SSD Write Endurance 25nm Vs 34nm

[RaZz!]

unfortunately i'm not in the position to offer any sponsorship, i'm sorry

i really had to argue with myself whether to buy a SSD or not. now i'm a proud owner of a crucial m4 128gb and to think about putting it under such an endurance test raises my hackles

so it's even more creditable that you guys do this kind of research and share your results!

now that i'm already here in this thread: is there a way to monitor the amount of data written to a SSD? the tool SSD Life seems to offer exactly that, unfortunately it says my M4 doesn't provide any statistical data about data writes. so there's no chance?

[B.A.T]

Download SSDLife or Crystal Disk Info. Look at line 173/AD just like my last result post at page 147.
Take the P/E number and multiply with the total capacity of your ssd and you got how much it's been written to.

My m4 got 8400 P/E used.
8400*64020803584 byte = 489,1032859683037 TiB.

[canthearu]

now that i'm already here in this thread: is there a way to monitor the amount of data written to a SSD? the tool SSD Life seems to offer exactly that, unfortunately it says my M4 doesn't provide any statistical data about data writes. so there's no chance?

The smart values have a media wear count, that raw value documents the average wear cycles for your drive. Multiply that value by the drives size for rough estimate for the amount written.

[canthearu]

Torqx day 11

Drive hours: 262
GiB written: 13,796.90 (13.4739 TiB)
Avg MB/s: 17.22

Bad blocks: 83
Wear cycle counter: 0/913/1379

Intel 520 - day 2

Drive hours: 70
Avg MG/s: 74.92 MB/s

Host GB written (F1): 14,653.4 GiB (14.31 TiB , 468909 raw)
NAND writes (F9): 10,374 GiB (10.13 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 100 (0 raw)

[Christopher]

As B.A.T. said, the only way to measure the total writes to the M4 is through the wear leveling attribute. Multiplying it by 128 would give you a really rough idea of the total writes. In this testing you'll get a more accurate count that way as the writes are mostly sequential which creates less write amplification. In daily use you would see a little more. But a rough estimate is the best you can do (with the M4).

[kensiko]

Welcome canthearu, I will follow the 520 closely

What happened to the XS forum, was there an update ? I didn't receive my weekly update showing me the link to the new posts.

[Ao1]

Intel 520 - day 2

Drive hours: 70
Avg MG/s: 74.92 MB/s

Host GB written (F1): 14,653.4 GiB (14.31 TiB , 468909 raw)
NAND writes (F9): 10,374 GiB (10.13 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 100 (0 raw)

So it looks like the 520 also has an initial "credit" period.

[RaZz!]

thanks for the tip.

however, the raw value just shows 0's:



hum...

[Christopher]

RaZz!

you just haven't written that much yet... Less than 128GB host writes I'd guess.

[RaZz!]

aight. i already thought it'd be smth like that.

i just wanted to make sure as i left the windows temp folder as well as the browser cache on my ssd. so i'll check these values from time to time. most SSD guides recommend moving all this stuff off the SSD, but if the writes aren't excessively high, it doesn't make any sense to do that. why buy a fast SSD when you move everything to a slower HDD?

[Bun-Bun]

I agree. I just move large stuff off SSD's to save space. Games/Docs/Videos ect.

[kensiko]

anyone else notice this: http://cseweb.ucsd.edu/users/swanson...BleakFlash.pdf

I read that.

I'm not worried at all, a lot of companies are looking for alternatives to flash. I think the memristor is a good one, there is even a website for it: http://www.memristor.org/

[Anvil]

Kingston SSDNow 40GB (X25-V)

746.43TB Host writes
Reallocated sectors : 05 25 up 2 from last update (4 days ago)
Available Reserved Space : E8 99
POH 6710
MD5 OK

33.31MiB/s on avg (~262 hours)

--

Will update the chart and listings tomorrow. (had a few days off )

[canthearu]

So it looks like the 520 also has an initial "credit" period.

Heh, E9 has gone down to 99 over the night. Obviously I've finished burning this "credit" period already

Of course, by the time most 60gig SSDs reach 10TB NAND writes, people have seen that shiny new 256gig SSD on the shelf, and exiled the old 60giger to the secondary machine they hardly use

[Christopher]

Okay, the new Turbo arrived today.

It's SMART data is much more promising than the last one. Here it is straight out of the box.

[Christopher]

MTRON 7000 PRO 16GB SLC, Day 19

GiB : 80 002.81
TiB : 78.1277
MBs : 56.90

187 92/91/138218085


It gets faster yet. Interesting.

[canthearu]

Looking through the past, it seems that lack of a SATA 3 controller may be causing my Intel 520 to run slower then it can.

The best computer I have currently is an Intel i5-760 with only SATA 2 ports.

At some stage in the future, I'll have to try and get a benchmark run or 2 on my intel 520 using an intel sata 3 board, see if it makes a difference, in particular to the overall endurance MB/s.

[Christopher]

Looking through the past, it seems that lack of a SATA 3 controller may be causing my Intel 520 to run slower then it can.

I would guess moving to a SATA III port would increase speed, but only by a few MB/s avg.

I've experimented with the Chronos D and Samsung 830 on SATA II ports, and you do lose a few MBs even though you're not constrained by 3gbps bandwidth.

If you wanted to, and I'm not saying you should, Apricorn makes a Velocity Solo PCIe x1 SATA III controller card (the SSD mounts to it). You wouldn't get max reads, but you'd get a modest increase in write speed. I was thinking about getting one for other purposes, but it could work for you (or something like it).

[canthearu]

Yeah, not too thrilled about using Marvel SATA ports for SSDs.

I would just want to run a few benchmarks and to do a little bit of endurance testing to see the speed difference. It probably isn't a lot

[Christopher]

ANOTHER DEAD MTRON

I powered down the rig to install a second SATA power lead and install the new Turbo. When I booted back up, the MTRON 7025 was keeping Windows from booting. It was being detected in the UEFI and the activity light is on, but it's basically dead. Sounds similar to B.A.T.'s MTRON.

I tried it in a USB enclosure and on several different systems, but to no avail. It's locked up or something, but it doesn't work.

And no, I didn't short it out!

So cross the MTRON of the list.

[canthearu]

I tried it in a USB enclosure and on several different systems, but to no avail. It's locked up or something, but it doesn't work.

And no, I didn't short it out!

So cross the MTRON of the list.

Wow, that is highly unexpected, considering the drive was only just starting to get worn in. I thought I'd never see the death of that drive.

[Christopher]

Wow, that is highly unexpected, considering the drive was only just starting to get worn in. I thought I'd never see the death of that drive.

It's almost assuredly a firmware problem. I've asked if I can send it back to the factory to have the firmware updated.

On the bright side, the new Turbo 1.7FW 64GB is up and running. Here is the SMART data straight out of the box:


And then after formatting, here are some benches:

[canthearu]

Torqx day 12

Drive hours: 286
GiB written: 15,201.03 (14.8452 TiB)
Avg MB/s: 16.91 MB/s

Bad blocks: 83
Wear cycle counter: 0/999/1509

Intel 520 - day 3

Drive hours: 94
Avg MB/s: 75.42 MB/s

Host GB written (F1): 21,005.375 GiB (20.51 TiB , 672172 raw)
NAND writes (F9): 14,860 GiB (14.51 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 98 (0 raw)

E9 has started to move.

[Ao1]

Here is a comparison from my testing and that done by Vapor. Vapor had a modified OCZ V2 drive with LTT disabled and itwas using Hynix nand. Mine was a V3 using IMFT nand. SF drives stop at MWI 10, whilst it appears the threshold for the 520 is 0. Too early to be sure but it certainly looks like the 520 is able to write more before a drop in the MWI is recorded.

(I did not have a smart logger at the time so the values are not precise)

[Anvil]

Endurance_cr_latest.png

I've kept the MTRON, in case it returns?

Doesn't look like the MTRON's are up for this kind of "testing", weird!

Endurance_PEsummary_20120228.png