SSD Write Endurance 25nm Vs 34nm

[Ao1]

Generic USB Drive – 4 GB
Writes: 889.80 GB
Average speed: 2.72 MB/s

Going to have to suspend for a week as I'm going to be traveling.

[Anvil]

The USB Drive is a bit interesting, as is SD/CF cards for DSLRs, they do take some beating.

I've been using a 32GB SanDisk Cruzer? for years, still working but is noticeably slower than it used to be, large files are OK but small files are taking forever.
Not sure how much is written but it was used on a daily basis and a few times a week for "backup" of a single ~2GB file, I have tried and there is no way of restoring performance.

[Ao1]

A USB thumb drive does not have a controller (AFAIK), so it’s about as close as you can get to testing NAND only without specialist equipment. I was interested to see how well the NAND stacked up with no controller vs a sophisticated SSD controller. TBH I knew I was going to be travelling, but I didn’t think it would have lasted until then. It’s a bit annoying to have to stop, but theoretically it should be well within P/E cycles, so hopefully no big deal.

[canthearu]

Ummm, no, USB thumb drives certainly have controllers. They perform wear leveling and command multiplexing like destop drives. Most won't be as complicated nor as advanced as the sandforce and marvell SSD controllers, but the basic functionality is there.

[Christopher]

USB/CF/SD:SSD :: WW I Biplane:F22 Raptor

Some mobile devices have a couple of flash ICs that are managed and operated by the SoC as I understand it, while others have discreet flash controllers.

[Kain665]

USB/CF/SD:SSD :: WW I Biplane:F22 Raptor

Some mobile devices have a couple of flash ICs that are managed and operated by the SoC as I understand it, while others have discreet flash controllers.

I beg to differ. SD/CF controllers sure, they are as simple as they get. But USB drive controllers are more complex, especially with the advent of extremely large USB 3.0 drives. More akin to a P51 Mustang in your analogy.

[Zaxx]

I've seen a few usb 3.0 sticks that are actually lil mini ssds...think there is even one that uses a SF1200 if I remember correctly, not cheap tho.

[Christopher]

I beg to differ. SD/CF controllers sure, they are as simple as they get. But USB drive controllers are more complex, especially with the advent of extremely large USB 3.0 drives. More akin to a P51 Mustang in your analogy.

That's quite true. Many USB 3.0 devices really blur the line, I was more referring to "old-style" USB drives. There are quite a few in-between solutions and some are more SSDs (or are SSDs). But that USB stick I've had floating around for the past 6 years? Probably not that bright.

[Ao1]

http://en.m.wikipedia.org/wiki/Thumb_drive

l'm using a usb 2 thumb drive. The controller is very basic, certainly nothing like a ssd controller.

[DooRules]

My 64 gb Kingston USB 3 DataTraveller

[canthearu]

Intel 520 60GB - Day 103

Drive hours: 2,421
ASU GiB written: 768,721.26 GiB (750.71 TiB)
Avg MB/s: 93.21 MB/s
MD5: OK

Host GB written (F1): 774,055.06 GiB (755.91 TiB, 24769762 raw)
NAND writes (F9): 547,971 GiB (535.13 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Raw Error Rate (8B): 120 normalised
Avaliable Reserved Space (AA): 100 normalised
Media Wearout Indicator (E9): 1

Sandisk Extreme G25 - 120gb - Day 3

Drive hours: 73
ASU GiB written: 30,467.31 GiB (29.75 TiB)
Avg MB/s: 120.38 MB/s
MD5: OK

Host GB written (F1): 30,712 GiB (29.99 TiB)
NAND writes (E9): 22,942 GiB (22.40 TiB)

Retired Block Count (05): 4
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 104 normalized
Media Wearout Indicator (E7): 97 normalized
Wear Range Delta (B1): 4 raw
Reported Uncorrectable Errors (BB): 0 raw

I got an erase failure this time that knocked out a couple more blocks on the sandisk extreme. At least it promises to be an interesting ride if nothing else! My calculations at this stage seem to indicate the sandisk firmware is programmed for 3000 write cycles (like the corsair force 3)

[kaktus1907]

Silicon Motion SM3255P controller
Target Performance

• MLC sustained read rate: 22MB/s*
• MLC sustained write rate: 12MB/s*
• TLC sustained read rate: 18MB/s*
• TLC sustained write rate: 9MB/s*

datasheet


Micron 25nm 32gbit 3K P/E NAND asynchronous

[B.A.T]

Todays update:

m4
1021.9420 TiB
4292 hours
Avg speed 73.66 MiB/s.
AD gone from 28 to 25
P/E 17610.
CE 26
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
696.74 TiB
23602 hours
Reallocated sectors : 136 to 142
Available Reserved space: 76 to 75
MWI= 57 to 50
MD5 =OK
40.29 MiB/s on avg


Intel X25-E 64GB
226.74 TiB
690-30=660 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI= 98
MD5 =OK
91.20 MiB/s on avg


Mtron Pro 7025 32GB
ASU write 18.1485 TiB
Host write 33,0259 TiB
Total host write 80,5614 TiB
Day 5
MWI= 99
MD5 =OK
47.13 MiB/s on avg


8 hours left to 1 PiB

[Kain665]

That 520 is barreling towards 1pb without a single reallocation. I wonder when it'll pop...

[Christopher]

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

Plextor M3P 128GB Day 11
289,386.14 GiB
282.86 TiB

271.48 MB/s Avg, down from 273.49 MB/s in the other rig

298 Hours

Rellocated Event Count 0

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

Samsung 830 256GB Day 80

1,967,199.61 GiB
1,921.07 TiB

293.82 MB/s Average
1945 Hours

8542 Wear Leveling Count
MWI 1

7/0 Erase/Program Fail
Used Reserved Block Count: 14/28672 sectors

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

The march to 2PB continues. It actually has 1,980,526 GiB of LBA writes, but only 1,967,199 in ASU.

[Christopher]

That 520 is barreling towards 1pb without a single reallocation. I wonder when it'll pop...

Well, it's "only" at 535TB of NAND writes

The first 64GB M4 died without a reallocation, and the second one is at 1000TB without a reallocation yet. By comparison, my 830 256GB "only" made it to 1PB without a reallocation, but it's a 256. It's almost at 2PB with 7 reallocations, and I'm happy with that.

[Anvil]

Most Crucial's are already flagged with Bad Blocks, found in the "Factory Bad Blocks" SMART attribute.

So, there are some bad blocks, I don't think I've got a single Crucial w/o some blocks flagged. (this should be adapted by all vendors)

[Christopher]

Very few drives actually show the factory bad block count. Indilinx on Cognac firmware and the Microns/Crucials are the lone drives which do (that I can think of off the top of my head).

I think some of the keys to success are successfully identifying questionable blocks during manufacturing. But it's fairly impressive that so many drives are making it past MWI before flagging bad blocks at "runtime"

[Anvil]

I've probably mentioned that some of the G2's had bad blocks right from the start. (X25-V's in particular)

Endurance_cr_20120608.png

The m4 is ahead of the X25-V, it made it just in time for 1PiB

[canthearu]

@Christopher

Your plextor is about to exceed the octane's life ... that is a good thing.

[Christopher]

@Christopher

Your plextor is about to exceed the octane's life ... that is a good thing.

I certainly hope so. That would be extremely depressing were it not to surpass the Octane.

[B.A.T]

And that ladys and gentlemen, is 1 PiB

m4
1024.8467 TiB
4304 hours
Avg speed for the whole run is 69.35 MiB/s.
AD 24
P/E 17659.
CE 26
MD5 OK.
Reallocated sectors : 00

[Christopher]

Bravo! Think about all the drives which never made it this far... The list is long. And now we have two drives past 1PiB and one closing in on 2PiB.

[B.A.T]

It's been a bumpy ride, that's for sure

I hope the G1 also makes it to 1 PiB. 300TiB left and it's strugling some.

[canthearu]

Time for a retention test on that m4 *mwhahahaha*

But grats on the 1PB ... I didn't think the m4 could make it.