SSD Write Endurance 25nm Vs 34nm

[devsk]

Well, in that case the M225 will go even farther with it's 51nm NAND (10k P/E).

The controller is not the same.

[SynbiosVyse]

0-Fill
110.64 hours
103.1149 TiB written
139.78 MB/s
MD5 ok

05: 0
B1: 29
E9: 41984
EA/F1: 106048
F2: 128

[Vapor]

0-Fill
110.64 hours
103.1149 TiB written
139.78 MB/s
MD5 ok

05: 0
B1: 29
E9: 41984
EA/F1: 106048
F2: 128

I really don't understand why you're testing with 0-fill.

Also, what's the MWI at?

[SynbiosVyse]

MWI: 76%

I misunderstood Ao1's post, I thought he wanted to see the results if I changed to 0-fill.

I changed back to 46% compression and the speeds are still 57.06 MB/sec. Basically the same as what it was before at higher MWI levels. From my understanding LTT has not kicked in yet.

[Ao1]

If LTT has not kicked in now I somehow doubt it's going to. Could you please confirm write speed if you switch to 100% (Incompressible)? This is a retail drive, yes?

[SynbiosVyse]

100% incompressible is about 41.13 MB/sec

As far as i know, it is a standard retail drive.

[Vapor]

I think from here on out, sticking with 100% incompressible (to mitigate the skew from 53TiB of 0-fill) might not be a bad idea....or at least until MWI zeros out. (then return to 46% or move to 67%)

[bluestang]

^^ This only applies to the SF-based drives ??

[Vapor]

^^ This only applies to the SF-based drives ??

Right

All other drives can use any setting they want, it won't make a difference.

[bluestang]

Wheew!! (wiping browe)...thought I screwed up there for a second!

[Anvil]

I have considered alternating the fill-value for each session, not that it makes much of a difference.

[Ao1]

So, 110 hours to get to ~75% = 440 hours to 0% or 18 days. LLT cannot have been set on this particular drive, so is it a peculiarity on this particular drive or do Corsair have a different policy on LTT?

[Vapor]

So, 110 hours to get to ~75% = 440 hours to 0% or 18 days. LLT cannot have been set on this particular drive, so is it a peculiarity on this particular drive or do Corsair have a different policy on LTT?

110 hours is just how long it's been since he changed to 0-fill setting. MWI only moved from 81 to 76 in that time...but NAND write speed was also very low (~18.9MiB/sec).

With the 46% setting (had nearly the same NAND write speed as 100%), his drive was on pace to hit MWI=1 after a total of 41.5 days.


C300 Update

247.522TiB host writes, 13 MWI, 4351 raw wear, 2048/1 reallocations, 62.95MiB/sec, MD5 OK


SF-1200 nLTT Update

128TiB host writes, 88.28TiB NAND writes, 58 MWI, 1412.5 raw wear (equiv), wear range delta 3, 56.05MiB/sec, MD5 OK

[B.A.T]

m4 update:

380.0042 TiB
1248 hours
Avg speed 90.19 MiB/s.
AD gone from 137 to 134.
P/E 6661.
MD5 OK.
Still no reallocated sectors



Kingston V+100:

21.1626 TiB
81 hours
Avg speed 76.51 MiB/s.
AD no change - 177.
P/E ?
MD5 OK.


The Kingston is a strange one.... No change on AD.....

I'll be gone for the weekend and the next update will be monday.

[bulanula]

Well, in that case the M225 will go even farther with it's 51nm NAND (10k P/E).

Not really because the Indilinx controller has a really high WA. Ideally, you would need NAND that is rated for many erase cycles like 50nm SLC in the Intel X25-E drive with a controller that is modern ( supports TRIM ) and has a really low WA like the Sandforce which uses compression and gets quite a low WA of around 0.6 or a controller like the Intel / Crucial ones with a WA of 1.1. Based on this theory, I would expect the C300 to outlast the M4 because the controller is almost the same between the M4 and the C300 while the C300 has better lasting 34nm NAND ( compared to the M4 with lesser 25nm NAND ).

[bluestang]

M225->Vertex Turbo 64GB Update:

145.27 TiB
540.5 hours
MWI 27 (drops by 1 for every 50 raw wear)
3678 Raw Wear
111.22 MB/s avg for the last 16.3 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count(Realloc Sectors) at 2 (Bank 6/Block 2406; Bank 3/Block 3925)

[SynbiosVyse]

Vapor: so you recommend I run it at 100% (incompressible) until MWI hits 0%?

I just changed it to 100% and the average write speed has dipped a little since yesterday. It's running at about 46.22 MB/sec with dips to 42.85 MB/sec or so. MWI is at 74% at the moment.

107.4214 TiB written according to app.

E7 normalized: 74%
05: 0
E9: 45376
EA/F1: 110528
B1: 31
MD5 still ok

[Anvil]

@SynbiosVyse

There is no point in running 0Fill on the SF controller, 46% or more should be fine.

--

264.79TB Host writes
Reallocated sectors : 6
MD5 OK

33.25MiB/s on avg (28 hours)

[Vapor]

C300 Update

264.1TiB host writes, 11 MWI, 4462 raw wear indicator, 2048/1 reallocations, 62.95MiB/sec, MD5 OK


SF-1200 nLTT Update

133.5TiB host writes, 92.594TiB NAND writes, 56 MWI, 1481.5 raw wear (equiv), wear range delta 3, 56MiB/sec, MD5 OK

[bulanula]

It seems like the Indilinx drive is indeed going quite far. I would not have expected that because of its really high WA. Maybe you upgraded to the new firmware version that is supposed to drastically lower WA with better algorithms or maybe TRIM really does help ?

[Computurd]

t seems like the Indilinx drive is indeed going quite far.

thank god! ive got 8 x 30GB vertex in RAID 0. I have been thinking they were close to dead for awhile! Obviously they should be fine

[SynbiosVyse]

100% (Incompressible)
28.70 hours
111.4313 TiB written total
40.80 MB/s
MD5 ok

05: 0
B1: 35
E7: 70%
E9: 50304
EA/F1: 114624
F2: 128

[bluestang]

It seems like the Indilinx drive is indeed going quite far. I would not have expected that because of its really high WA. Maybe you upgraded to the new firmware version that is supposed to drastically lower WA with better algorithms or maybe TRIM really does help ?

I think both, TRIM definitely helps the WA and the newer firmawres (M225 v1916 and 2030/Vertex v1.5 and 1.6) have better wear algorithms. The problem with that is they also added support for 34nm NAND which caused them to change P/E to 5k from 10k. If I had stayed on 1819 (or maybe 1916) then my MWI would have ran out around ~400TB

thank god! ive got 8 x 30GB vertex in RAID 0. I have been thinking they were close to dead for awhile! Obviously they should be fine

Have you (or can they be) upgraded to Turbos?

[Computurd]

there are 5 of them that i ordered the day they became publicly available. So im sure i cant upgrade those. the other 3 can be upgraded, but i keep them all as a matching set as it is a raid 0 volume. I havent used it in months tbh, it just sets there.
there will be a time that i get back to them though L:OL
they are all on the .6 firmware, so i think they should be fine as far as WA goes.

[Anvil]

@ct
Wishful thinking?
I bet you won't be booting off that old array unless it's for sentimental reasons
I've still got a few V1's and they are doing just fine! (fw 1.3 and 1.6)

--

271.52TB Host writes
Reallocated sectors : 6
MD5 OK

32.48MiB/s on avg (89 hours)