SSD Write Endurance 25nm Vs 34nm

[Ao1]

I have projected Anvil’s MWI to 10 to get the following:

SymbiosVyse 25nm Corsair 40GB
Data written to MWI 10 (Expiry of P/E cycles) No static data, disproportionate wear between least and most worn blocks, data switched from 0 fill to 100 fill.
Host writes = 191GB/ Nand Writes = 143MB

Christopher – 34nm Muskin Chronos 60GB 46% fill, static data
Host writes increase over SymbiosVyse = 11.5%
Nand write increase over SynbiosVyse = 14.7%

Anvil – 25nm Corsair F3 120GB 46% fill, static data
Host writes increase over SymbiosVyse = 127.2%
Nand write increase over SynbiosVyse = 131.5%

See Anvil's post below for charts

[Anvil]

Kingston SSDNow 40GB (X25-V)

448.92TB Host writes
Reallocated sectors : 12

MD5 OK

33.68MiB/s on avg (~40 hours)

--

Corsair Force 3 120GB

01 86/50 (Raw read error rate)
05 2 (Retired Block count)
B1 69 (Wear range delta)
E6 100 (Life curve status)
E7 20 (SSD Life left)
E9 301049 (Raw writes)
F1 400714 (Host writes)

MD5 OK

103.08MiB/s on avg (>275 hours) ~97TiB Host writes during this session.

power on hours : 1181

No movement on B1 this time.

--

Thanks for the charts Ao1

Looks like my drive is more consistent than SynbiosVyse's but we do know that he played with compression.
One thing that distinguishes "my" drive from the other SF drives is that it has RAISE, not sure what to make out of that difference, yet.

Do you mind if I put them on a server and re-post them as images?

[Anvil]

I've uploaded the charts Ao1 made for better accessibility/visibility

[bluestang]

M225->Vertex Turbo 64GB Update:

664.31 TiB (730.42 TB) total
1749.87 hours
13197 Raw Wear
118.88 MB/s avg for the last 18.87 hours (on W7 x64)
MD5 OK
C4-Erase Failure Block Count (Realloc Sectors) at 11.
(1=Bnk 6/Blk 2406; 2=Bnk 3/Blk 3925; 3=Bnk 0/Blk 1766; 4=Bnk 0/Blk 829; 5=Bnk 4/Blk 3191; 6=Bnk 7/Blk 937; 7=Bnk 7/Blk 1980; 8=Bnk 7/Blk 442; 9=Bnk 7/Blk 700; 10=Bnk 2/Blk 1066; 11=Bnk 7/Blck 85)

[Christopher]

I'm not sure that the Mushkin doesn't have RAISE. I know most 60's weren't supposed to have it, but given the available capacity of 55GB and the total NAND of 64GB, I think it has it.

Incidentally, I believe the Chronos Deluxe 60 has been discontinued. Here in the US, you can buy a 120GB toggle NAND model for $180 to $220, but the 60GB toggle NAND model is $140. That's not a great deal, and I think once retailer stock ends you won't see them anymore.

The Mushkin has been running for exactly 10 days straight as of a minute ago, 240hours
That's not bad, as uninterrupted I pull down 10.5 to 10.8TB a day.

[Christopher]

I just got another retired block, but without a corresponding program or erase failure.

Retired block count = 5

[Christopher]

I just got another retired block, but without a corresponding program or erase failure.

Retired block count = 5

[Ao1]

When Intel talk about relocated sectors are they talking about pages or a block? One_Hertz has 7,257 relocations and 25nm NAND. If a page = a sector that equates to 56.7MiB. If a block = a sector that would equate to 14.2GiB.

34nm = 4kiB page & a 0.5MiB block, so a retired block count of 5 = 2.5MiB. If the Chronos has 9GiB spare area and RAISE is deactivated it looks like it will have some way to go yet.

JESD218 states (for endurance) that the SSD must:

The SSD maintains its advertised user capacity.
The SSD maintains the required UBER for its application class
The SSD meets the required functional failure requirement (FFR) for its application class.
The SSD retains data with power off for the required time for its application class. (1 year for client, 3 months for enterprise)

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 44
3.3.2FW

05 5
Retired Block Count

B1 12
Wear Range Delta

F1 434195
Host Writes

E9 334988
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 130.18MB/s Avg
Intel RST drivers, Celeron G530 Biostar TH67+

1023 Hours Work
Time 1 month 12 days 15 hours

12 GiB Minimum Free Space

[Computurd]

i sincerely appreciate your new charts, but wondering if there is a normalized value chart? that would help my brain to wrap around the data a bit better...
very hurt at the loss of Vapor on the project, i hope that everything is fine, his health and that of his family as well.
Is there any type of communication or is he incommunicado? If we could at least get a PM shot his way maybe he could send the graph files that he made so that they can be updated by others...i mean damn his charts are the best i think ive seen.

Vapor=HELP! We miss you!

[sergiu]

When Intel talk about relocated sectors are they talking about pages or a block? One_Hertz has 7,257 relocations and 25nm NAND. If a page = a sector that equates to 56.7MiB. If a block = a sector that would equate to 14.2GiB.

The drive has somewhere between 6.5-8GiB spare area. But spare area indicator was calibrated to go to 1 for 2.74GiB which is the difference between GiB and GB. If a sector is a page, then it makes no sense to decrease MWI to 1 so fast. But if one reallocation is a block, then it makes sense for 512KiB blocks, which scales perfectly, as there were ~5600 reallocations when MWI hit 1. Could it be that drive has actually 34nm NAND? or this was already physically checked?

[Ao1]

^ Intel 25nm NAND is spec’d at 5K P/E cycles. It makes sense that a relocation would be a block rather than a page, but if that is the case the capacity of the drive must have been reduced.

I have emailed Vapor. Let’s see if he replies. I hope all is well. If we have not heard from him by next week I think someone else should take over the charts until such time as Vapor returns. I could host the excel charts on my Live Sky Drive account, which means anyone can get access to them. If anyone testing has excel charts that they would like to post send me a pm and I will give you my email address.

[One_Hertz]

I fired up my rig again last night after several days of it staying off to check data retention. All is well. I have yet to get to the samsung because I don't seem to have a 5-6 hour block of time free to play with it

534TB, 7298 reallocated sectors.

[sergiu]

^ Seems that reallocation rate is decreasing... Could you post the values of all SMART parameters, I am interested more in Erase Fail Count. If reallocation sector count is counting page reallocation, then erase fail must count block reallocation.

[Christopher]

1_Hz,

Are you going to resume testing the 320 now? I don't think it's quite so dead yet.

[One_Hertz]

yes, I have resumed the testing.

I am trying to upload a screenshot but it just says upload failed...

The Erase fail count is at 33 and reallocations are now 7299.

[johnw]

I fired up my rig again last night after several days of it staying off to check data retention. All is well. I have yet to get to the samsung because I don't seem to have a 5-6 hour block of time free to play with it

I'm glad you tested the power-off data retention on the Intel. Do you remember how many days it was powered off?

[Anvil]

I'm planning on doing the data retention test on both of my drives at ~500TiB, I was thinking of 4-5 days completely disconnected.

Just for the record, the Toshiba toggle mode NAND used on the SF2 drives are 32nm, not 34nm.

[Anvil]

I rebooted the pc last night so the drives have been up for 24 and a few minutes.
(the pc was online for 17 days)

Kingston SSDNow 40GB (X25-V)

453.22TB Host writes
Reallocated sectors : 12

MD5 OK

35.28MiB/s on avg (~25 hours)

--

Corsair Force 3 120GB

01 94/50 (Raw read error rate)
05 2 (Retired Block count)
B1 66 (Wear range delta)
E6 100 (Life curve status)
E7 17 (SSD Life left)
E9 311167 (Raw writes)
F1 414179 (Host writes)

MD5 OK

106.73MiB/s on avg

power on hours : 1218

B1 is down from 69 to 66 this time, same reduction as E7 (MWI).

[Christopher]

I got a Windows detected a hard disk problem pop up message, but everything appears to be fine. It triggered for the Mushkin, but for no apparent reason. All SMART stuff seems to be unchanged from last night, still 5 retirements.

Mushkin Chronos Deluxe 60 Update, Day 45
3.3.2FW

05 5
Retired Block Count

B1 11
Wear Range Delta

F1 442911
Host Writes

E9 341716
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 128.95MB/s Avg
Intel RST drivers, Celeron G530 Biostar TH67+

1043 Hours Work
Time 1 month 13 days 11 hours

12 GiB Minimum Free Space

[Christopher]

I got an email from CDI with the same Health Bad warning, not sure why either.

Why would Windows and CDi just now decide the drive was about to die? Its been at MWI 10 for weeks.

[B.A.T]

And I'm back. First something completely off topic



On topic

Second: The M4 is dead. Just like the Samsung 470 it couldn't last without power for some time, in this case 9 days. I've tried to power on and off, and i several pc but no luck. Do you have any ideas I should try?

Third: The MTRON is something wrong with. I've tried it in my foldingrig but without any luck. I get the same problems as before and the system feels sluggish and sometime hangs. I'm going to send it back to GullLars and he can rma it. They come with 5 years warranty som he might get something new and shiny

The Intel X25-M G1 and the Kingston V+100 is back in the test again and all looks normal at this moment.

[Hopalong X]

Great picture of the new Tech support crew.

The M4 was giving problems before this? If so, I guess I missed the posting.

[devsk]

And I'm back. First something completely off topic

awww! So cute! Congrats again man!

On topic

Second: The M4 is dead. Just like the Samsung 470 it couldn't last without power for some time, in this case 9 days. I've tried to power on and off, and i several pc but no luck. Do you have any ideas I should try?

Wow! That's out of the blue! I was hoping it would make it to 1PB. Looks like Vertex Turbo is the one which has the 1PB torch in hand now.

May be we should take a break on Vertex Turbo and have it sit idle, powered off for few days to see if it still keeps going! I have a vague memory that bluestang has been powering it off during the day on working days.

[Ao1]

It seems that the MWI might not be so overly conservative after all. 1 year to hold data without a charge after the MWI has expired seems excessive. I wonder how much the MWI would change if the requirement to hold data changed to 3 months as it is with enterprise.

Zads