SSD Write Endurance 25nm Vs 34nm

[Christopher]

BAT,

No, my FW updates went off without a hitch, I just burned the ISO files to CD or on USB and booted. In fact, I can even update the FW of the dead MTrON (the one I don't fry). I got my Fw updates from Imation.

[B.A.T]

I've taken a look at their homepage but where can I find the download page with fw?

[Christopher]

I've taken a look at their homepage but where can I find the download page with fw?

The last time I looked, MTRON didn't even have a site anymore.

[B.A.T]

I ment Imation homepage. The Mtron homepage is down wich makes it not easy to find anything useful.

[Christopher]

Were you trying to flash your MTRON with the Imation FW?

[Christopher]

Samsung 830 256GB Day 120

(GiB) 2,945,117
(TiB) 2,876
(PiB) 2.83


(Avg) 297.81 MB/s

(B1) Wear Leveling Count: 12,761

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 2925


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

[B.A.T]

Were you trying to flash your MTRON with the Imation FW?

Nope, I used the Mtron fw posted here.

I've been desperately trying to find the 0.18 or 0.19 ver to see if that will make a difference

[B.A.T]

Please find the 0.20r1 firmware links.

This firmware package support all Mtron SSD.

http://www.mediafire.com/?xw3ibgbp3bvbwuc

Hi! Do you by any chance have the 18r or 19r? The 20r returns an error and won't flash.

[Christopher]

I'll try flashing one of my Imations with it.

[B.A.T]

Good luck

[B.A.T]

Here are todays update:

m4
1239.4681 TiB
5174 hours
Avg speed 74.52 MiB/s.
AD 164 to 158
P/E 21325.
C3 4457
CE 58
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
802.43 TiB
24461 hours
Reallocated sectors : 38 to 73
Available Reserved space: 51 to 49
MWI= 194 to 184
MD5 =OK
31.37 Mi/s on avg


Intel X25-E 64GB
526.13 TiB
1642-30=1612 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI= 96
MD5 =OK
91.05 MiB/s on avg

[Anvil]

Intel 330 120GB

554.08TB Host writes
2.00TiB Host reads
Reallocated sectors : 05 20 +2
Available Reserved Space : E8 100
MWI 35
[B5] 10
[B6] 10
[F1] Total LBAs Written 18156165
[F2] Total LBAs Read 65675
[F9] Total NAND Writes 399966GB // ~390TiB
POH 1345
MD5 OK

125.19MiB/s on avg (~217 hours)

--

Kingston SSDNow 40GB (X25-V)

1129.12TB Host writes (36999097*32)
Reallocated sectors : 05 61 // 1 up
Available Reserved Space : E8 98
POH 10079
MD5 OK

33.43MiB/s on avg (~217 hours)

[Christopher]

I feel like the 830 should have more reallocations than just 9 at almost 3PB. But if we scale it down, it's a lot like the 64GB M4 with only .75PiB on it. Its even 1/4 the average speed of the M4.

[Christopher]

Samsung 830 256GB Day 121

(GiB) 2,969,816
(TiB) 2,900
(PiB) 2.85


(Avg) 298.05 MB/s

(B1) Wear Leveling Count: 12,867

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 2949


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

[canthearu]

Sandisk Extreme G25 - 120gb - Day 45

Drive hours: 1076
ASU GiB written: 465,183.49 GiB (454.28 TiB)
Avg MB/s: 125.50 MB/s (70.38 hours)
MD5: OK

Host GB written (F1): 467,742 GiB (456.78 TiB)
NAND writes (E9): 349,098 GiB (340.92 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 116 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 4 raw
Reported Uncorrectable Errors (BB): 0 raw

Been a hard week at work. Will be taking it out on a new SSD. Stay tuned!

[Anvil]

[B]
Been a hard week at work. Will be taking it out on a new SSD. Stay tuned!

I might start looking for another drive as well.

--

@christopher
Your Samsung has been performing better than expected, drives of such capacities are in a different league from smaller drives like 64GB drives.

I'm still thinking of creating an option for running at a reduced/fixed throughput and or combining with reading data on top of writing.

[canthearu]

Anvil, my sandisk extreme seems to be missing from the list of drives on the first page?

Can you fix?

[Anvil]

Ugh, will check whats happened when I get back home. (tonight)

edit:
Should be OK now, not sure how that could have happened.
(I do keep backups and that made it a quick fix)

[canthearu]

Ok, new drive time:

I was kind of disappointed in how the first vertex 4 went ... so it is time to kill another:

This time, I'm going to change the rules.

Drive settings:
Firmware version: 1.5
Overprovisioning: 60 GiB
Static Data: 16 GiB

I am overprovisioning the drive so it stays in performance mode permanantly. I think the constant switching between performance and storage mode, as well the garbage collection the switch caused, was making the drive use excessive write cycles.

There are many new factors that may influence the way this drive goes:

a) If the drive is using NAND in an SLC type fashion, then each erase does twice the damage over normal MLC, as each erase cycle would only be wiping one bit, rather then 2

b) The drive may take a lot less damage during writes if the NAND is being used in an SLC type fashion, as each cell is only written once each pass.

c) Reading the data may be easier if each cell is only written once. Thus the drive may last longer due to less ECC being needed.

How this ends may really be up to how the NAND behaves.

[Anvil]

That one should be interesting!
It might help on finding out what's going on "behind the scenes" on the V4.

Over-provisioning has been on my list of things to do for quite some time, to get to the interesting parts one needs 2 drives where OP is set differently. (both could be OPd)

Endurance_cr_20120720_1349.png

[sergiu]

Ok, new drive time:

a) If the drive is using NAND in an SLC type fashion, then each erase does twice the damage over normal MLC, as each erase cycle would only be wiping one bit, rather then 2

b) The drive may take a lot less damage during writes if the NAND is being used in an SLC type fashion, as each cell is only written once each pass.

c) Reading the data may be easier if each cell is only written once. Thus the drive may last longer due to less ECC being needed.

How this ends may really be up to how the NAND behaves.

How is possible to use it as SLC? is there any firmware configuration that allows the cells to be programmed in "SLC" mode?

[Zaxx]

Yes, in a roundabout way. There was an excellent summary and re-review of the 256GB V4 (w/FW 1.5) explaining how the performance/storage mode works as well as the mimic-ing of SLC on RWL but now the link just reloads the home page. :/ The new 1.5 was supposed to extend performance mode past the 50% barrier by tweaking garbage collection as well as give a little speed bump.

[Christopher]

How is possible to use it as SLC? is there any firmware configuration that allows the cells to be programmed in "SLC" mode?

It's more accurate to call it MLC-1 than SLC. When flash stores two bits, the first bit gets programmed fast while the second takes much longer. So if you just write to that first bit, you get similar performance to SLC.

[B.A.T]

Yes, in a roundabout way. There was an excellent summary and re-review of the 256GB V4 (w/FW 1.5) explaining how the performance/storage mode works as well as the mimic-ing of SLC on RWL but now the link just reloads the home page. :/ The new 1.5 was supposed to extend performance mode past the 50% barrier by tweaking garbage collection as well as give a little speed bump.

Tom's Hardware wrote something about this.

Link

[Zaxx]

Tom's Hardware wrote something about this.

Link

Actually that's where I saw it...musta had a toke or two.

The 256GB V4 re-review w/fw1.5 is still a good read if you can get to it...I informed him of the broken link.