SSD Write Endurance 25nm Vs 34nm

[Christopher]

The Recovery time white paper discusses MLC-1 briefly.

[Christopher]

Samsung 830 256GB Day 122

(GiB) 2,995,490
(TiB) 2,925
(PiB) 2.87


(Avg) 298.77 MB/s

(B1) Wear Leveling Count: 12,978

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 2973


-------------------

[Anvil]

Intel 330 120GB

569.94TB Host writes
2.04TiB Host reads
Reallocated sectors : 05 20
Available Reserved Space : E8 100
MWI 34
[B5] 10
[B6] 10
[F1] Total LBAs Written 18675804
[F2] Total LBAs Read 66981
[F9] Total NAND Writes 411416GB // ~402TiB
POH 1381
MD5 OK

125.19MiB/s on avg (~253 hours)

--

Kingston SSDNow 40GB (X25-V)

1133.32TB Host writes (37136841*32)
Reallocated sectors : 05 62 // 1 up
Available Reserved Space : E8 98
POH 10116
MD5 OK

33.39MiB/s on avg (~253 hours)

--

Endurance_cr_20120721.png

I330_MWI_vs_RSC.png

edit:

I'll update the chart later today if there are any updates...

[canthearu]

Sandisk Extreme G25 - 120gb - Day 46

Drive hours: 1103
ASU GiB written: 476,913.77 GiB (465.74 TiB)
Avg MB/s: 126.09 MB/s (26.39 hours)
MD5: OK

Host GB written (F1): 479,554 GiB (468.31 TiB)
NAND writes (E9): 357,930 GiB (349.54 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 114 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 3 raw
Reported Uncorrectable Errors (BB): 0 raw

Vertex 4 - 50% OP - 128gb - Day 1

Drive hours: 31
ASU GiB written: 23,158.21 GiB (22.62 TiB)
Avg MB/s: 248.61 MB/s (26.28 hours)
MD5: OK

Host GB written (E8): 24,088.93 GiB (23.52 TiB, 50518137835 raw)

Raw Read Error Rate (01): 6 raw
Reallocated Block Count (05): 0 raw
Remaining Life (E9): 94 normalised

The second vertex 4 is going along quite a lot faster then the first one

[Christopher]

That is impressive Canthearu... I wonder how long it can keep that up?

[B.A.T]

Here are todays update:

m4
1252.2432 TiB
5224 hours
Avg speed 74.47 MiB/s.
AD 158 to 150
P/E 21540.
C3 4457
CE 58
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
807.84 TiB
24511 hours
Reallocated sectors : 73 to 116
Available Reserved space: 49 to 46
MWI= 184 to 169
MD5 =OK
31.39 Mi/s on avg


Intel X25-E 64GB
541.83 TiB
1692-30) 1662 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI= 96
MD5 =OK
91.10 MiB/s on avg

[canthearu]

That is impressive Canthearu... I wonder how long it can keep that up?

Not as good as the plextor m3p was (or my oem lite-on)

Those marvell+toggle nand drives are just unbeatable.

[Christopher]

Not as good as the plextor m3p was (or my oem lite-on)

Those marvell+toggle nand drives are just unbeatable.

That's probably true, but real MLC-1 also gets SLC like endurance. Perhaps only writing to one half of the drive will change the outcome this time?

[Christopher]

Samsung 830 256GB Day 123

(GiB) 3,020,675
(TiB) 2,949
(PiB) 2.90


(Avg) 298.62 MB/s

(B1) Wear Leveling Count: 3,086

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 2997


-------------------
We're closing in on 3PiB now. 48 more hours to 3000 TiB and another three days to 3PiB.

[AMD_Freak]

Not as good as the plextor m3p was (or my oem lite-on)

Those marvell+toggle nand drives are just unbeatable.

Plextor and lite-on are 1 in the same { PLDS}

[Zaxx]

Samsung 830 256GB Day 123

(GiB) 3,020,675
(TiB) 2,949
(PiB) 2.90


(Avg) 298.62 MB/s

(B1) Wear Leveling Count: 3,086

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 2997


-------------------
We're closing in on 3PiB now. 48 more hours to 3000 TiB and another three days to 3PiB.

Very impressive...the 830 has proven to be a very solid performer to say the least. I wish I had the extra resources to test 1 or 2 more large drives (240/256) and see just how big of a roll drive size plays in regards to endurance.

[B.A.T]

Here are todays update:

m4
1258.2423 TiB
5248 hours
Avg speed 74.41 MiB/s.
AD 150 to 147
P/E 21642.
C3 4457
CE 58
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
810.42 TiB
24535 hours
Reallocated sectors : 116 to 134
Available Reserved space: 46 to 45
MWI= 169
MD5 =OK
31.47 Mi/s on avg


Intel X25-E 64GB
549.24 TiB
1716-30=1686 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI= 96
MD5 =OK
91.15 MiB/s on avg

[zhadoom]

Plextor and lite-on are 1 in the same { PLDS}

PLDS = Philips Liteon Digital Solutions
http://www.pldsnet.com/

Interesting that Plextor are advertised at that site too.

I doesn't know of any relation between Plextor and PLDS.

[Christopher]

Samsung 830 256GB Day 124

(GiB) 3,046,610
(TiB) 2,975
(PiB) 2.92


(Avg) 299.14 MB/s

(B1) Wear Leveling Count: 13,198

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 3021


-------------------

[Anvil]

Intel 330 120GB

589.82TB Host writes
2.09TiB Host reads
Reallocated sectors : 05 22 +2
Available Reserved Space : E8 100
MWI 31
[B5] 11
[B6] 11
[F1] Total LBAs Written 19327088
[F2] Total LBAs Read 68626
[F9] Total NAND Writes 425764GB // ~416TiB
POH 1427
MD5 OK

125.20MiB/s on avg (~299 hours)

--

Kingston SSDNow 40GB (X25-V)

1138.59TB Host writes (37309611*32)
Reallocated sectors : 05 65 // +3
Available Reserved Space : E8 98
POH 10162
MD5 OK

33.35MiB/s on avg (~299 hours)

--

Endurance_cr_20120723.png

[canthearu]

Hmmm, I seem to be getting some performance degredation ... something triggered the drive to shift to storage mode rather then performance mode. The attached picture shows how each run starts fast at about 400meg per sec, then drops to around 200meg per sec.

Will secure erase and see if that fixes it!

[sergiu]

Probably the drive just needs extra overprovisioning. Those 60GB are enough just to make the drive enter in MLC-1 mode. To keep it there, I would guess you would need to use even more over provisioning, like 70-75GB.

[canthearu]

Secure erase fixed it:

[Anvil]

At what point did it start slowing down?
(history might repeat itself)

[Kain665]

Sandisk Extreme G25 - 120gb - Day 46

Drive hours: 1103
ASU GiB written: 476,913.77 GiB (465.74 TiB)
Avg MB/s: 126.09 MB/s (26.39 hours)
MD5: OK

Host GB written (F1): 479,554 GiB (468.31 TiB)
NAND writes (E9): 357,930 GiB (349.54 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 114 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 3 raw
Reported Uncorrectable Errors (BB): 0 raw

Vertex 4 - 50% OP - 128gb - Day 1

Drive hours: 31
ASU GiB written: 23,158.21 GiB (22.62 TiB)
Avg MB/s: 248.61 MB/s (26.28 hours)
MD5: OK

Host GB written (E8): 24,088.93 GiB (23.52 TiB, 50518137835 raw)

Raw Read Error Rate (01): 6 raw
Reallocated Block Count (05): 0 raw
Remaining Life (E9): 94 normalised

The second vertex 4 is going along quite a lot faster then the first one

Interesting to see what the V4 128 in SLC mode can do...

[B.A.T]

Here are todays update:

m4
1264,4222 TiB
5272 hours
Avg speed 74.35 MiB/s.
AD 147 to 144
P/E 21747.
C3 4457
CE 58
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
813.02 TiB
24559hours
Reallocated sectors : 134 to 153
Available Reserved space: 45 to 44
MWI= 169 to 152
MD5 =OK
31.39 Mi/s on avg


Intel X25-E 64GB
556.88 TiB
1740-30=1710 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI= 96
MD5 =OK
91.16 MiB/s on avg

[Christopher]

Samsung 830 256GB Day 125

(GiB) 3,071,152
(TiB) 2,999
(PiB) 2.95


(Avg) 298.87 MB/s

(B1) Wear Leveling Count: 13,304

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 3045


-------------------
The drive actually has 3012 TiB in LBA writes. I lost 13 TIB in ASU along the way.

[Anvil]

Intel 330 120GB

600.30TB Host writes
2.12TiB Host reads
Reallocated sectors : 05 24 +2
Available Reserved Space : E8 100
MWI 30
[B5] 12
[B6] 12
[F1] Total LBAs Written 19670569
[F2] Total LBAs Read 69534
[F9] Total NAND Writes 433332GB // ~423TiB
POH 1451
MD5 OK

125.21MiB/s on avg (~324 hours)

--

Kingston SSDNow 40GB (X25-V)

1141.37TB Host writes (37400662*32)
Reallocated sectors : 05 65
Available Reserved Space : E8 98
POH 10186
MD5 OK

33.34MiB/s on avg (~324 hours)

--

[canthearu]

Sandisk Extreme G25 - 120gb - Day 49

Drive hours: 1173
ASU GiB written: 506,710.51 GiB (494.83 TiB)
Avg MB/s: 125.81 MB/s (7.19 hours)
MD5: OK

Host GB written (F1): 510,518 GiB (498.55 TiB)
NAND writes (E9): 381,058 GiB (372.13 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 109 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 3 raw
Reported Uncorrectable Errors (BB): 0 raw

Vertex 4 - 50% OP - 128gb - PAUSED

Drive hours: 101
ASU GiB written: 61,811.98 GiB (60.36 TiB)
Avg MB/s: 0 MB/s (0 hours)
MD5: OK

Host GB written (E8): 62,941.02 GiB (61.47 TiB, 131996886814 raw)

Raw Read Error Rate (01): 5 raw
Reallocated Block Count (05): 0 raw
Remaining Life (E9): 84 normalised

I can't seem to keep the vertex 4 out of storage mode. Secure wipe only helped for half an hour or so, then performance dropped again. I'll try a firmware reflash followed by another secure wipe. See if that helps.

Strange thing is that it ran fine for 3 days .....

If I can't get this working right, I may need to abandon breaking this V4 ... which would mean I would need to go find another victim in its place.

[Anvil]

So, having secure erased it performs as expected in benchmarks?

Can you get it to perform in benchmarks once it has entered the so called storage-mode. (by normal means, like, i.e. by trimming a large file,...)