GB written (F1): 47,965.41 GiB (46.84 TiB , 1534893 raw)
NAND writes (F9): 33,931 GiB (33.14 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 92

[B.A.T]

And it's confirmed. The Kingston is dead after 6 months of strugle. I've tried my other rig and with a usb-SATA adapter on both my laptops but with no luck. It won't connect and that is the end of testing. I have no evidence of what caused the death but my guess is a faulty controller which have been bugging me during the test.

The last update was:
Kingston V+100
368.6335 TiB
2343 hours
Avg speed 79.58 MiB/s
AD still 1.
168= 1 (SATA PHY Error Count)
P/E?
MD5 OK.
Reallocated sectors : 00

[Christopher]

B.A.T.

Good riddance.

I think the MTRON 7000 PRO died from a faulty IDE to SATA controller. I was able to flash the FW (Imation was able to get it to me) to the latest, but to no avail. The drive isn't dead... It just doesn't work.


To my knowledge, every drive will create an internal panic log when it panic locks. So they'd know at the factory when a drive panics, which is undoubtedly how most drives die. In the end, most drives just can't handle lots of bad news at once, methinks.

[canthearu]

I thank B.A.T as well for the test and I also want to lament that the Toshiba controllers don't have any real NAND wear information.

I have an old SSDNOW V+ in my laptop and it bugs me that it says nothing important.

[B.A.T]

It was supposed to be the value leader but they saved probably a little to much and sacrificed good RR/RW 4K speed and bought a controller with little smart support and unstable behaviour.

But I like the fact that I don't need to restart the rig every 24-28 hours anymore

No we are 6 hours into the first retention test for the rest of the pack. It'll be finished in another 6 hours. I'll try 24h to morrow.

[Christopher]

Vertex Turbo Day 5 Lazy Update
Vertex Turbo 64GB Day 5

35440.28 GB
82.45 MB/s

MWI 88

0 Erase Failure
0 Program Failure
0 Write Failure

Avg Erase Count
635

122 Hours



1 average PE cycle for every 55.8GB.
MWI is 88, or -1 MWI for every 50 aPE count.

I wish i had used the SMART VIEWER on the first Turbo, as I think one die (or one plane of one die) was faulty. Either way, this new one is much better I think, in terms of numbers you do not want to see on a new drive (like bad blocks after the factory burn in). The last one was doomed from the start.

[canthearu]

My Intel 520 has hit 90 MWI, logs in CSV of its smart values in progression to this point can be downloaded from

http://canthearu.com/intelssd520log.zip

[Christopher]

My Intel 520 has hit 90 MWI, logs in CSV of its smart values in progression to this point can be downloaded from

http://canthearu.com/intelssd520log.zip

Looks pretty stable and predictable. If you can ever get around to it, check out the workload timer. You can use smartmontools or better yet, HD Sentinel, to reset the timer. Then you can divide the number by 1024 and get the percentage of wear over the period timed (you have to run it for at least one hour -- it does work on G1 Intels, like the X25-Es). I don't know whether the workload calculator just uses MWI or some deeper metric when presenting the percentage of wear over the timed period, but I think it would be interesting either way.

[canthearu]

Patriot Torqx-2 64GB - Day 17

Drive hours: 406
GiB written: 22,715.54 GiB (22.1831 TiB)
Avg MB/s: 19.42 MB/s

Bad blocks: 83
Wear cycle counter: 0/1465/2217

Intel 520 60GB - Day 8

Drive hours: 213
Avg MB/s: 90.90 MB/s

Host GB written (F1): 55,272.56 GiB (53.98 TiB , 1534893 raw)
NAND writes (F9): 39,108 GiB (38.19 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 90

[B.A.T]

Todays update:
The 12h retention test is over and everything is OK. I'll start the 24h tonight.

Intel X25-M G1 80GB
393,0694 TiB
21360 hours
Reallocated sectors : 00
MWI=94 to 92
MD5 =OK
40.80 MiB/s on avg


m4
507.8998 TiB
2024 hours
Avg speed 64.88 MiB/s.
AD gone from 63 to 60.
P/E 8907.
MD5 OK.
Reallocated sectors : 00

[Christopher]

Vertex Turbo 64GB Day 6

41029.83 GB
82.46 MB/s

MWI 86

0 Erase Failure
0 Program Failure
0 Write Failure

Avg Erase Count
735

141Hours

[Anvil]

Kingston SSDNow 40GB (X25-V)

764.26TB Host writes
Reallocated sectors : 05 27
Available Reserved Space : E8 99
POH 6866
MD5 OK

33.22MiB/s on avg (~419 hours)

No further development on reallocated sectors.

[canthearu]

Patriot Torqx-2 64GB - Day 18

Drive hours: 432
GiB written: 24,454.69 GiB (23.8816 TiB)
Avg MB/s: 20.20 MB/s

Bad blocks: 83
Wear cycle counter: 0/1576/2389

Intel 520 60GB - Day 9

Drive hours: 239
Avg MB/s: 91.86 MB/s

Host GB written (F1): 63,403.31 GiB (61.92 TiB , 2028906 raw)
NAND writes (F9): 44,872 GiB (43.82 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 88

I've tested the timed workload function: In 1520 minutes, I have consumed 1.8389% of the drives MWI. Seems to work.

[Anvil]

Endurance_cr_20120305.png

I'll remove the V100+ later tonight.

[bulanula]

Don't know about you but I am sick of controllers like Toshiba or older Samsung drives.

No SMART data at all

[Christopher]

The 520 should take another 45 days to hit MWI 0.