SSD Write Endurance 25nm Vs 34nm

[bluestang]

^^ 01, CA and CB are staying pretty steady though. How did the last one look in those areas when C5 started to climb?

[canthearu]

@Christopher

NAND hasn't even exceeded the supposed erase cycle specification yet

I think it might be time to put the barefoot drives out to pasture. (my intel 520 on the other hand is still ticking away just fine, will have to post an update today ... been busy/lazy)

[Christopher]

^^ 01, CA and CB are staying pretty steady though. How did the last one look in those areas when C5 started to climb?

They looked just fine on the first one. CA and CB have been at the same counts for almost six weeks.

@Christopher

NAND hasn't even exceeded the supposed erase cycle specification yet

I think it might be time to put the barefoot drives out to pasture. (my intel 520 on the other hand is still ticking away just fine, will have to post an update today ... been busy/lazy)

Maybe... it's not like people are still Barefoot crazy. I still heart them -- the VertexEX is where it's at.


-----------
It's now at 3 erase/38 read fails 9025 AVG PE count.

The 830 tossed another set of blocks too. Sadness!

[Anvil]

I think it might be time to put the barefoot drives out to pasture. (my intel 520 on the other hand is still ticking away just fine, will have to post an update today ... been busy/lazy)

It's been almost a week since your last update

The new generation drives are definitely the most interesting ones, a few older ones does not hurt though.

Endurance_cr_20120511.png

[canthearu]

Oh dear - it has been a week, I better get back in the swing of things than.

Patriot Torqx-2 was offline for several days, but I have restarted it now. It passed the "too busy to reinstall" 2-3 day retention test. It is also running a bit slower.

Patriot Torqx-2 64GB - Day 87

Drive hours: 1,959
ASU GiB written (first run): 137,907.07 GiB (134.68 TiB)
ASU GiB written (current run): 471.20 GiB (0.46 TiB)
ASU GiB written (Total): 138,378.27 GiB (135.14 TiB)
Avg MB/s: 23.41 MB/s
MD5: OK

Bad blocks: 6 grown, 83 factory
Wear cycle counter: 8768 average erase cycles, 13491 maximum erase cycles.
Total ECC Error Count: 1
SATA PHY error count: 250

Intel 520 60GB - Day 79

Drive hours: 1,843
ASU GiB written: 585,640.55 GiB (571.91 TiB)
Avg MB/s: 95.02 MB/s
MD5: OK

Host GB written (F1): 589,806.25 GiB (575.98 TiB, 18873800 raw)
NAND writes (F9): 417,543 GiB (407.76 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Raw Error Rate (8B): 112 normalised
Avaliable Reserved Space (AA): 100 normalised
Media Wearout Indicator (E9): 1

Vertex 4 128GB - Day 26

Drive hours: 648 (556 + 92)
ASU GiB written: 276,622.26 GiB (270.14 TiB)
Avg MB/s: 122.87 MB/s
MD5: OK

Sector GB written (E8): 279,130.15 GiB (272.59 TiB, 486156078177 + 99222277356 raw)
Reallocated sectors (05): 0
Raw Read Error Rate (01): 6
Remaining Life (E9): 60 (90 current - 30)

For the vertex 4, all the smart values were reset after firmware upgrade. Overall speed is a lot slower. Regret doing it.

[Anvil]

I had issues with my Vertex 4 on the shipped fw (1.3), the 1.4RC fw looks much better. (maybe not for this test though)

Had to flash off the X79 as it gave all sorts off issues, including being locked.
Flashing on a Z68 worked instantly though, it cleared most if not all attributes.

[Ao1]

It sucks when a major f/w update gets released after you have started testing. From an endurance perspective it is hard to know what impact the f/w change might have on the end result. Then again it seems a bit pointless to test on outdated f/w.

I’ve read a few snipets recently about some significant changes to NAND structure that are on the cards to cope with smaller geometries. (3D nand). The tech is continuing to evolve rapidly, so I wouldn’t be surprised if drives continue to be outdated every 12months and remain at high price levels for a long time yet.

[Christopher]

The Vertex Turbo is dead! And it took the endurance rig with it when it died.

That's three drive failures in three days for me. What are the chances?



511,079.0GiB

[Anvil]

Those signs looks to be a sure way of telling that it's about to self destruct.

What happened to the rig?

[Christopher]

Those signs looks to be a sure way of telling that it's about to self destruct.

What happened to the rig?

It just locked it up. It didn't "kill" the rig, I just had to restart.

I'm going to try d-flashing the VT. That SMART data was just taken from my main rig... It shows up in volume management as uninitialized and I get a "semaphore message timeout" when I try to initialize it. I was able to pull the most current SMART data off of it though. Perhaps a d-flash will get it back in the game.

[Anvil]

Well, giving up is not in the spirit of this thread

[Christopher]

Some days I feel as though I'm in the Bermuda Triangle of solid state storage.

All the read and erase failures were in Bank 6 (C1). A dflash failed, so I laundered the flash. Subsequent dflashing attempts also failed.

Even if I had been successful, it would be like taking a lame racing horse and shooting it full of steroids and pain-killers and making it race once more; nothing good is going to happen, and I'll just have to shoot it in the head.

But that doesn't mean I'm going to stop trying to dflash it though, but I'll just have to wait a day or so.

[B.A.T]

Too bad about the VT....

What will be your next ssd?

[Christopher]

Too bad about the VT....

What will be your next ssd?

At this rate, I'm just going to have to start testing M4s and X25-Vs!

I'm thinking about doing a 128GB M3P or a 120GB Mushkin Chronos Deluxe.

[B.A.T]

The M3P is a very exciting canditate.

[Hopalong X]

At this rate, I'm just going to have to start testing M4s and X25-Vs!

I'm thinking about doing a 128GB M3P or a 120GB Mushkin Chronos Deluxe.

Sorry to hear the bad news.

Scroll down to the SSD's. Samsung 830 128GB $160. Of course you are already doing the 256GB aren't you? Just a thought.
http://promotions.newegg.com/NEemail...E0-_-PromoWord

There is a HyperX 120GB also.

[Christopher]

I already have the M3P and Chronos Deluxe. I'm a fan of the Kingston though, and naturally, I also really dig the 830. I very much would like to test an 830 128GB, but I'm already doing the 256GB.

[B.A.T]

Todays update:

m4
860.8958 TiB
3650 hours
Avg speed 74.64 MiB/s.
AD gone from 119 to 117.
P/E 14853.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
603.36 TiB
22962 hours
Reallocated sectors : 154
Available Reserved space: 90
MWI=124
MD5 =OK
41.94 MiB/s on avg


Intel X25-E 64GB
7.23 TiB
(49-30)=19 hours
Reallocated sectors : 0
Available Reserved space: 100
MWI=99
MD5 =OK
106.09 MiB/s on avg




The retention test went fine and the G1 is running fine again. Due to bad weather the power was out a couple of hours during the night. Everything is ok now.

[Christopher]

Wow, BAT I really thought the X25-E would do an average of around 160MB/s. I really thought it would be faster on average.


Samsung 830 256 Day 55

1,312,193.38 GiB
1,281.43 TiB

292.49 MB/s Average
1298 Hours

5700 Wear Leveling Count
MWI 1

4/0 Erase/Program Fail
Used Reserved Block Count: 8/16384 sectors
----------------------

The 830 is all alone now.

[bluestang]

That's too bad about the VT. Only 7 Initial and 3 Erase Bad Blocks. That's not really alot. It gave up cause of all the Bad Read Blocks.

I guess if we want one that works like it's supposed to ...it has to actually be a Crucial

[canthearu]

Christopher: "Killer of SSD" indeed!

I think you have killed more SSDs then I've owned!

Very sad about the VT though, I thought it would have lasted a fair bit longer to be honest.

[Christopher]

Christopher: "Killer of SSD" indeed!

I think you have killed more SSDs then I've owned!

Very sad about the VT though, I thought it would have lasted a fair bit longer to be honest.

I'm responsible for the deaths of 10 or 11 SSDs. Only a couple of those were drives I wasn't trying to kill. I'm up to three dead in the past three days!

[canthearu]

Ha, I only own 10 SSDs

[B.A.T]

Wow, BAT I really thought the X25-E would do an average of around 160MB/s. I really thought it would be faster on average.

The loop start at around 130 MiB/s and climbs to around 160 MiB/s but the 9 sec waiting between each loop pulls the avg down. I can try to shorten that some more.

[Christopher]

B.A.T., you are the Bravest Man on the Face of the Earth.... don't be afraid to experiment. Considering the fact that there is 80GB of SLC on the 64GB E, I wouldn't hesitate to drop the the wait between loops or reduce pause per 1K files deleted. You need all the help you can get.