SSD Write Endurance 25nm Vs 34nm

[Hopalong X]

Sounds like I need some kind of cape

You had something like this in mind?

[Anvil]

Maybe we should get some T-Shirts, I'm sure we could find some Funny slogan.

[B.A.T]

You had something like this in mind?

That is probably the smartest thing to wear with this hobby

By the way - It's in my mailbox tomorrow

[Hopalong X]

The New Adventures of B.A.T.M.A.N.

Bats in his Belfry.

Coming to a theater near you fall of 2012.


Sorry B.A.T.
Chris and Anvil made me do it.

[Christopher]

We should have a t-shirt design contest. The winner gets two broken MTRON 16GBs!

[Hopalong X]

Well I already posted my picture entry
B.A.T.S. in his Belfry is my slogan entry.

[Christopher]

We have an erase failure on the Vertex Turbo
MWI is now down to 99, erase fail = 1, and it happened somewhere near 8738 avg PE.

So within a few days, both the 830 and the VT have failed their first blocks.

[B.A.T]

And I'll throw in a Kingston V+100 - The perfect paperweight

Who want to be the judge?

Todays update:

m4
845.7021 TiB
3581 hours
Avg speed 74.49 MiB/s.
AD gone from 130 to 127.
P/E 14578.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
595.66 TiB
22914 hours
Reallocated sectors : 153 to 154
Available Reserved space: 90
MWI=124
MD5 =OK
45.25 MiB/s on avg

[Anvil]

Kingston SSDNow 40GB (X25-V)

940.08TB Host writes
Reallocated sectors : 05 41
Available Reserved Space : E8 99
POH 8425
MD5 OK

33.46MiB/s on avg (~122 hours)

[Christopher]

Samsung 830 256 Day 52

1,240,699.36 GiB
1,211.62 TiB

296.62 MB/s Average
1225 Hours

5391 Wear Leveling Count
MWI 1

2/0 Erase/Program Fail
Used Reserved Block Count: 4/8192 sectors


OCZ Vertex Turbo 64 Day 69

489,383.22 GiB
477.91 TiB

85.24 MB/s Avg
1718 hours

8743 Avg PE count
-99 MWI

0/0/1 Read/Program/Erase Fail

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

The VT has finally dropped a block. If you had put money on the VT having the first erase fail, you would have lost.

Nevertheless, both drives have really impressed me so far in that respect. The 830 didn't lose a block until after a PiB, and the 64GB 830 made it almost halfway to 1 PiB before popping a block.

[Anvil]

Endurance_cr_20120509.png

Christopher,

I find the 830 to be the most impressive, nothing wrong with the VT though, as long as the fw does it's job it'll definitely put up a fight.
I'm not saying that as if the fw is bad, it's just that fw's have come a long way.

[Christopher]

The Cognac firmware has come about a thousand miles since 2009, but I think I have to credit the Vt's performance to the flash. The last one was obviously defective -- if it arrives on my door with runtime bad blocks AFTER burn in, something is up.

But yeah, the 830 is so sexy -- it's a shame more drives aren't using Samsung flash like the VT and 830.

The VT made it to almost 8.7K PE cycles before a bad block, and the 830 made it to almost 5.2K PEs. Not bad considering one is using 51nm and the other is using 27nm.

[Anvil]

Yes, the NAND should be excellent on the VT.

[canthearu]

Hmmmm, in other news ... the torqx 2 may not be quite dead yet.

Just did a secure erase and it seems to be back ... but with a couple more popped blocks and a read error.

And a graph showing transfer rates over time for the intel 520 and vertex 4 with new firmware.



Notice how the vertex 4 starts off real quick, but ends up going slower then the intel 520 towards the end of each loop

[Anvil]

Hmmmm, in other news ... the torqx 2 may not be quite dead yet.
...
Notice how the vertex 4 starts off real quick, but ends up going slower then the intel 520 towards the end of each loop

I figured you'd get some response from the Torqx, is it fit for the test?

So the V4 starts off at ~400MB/s for a minute or so and then levels at ~100MB/s, what does look a bit odd is the pause at the end of the loop, it looks to be lasting longer than the SF drive?

[canthearu]

Yeah, TRIM certainly isn't instant on the vertex 4. A few significant pauses during TRIM occur.

On the 1.3 vertex 4 firmware, performance is a bit more consistent (at around 180meg per sec)... which seems to be better for endurance testing then the current situation.

The Torqx 2 seems fit enough to go back into the endurance test .... I'll put it back in tomorrow. (SMART values are starting to show the scars of old age ... tis an old battle wound :P)

[Anvil]

How many files are generated during a single loop on each drive? (520 vs V4)
(might make up for the time it takes for deleting/TRIMming the files)

[canthearu]

Yeah, the V4 does 18240 files per cycle.

The 520 does 12940 files per cycle.

The V4 (and 520) is quicker deleting when TRIM is disabled though.

[Christopher]

--------------------------------
Samsung 830 256 Day 53

1,261,175.94 GiB
1,231.61 TiB

296.40 MB/s Average
1245 Hours

5480 Wear Leveling Count
MWI 1

2/0 Erase/Program Fail
Used Reserved Block Count: 4/8192 sectors

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

OCZ Vertex Turbo 64 Day 70

495,280.250 GiB
483.67 TiB

85.08 MB/s Avg
1738 hours

8850 Avg PE count
-99 MWI

0/0/1 Read/Program/Erase Fail

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

[B.A.T]

Todays update:

m4
851.2467 TiB
3602 hours
Avg speed 74.49 MiB/s.
AD gone from 127 to 123.
P/E 14673.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
599.04 TiB
22935 hours
Reallocated sectors : 154
Available Reserved space: 90
MWI=124
MD5 =OK
45.27 MiB/s on avg


I was too late for the post-office today so you'll have to wait until tomorrow for the fun to begin

[Christopher]

I was too late for the post-office today so you'll have to wait until tomorrow for the fun to begin

That's not very becoming for the Bravest Man on the Face of the Earth.

[B.A.T]

I had to work some overtime... But it's going to be a glorious weekend.

[Christopher]

I posted about this in the Plextor M3P thread, but it's worth pointing out here... because things have changed, and the M3P may now be a testing candidate. I'll have to run some tests, but the M3P 1.03 firmware added a whole mess of new SMART attributes -- MWI, temp, etc (viewable through the new Plextool). I can probably figure out the attributes on my own, but one may be host writes.

UPDATE
Yes, the M3P now has host writes! In 64MB increments! Attribute B1... thanks, Plextor! I don't know why I'm so excited-- it still doesn't have a PE cycle counter.

[Christopher]

I got one erase failure more on the VT, and one read failure. I hope it's not about to hit a brick wall and die slow over the next two weeks.

So that's 2 erase and one read fail.

[Computurd]

i fear for the M3P now...