SSD Write Endurance 25nm Vs 34nm

[Anvil]

@Christopher

I've had BSOD's, not many though (2 o 3) in about 6 months and only on specific drives.

There really was just one problematic drive and it was an early Agility 3, it had all the issues and so it was returned.
(besides a few BSOD's the really annoying issue was that it wasn't detected on cold boots)

The GT had a BSOD-like shutdown, while working the computer stopped responding, it suddenly started logging off and in the midst of logging off it BSOD'd. (no data lost though)
In case of the GT I had forgotten to apply the LPM fix and it has not happened since.

So, I've been bitten but I don't really count the GT as a problem as the LPM fix was and still looks to be needed for modern Intel MB's so make sure to apply it.
(the LPM fix was also needed for other drives like the M4 during early FW's)

I do like the Samsung drives, they work really well on laptops and so I'm currently using a few in a MacBook, with great success.

edit:
I've got a lot of SF-2 based drives, not just the Corsairs.
Quite a few Vertex 3's, Vertex 3 MAX IO, Patriot WildFire (toggle mode), they all are 240GB drives and none of them have caused any issues, so far.
I've had the V3's since they were made available, so about 6 months.

[Christopher]

I think Intel toolbox's automatic Windows optimizations automatically disables LPM, but I'll double check. Looking through Mushkin's support forum didn't yield too many real problems. In particular, many Mushkin users don't seem to grasp the difference between SATA III AND SATA II, nor between incompressible and compressible. That tells me there aren't too many real problems. Incidentally, there don't seem to be any 60GB deluxes out in the wild yet. I don't think they're making many of these things anyway.

[bulanula]

I'm not saying that's what they did, just that they could have. It's different to get a drive from RMA which may be used (because somewhere, in the fine print, they make this fact known), but it's another thing entirely to sell as new drive which may be used. When you do that, it's called REFURBISHED.

LOL. OCZ gets better and better at this kind of dodgy stuff. First the 25nm bait and switch. Then Spectek. Now this. EPIC business practice / profits and screw the consumer is too small a word

Would love to be a OCZ shareholder. Their tactics are quite successful so long as the media is paid not to make a big hype about it etc. Pretty good PR. It is obvious that the drives are not new. Imagine getting a brand new Commodore 64 today = impossible. Just like my 5870s = all sold as "NEW OEM" and none were indeed new. LOL

[Christopher]

just to be fair to OCZ, I only speculate that this could be going on, but it should be said that there are very few drives like mine. The one place that had 120 Turbos in stock only had 61 of them @$130 each. Maybe they keep some around for RMAs at OCZ, and decided to liquidate remainder stock. Or maybe they just keep some Indilinx + Intel nand around because they can be Agility or Vertex. Who knows? They did give some refunds for people who purchased JMicron based Core drives, so they've stepped up in the past. In fact, of the Turbo and Agility, the Agility must be newer, because OCZ had a problem with the new style plastic shells being out of spec and not fitting in some laptops. My new agility has the correct size plastic shell, while the Turbo doesnt . OCZ might be operating above board, and I'm more upset about getting the plastic chassis than getting possibly used drives.

[Christopher]

Has anyone ever used a Vertex LE?

They're even cheaper than the Turbo at the 100GB capacity, which I believe is 128 with a lot more spare area.

As long as they're using 34nm NAND and not 25nm, I can't see how it'd be a bad deal.

[Computurd]

I am using a Vertex LE with the extended capacity firmware. no problems from it, ever.

[bluestang]

M225->Vertex Turbo 64GB Update:

335 TiB (368.33 TB) total
973.13 hours
7203 Raw Wear
110.68 MB/s avg for the last 65.15 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count (Realloc Sectors) at 4
(Bank 6/Block 2406; Bank 3/Block 3925; Bank 0/Block 1766; Bank 0/Block 829)



I'm really impressed on the write speed of this old Indilinx drive from beginning to end of loop...
Beginning of loop:


Middle of loop:


End of loop

[Christopher]

I am using a Vertex LE with the extended capacity firmware. no problems from it, ever.

You think the 100GB/128GB version (93 after over provisioning) is worth $109?

[Christopher]

M225->Vertex Turbo 64GB Update:

335 TiB (368.33 TB) total
973.13 hours
7203 Raw Wear
110.68 MB/s avg for the last 65.15 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count (Realloc Sectors) at 4
(Bank 6/Block 2406; Bank 3/Block 3925; Bank 0/Block 1766; Bank 0/Block 829)

I'm really impressed on the write speed of this old Indilinx drive from beginning to end of loop...
Beginning of loop:

I ran my 60GB Agility for about 8 hours, and it wasn't nearly so fast, but still fast in it's own right, and still fast for a 64GB drive:



The big difference is the move from Samsung to Intel NAND.

[bulanula]

Really amazing that the Indilinx just keeps on going. I expected it to have died by now etc.

[Christopher]

Really amazing that the Indilinx just keeps on going. I expected it to have died by now etc.

With TRIM, I'd expect the M225 -> Vertex Turbo to last as long as (at least) as the M4.

The only problem with the Indilinx controller is poor 4K performance. OCZ claims they're going to release their Arrowana FW for Indilinx's with Intel NAND, and that 4K performance will be increased 500%.

I think that same poor 4K performance will end up making the Turbo last longer in this test.

[Anvil]

Kingston SSDNow 40GB (X25-V)

329.5TB Host writes
Reallocated sectors : 7
MD5 checking is disabled until later tonight. (moving to a new rig)

34.15MiB/s on avg (~26 hours)

[bulanula]

The Intel X25-V keeps on going as well but it really is writing very slowly etc. Funny the Indilinx already passed it.

[Christopher]

I just noticed something about the M225 --> Vertex Turbo, something I've noticed with other Indilinx drives: The difference between maximum and minimum erases.

Intel made a big deal about this in 2008:





It would be more helpful if Intel's SMART attributes had min and max erases as well, so there's no way to really know.

[bluestang]

I can give you a .csv file showing all my blocks and their current erase count. I can even spit out total erase count for each Bank. Want them?

[Anvil]

The Force 3 just changed to 99% "Life left"

SSDLife now estimates that it will last for 6 months 22 days. (April 08, 2012)

SSDLife_2011-09-19-22-06.png

[Christopher]

I can give you a .csv file showing all my blocks and their current erase count. I can even spit out total erase count for each Bank. Want them?

How do you do that?
I was just looking at the SMART data...

I noticed that your avg. was very close to your maximum (235/7574/7203avg). Mine were a lot worse before putting down some endurance writes. My drive's averages are sitting at ~550 min, ~4000 max, and ~1700 avg.

I think TRIM makes a big difference in the wear average with Indilinx drives, which makes sense. Mine was used for most of it's life in a Vista x86, Dell Latitude D630 system, which is how I explain the much greater WA and block wear differential.

[Ao1]

The Force 3 just changed to 99% "Life left"

SSDLife now estimates that it will last for 6 months 22 days. (April 08, 2012)

The V3 incurred around 11TB of nand writes before it dropped to 99% and I think the un-throttled V2 was around 6.1TB. It’s looking likely that the F3 does not have LTT. (I'm guess you are around 6TB of NAND writes?).

[Anvil]

Raw NAND writes are at about 15TB. (Host writes are at ~20TB)

The 46% compression setting looks to result in about ~74% NAND writes (~26% reduction) when I last checked.

We had a power outage so it's been off for almost 2 hours, I might end up starting the test tomorrow morning (in about 7-8 hours) as I'm in the middle of moving the drive to the other computer.

I'm almost confident that there is no LTT on the Corsairs (and a few other's as well)
Well, we'll know for sure in a week or so.

[SynbiosVyse]

441.59 hours
189.7459 TiB written
40.85 MB/s
MD5 ok

05: 0
B1: 102
E7: 10%
E9: 146240
EA/F1: 195264
F2: 320

[bulanula]

Good to see the Corsair almost crossing 200TB.

[bluestang]

How do you do that?
I was just looking at the SMART data...

I noticed that your avg. was very close to your maximum (235/7574/7203avg). Mine were a lot worse before putting down some endurance writes. My drive's averages are sitting at ~550 min, ~4000 max, and ~1700 avg.

I think TRIM makes a big difference in the wear average with Indilinx drives, which makes sense. Mine was used for most of it's life in a Vista x86, Dell Latitude D630 system, which is how I explain the much greater WA and block wear differential.

Links below for ya. First one is the data for my M225. The second is a program called SmartViewer for use with Indilinx drives.

M225-VertexTurbo_Erase_Counts.zip
SmartViewer_2_21.zip

Keep in mind that the mininum erase can be very low due to block failing (CDI attrib. C3 and C4) at a low erase count. Also, some blocks show zero erases because they are intitial bad blocks (B8).

[Christopher]

Thanks, Bluestang.

That's pretty awesome.

The mushkin didn't make it here today, will be here in about 20 hours.


Those things rare as hen's teeth @60GB. Newegg had some for a couple days, but are sold out, and direct from mushkin may be the only place to get them now that they're up on mushkin.com.

[Anvil]

Kingston 40GB
330.38TB Host writes
Reallocated sectors : 7

(it's been offline most of the day, still doing some tests on the new Endurance test-rig)

--

Note

There is a new beta that can be downloaded. Link (or from the ASU thread in this forum)
(it expires late November)

[Anvil]

The Corsair has been running most of the day.

E6 100 (Life Curve sStatus)
E7 98 (SSD Life left)
B1 13 (Wear Range Delta)
E9 20060 (Raw writes)
F1 26748 (Host writes)

SSDLife_2011-09-21-01-15.png SSDLife_SMART.png