SSD Write Endurance 25nm Vs 34nm

[B.A.T]

Todays update:

m4
888.4214 TiB
3758 hours
Avg speed 74.48 MiB/s.
AD gone from 107 to 102.
P/E 15319.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
618.88 TiB
23070 hours
Reallocated sectors : 156
Available Reserved space: 90
MWI=124
MD5 =OK
42.06 MiB/s on avg


Intel X25-E 64GB
45.09 TiB
157-30=127 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI=99
MD5 =OK
101.49 MiB/s on avg

[canthearu]

Patriot Torqx-2 64GB - Day 92 - Not running

Drive hours: 2,036
ASU GiB written (previous run): 144,175.53 GiB (140.8 TiB)
ASU GiB written (current run): 0.00 GiB (0 TiB)
ASU GiB written (Total): 144,175.53 GiB (140.8 TiB)
Avg MB/s: 0 MB/s
MD5: OK

Bad blocks: 20 grown, 83 factory
Wear cycle counter: 9067 average erase cycles, 13875 maximum erase cycles.
Total ECC Error Count: 1
SATA PHY error count: 250

Intel 520 60GB - Day 84

Drive hours: 1,964
ASU GiB written: 625,572.45 GiB (610.91 TiB)
Avg MB/s: 95.40 MB/s
MD5: OK

Host GB written (F1): 630,028.75 GiB (615.26 TiB, 20160920 raw)
NAND writes (F9): 446,021 GiB (435.57 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Raw Error Rate (8B): 115 normalised
Avaliable Reserved Space (AA): 100 normalised
Media Wearout Indicator (E9): 1

Vertex 4 128GB - Day 31

Drive hours: 759 (556 + 203)
ASU GiB written: 329,433.46 GiB (321.71 TiB)
Avg MB/s: 125.83 MB/s
MD5: OK

Sector GB written (E8): 332,337.96 GiB (324.55 TiB, 486156078177 + 210807149746 raw)
Reallocated sectors (05): 1
Raw Read Error Rate (01): 6
Remaining Life (E9): 48 (78 current - 30)

Time to attempt to resurrect the Torqx 2 drive --- again!

[Anvil]

Kingston SSDNow 40GB (X25-V)

960.52TB Host writes
Reallocated sectors : 05 43
Available Reserved Space : E8 99
POH 8611
MD5 OK

33.22MiB/s on avg (~98 hours)

Endurance_cr_20120517.png

Endurance_PEsummary_20120517.png

The Vertex 4 is getting close to 3K which is spec for the NAND. (according to some review sites)

[Christopher]

A few weeks back, I tried to calculate how many PE cycles the V4's firmware associated with MWI. It looked as though the MWI was based on 7K PE cycles. Canthearu's V4 only had 100TB back then, so maybe I did something wrong. But the flash is only 3K rated.

[Anvil]

Based on MWI it looks to aim for ~5K.
(~2700 cycles and 48% remaining)

[Christopher]

Samsung 830 256GB Day 60

1,427,874.66GiB
1,394.41 TiB

282.48 MB/s Average
1415 Hours

6204 Wear Leveling Count
MWI 1

5/0 Erase/Program Fail
Used Reserved Block Count: 10/20480 sectors

[canthearu]

Well, MWI is going down a lot quicker for the V4 on 1.4RC.

Typical OCZ to cheap out on the NAND and not tell anyone.

[Christopher]

Well, MWI is going down a lot quicker for the V4 on 1.4RC.

Typical OCZ to cheap out on the NAND and not tell anyone.

It does seem to be decreasing faster on 1.4RC6. In the calculations I performed, the MWi was at 88 and I assumed 115GB per PE cycle. At the time, it worked out to around 7K PE cycles that MWI was based on. They may have tweaked it for RC6.

[stevecs]

So, if I'm reading the charts correctly, currently the highest endurance drives are Intel & Samsung?

Also, just to clarify, Anvil, does your utility do any crc checks of the data in the read cycle (i.e. will it catch any errors if the drive mis-writes or mis-reads them back to your application?) Basically seeing if we have the data here that would also help bound bit error expectations on the drives that are being tested.

[Christopher]

So, if I'm reading the charts correctly, currently the highest endurance drives are Intel & Samsung?

The second chart shows estimated current PE cycle count for presently running drives (not the past drives). So far Intel, Crucial, and Samsung are doing the best, but then, 6 of the 8 are made by those three companies. And they each own fabs, so it shouldn't be that surprising.

[Anvil]

Also, just to clarify, Anvil, does your utility do any crc checks of the data in the read cycle (i.e. will it catch any errors if the drive mis-writes or mis-reads them back to your application?) Basically seeing if we have the data here that would also help bound bit error expectations on the drives that are being tested.

There is a MD5 test which is being performed every n loops, default is every 5 loops.
I perform the test on a ISO image, about 2GB in size.
(the file doesn't sit on the drive, it's copied to the drive when testing is due, in case there is an error it is reported)

[canthearu]

Nope, the Torqx 2 is finished. While it did secure erase and I was able to push slightly more data to it ... windows event viewer is continually reporting write errors and ASU is unresponsive. I think it also crashed my endurance rig/HTPC last night.

My last status update is probably accurate enough for this drive. Add on 5TiB if you care.

Torqx 2 = dead and unloved

[Christopher]

Like my MTRONS and Turbos and Octane, it too is an also-ran. The Phison needs some work.

[Anvil]

The departure of the Torqx 2 was expected, not overly impressive.

I'm debating what to use as the next drive, the SF drives have been done almost to death and so I'm looking at something else.


There are some great offers on 128GB Petrol, Octane and m4, some of them have been done (Octane) and the m4 is currently being tested in 64GB capacity.

Late last night I started thinking about testing 3K drives, i.e. Intel 330 or the Kingston HyperX 3K, they are both interesting drives but then we are back on the SF controller...
(still got the Force 3 though, could do some other compression ratio testing...)

[B.A.T]

3K units are very interesting and the plextor m3/m3p would have been a good contestant.

[kaktus1907]

3K units are very interesting and the plextor m3/m3p would have been a good contestant.

Im looking forward to that to see what toshibas 24nm nands doing

[B.A.T]

Todays update:

m4
895.2967 TiB
3785 hours
Avg speed 74.44 MiB/s.
AD gone from 102 to 98.
P/E 15435.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
622.76 TiB
23097 hours
Reallocated sectors : 156 to 157
Available Reserved space: 90
MWI=124
MD5 =OK
42.03 MiB/s on avg


Intel X25-E 64GB
54.54 TiB
184-30=154 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI=99
MD5 =OK
101.46 MiB/s on avg

[canthearu]

Still deciding if I should throw a sandisk extreme 120gig into the mix .... they are quite inexpensive, and would be interesting to test the 24nm NAND.

[Anvil]

That is an interesting drive!
I was considering that one but the shop that sells them extremely cheap doesn't sell to businesses. (I won't get a tax refund)

Looks like the Intel 330 could be it!
Plextors are still too expensive here, the M3P (~$260) is more or less 2x the price of a SanDisk Extreme and about $50 more than most other drives like the Intel 520 (~$210), almost $100 more than an m4 ($165).

[Christopher]

I happen to not be using my 128GB M3P at the moment... It has host writes and MWI now...

Actually, I'm not using most of my drives. But I really want to test the M3P!

They're $210 here at the moment, but the 128GB 830 is only $109 this weekend, so if I were making a buying decision, I'd take the 830. Since I already own the M3P, the choice is clear.

Can anyone think of a good reason not to test it?

[Anvil]

I won't be entering a M3P so feel free

Agility 4s are starting to surface, wonder whats up with that vs Vertex 4?

Not much info on the Agility 4, this is all I have found on the 128GB (Agility 4 Max R/W 400 MB/ 300 MB sec)

[Christopher]

Samsung 830 256GB Day 61

1,452,304.28GiB
1,418.26 TiB

297.51 MB/s Average
1438 Hours

6311 Wear Leveling Count
MWI 1

5/0 Erase/Program Fail
Used Reserved Block Count: 10/20480 sectors

[canthearu]

Intel 520 60GB - Day 86

Drive hours: 2,011
ASU GiB written: 637,689.27 GiB (622.74 TiB)
Avg MB/s: 95.42 MB/s
MD5: OK

Host GB written (F1): 642,255.88 GiB (627.20 TiB, 20552188 raw)
NAND writes (F9): 454,676 GiB (444.02 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Raw Error Rate (8B): 108 normalised
Avaliable Reserved Space (AA): 100 normalised
Media Wearout Indicator (E9): 1

Vertex 4 128GB - Day 33

Drive hours: 801 (556 + 245)
ASU GiB written: 345,315.27 GiB (337.22 TiB)
Avg MB/s: 123.96 MB/s
MD5: OK

Sector GB written (E8): 348,370.44 GiB (340.21 TiB, 486156078177 + 244429690650 raw)
Reallocated sectors (05): 4
Raw Read Error Rate (01): 6
Remaining Life (E9): 44 (74 current - 30)

Hmm, 3 more bad blocks in the last 48h. Hopefully this vertex 4 isn't following the same path as the octane before it.

[Anvil]

Might not be a good idea to flash to 1.4 as SMART values are reset.

The next 48hours should tell if there is something going on, there shouldn't be that much in common with the Octane except for the casing.

[canthearu]

The next 48hours should tell if there is something going on, there shouldn't be that much in common with the Octane except for the casing.

Except for the same people writing the firmware!