SSD Write Endurance 25nm Vs 34nm

[Anvil]

The throughput looks very stable on all your drives.

Do they vary much during a loop?

[B.A.T]

Todays update:

m4
832.7410 TiB
3530 hours
Avg speed 74.48 MiB/s.
AD gone from 137 to 134.
P/E 14359.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
587.72 TiB
22863 hours
Reallocated sectors : 151 to 152
Available Reserved space: 90
MWI=124
MD5 =OK
45.20 MiB/s on avg

[Christopher]

Samsung 830 256 Day 50

1,188,765.36 GiB
1,160.90 TiB

297.09 MB/s Average
1175 Hours

5167 Wear Leveling Count
MWI 1

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



OCZ Vertex Turbo 64 Day 67

474,463.19 GiB
463.34 TiB

85.24 MB/s Avg
1688 hours

8477 Avg PE count
-100 MWI

0/0/0 Read/Program/Erase Fail


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

The 830 finally popped some blocks.... I'll dig through the data to see when exactly.

While I was away over the weekend, I had trouble getting in touch with the E-Rig. When I got back and looked, the VT was doing .02MB/s and the average had dropped to 47MB/s. I stopped and restarted it, and it seems to be fine now.

[Anvil]

Endurance_cr_20120507.png

[canthearu]

The throughput looks very stable on all your drives.

Do they vary much during a loop?

Definitely.

The Intel 520 is definitely the most stable though, usually always running at about 103meg per sec in the loop.

The Torqx 2 is all over the place. (20meg per sec to 60meg per sec) Ussually faster at the beginning of the loop for a gig or so.

The Vertex 4 is mostly 182meg or so per sec during the loops, but slows significantly once free drive space falls below about 8gig or so. My average is probably so low because of the large amount of static data I am using meaning loops are shorter.

[B.A.T]

A small teaser of what I'll get in the mail next week


By the way, it's a 64GB flavour.

[Anvil]

My best guess would be the

Kingston SSDNow E Series SNE125-S2 64GB

Should be very interesting if that is the case.
(being an SLC it could literally last forever)

BTW, the price makes me wan't one, I paid close to 2x for the Intel X25-E 32GB. (a few years ago)

[B.A.T]

And you are correct

There are only a couple left now at around 2K NOK.

Here and here

[Anvil]

Looks like one of the stores are already out of stock

[Christopher]

And you are correct

There are only a couple left now at around 2K NOK.

Here and here

Oooohhh! I'm so jealous!

I did get a 32GB X25-E for $45 not too long ago, so I can't complain all that much.

[B.A.T]

My 64 cost me 350$ so you got a good deal.

Looks like one of the stores are already out of stock

I see you are a quick draw

[canthearu]

Interesting news:

Seems the Patriot Torqx 2 is dead. It crashed when I upgraded the vertex 4 firmware.

Reporting CRC errors and saying drive unformatted. Will try a secure erase before declaring complete death though, but not expecting much.

On that note, I'm playing with the new vertex 4 firmware. In some ways it is impressive, in others, it is a setback. For my 55 gig endurance run, I get about 27gig written at about 375meg per sec, and the other 27gig at about 100 gig per sec.

Hopefully, OCZ can get this firmware working a bit more consistently before releasing 1.4 proper, but it has promise.

[B.A.T]

Looks like it's joining the heard of dead ssd.

[Anvil]

Interesting news:

Seems the Patriot Torqx 2 is dead. It crashed when I upgraded the vertex 4 firmware.
...
Hopefully, OCZ can get this firmware working a bit more consistently before releasing 1.4 proper, but it has promise.

I usually disconnect all other drives when performing fw upgrades, let's just say that I had a few odd experiences that lead to data loss on other drives.

I ordered a V4 along with the SLC, I need to check out that drive

[Rubycon]

The 830 is a monster. Hopefully I made a good decision to stripe 8 of them (256GB) on this 1880ix-24.

[Christopher]

B.A.T.

You can't seriously be thinking of throwing an X25-E in your endurance rig... 80GB of 50nm SLC, OP'd to 59GB. If you're lucky, in year 5 your power supply will short it out. If not... I mean, unless you're running full span random 4K's, I don't think you stand a chance

[Christopher]

Samsung 830 256 Day 51

1,214,643.37 GiB
1,186.17 TiB

296.25 MB/s Average
1200 Hours

5279 Wear Leveling Count
MWI 1

2/0 Erase/Program Fail
Total Used Block Count 4
Reallocated Sectors 8192



OCZ Vertex Turbo 64 Day 68

481,958.16 GiB
477.66 TiB

85.10 MB/s Avg
1693 hours

8611 Avg PE count
-100 MWI

0/0/0 Read/Program/Erase Fail

--------------------------------
If the Vertex Turbo's firmware was still set for 10,000PE based MWI, it would still be at 24MWI.

The 830 just hit 1200 hours of testing, so 1,214,600GiB in 1200 hours is nothing to sneeze at.

[B.A.T]

Oh yes. Been at it for 10 months now and if it's impossible to breake I'll just put it in my foldingrig

Todays update:

m4
839.9220 TiB
3558 hours
Avg speed 74.50 MiB/s.
AD gone from 134 to 130.
P/E 14480.
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
592.11 TiB
22891 hours
Reallocated sectors : 152 to 153
Available Reserved space: 90
MWI=124
MD5 =OK
45.23 MiB/s on avg

[Christopher]

The way I figure it, you'd be able to avg around 160MB/s with a 32GB X25-E, or 13.5TB a day. That works out to 337.5x drive capacity (based on the 40GB of SLC onboard) a day. That means it would only take ~296 days to exhaust MWI. If the drive died exactly at MWI 1, then that would be do-able.

But I don't think it would work like that. If the X25-V can make it to 25,000PE cycles, why can't the X25-E make it to 500,000? Or 750,000? I think that 50nm SLC will pretty much last forever unless you're placing it under absurdly high WA full-span conditions.

Now double the amount of flash, but keep the same avg. speed... almost two years to MWI 1.



The man who decides to endurance test an X25-E would officially be known as the bravest man to walk the Earth.

[B.A.T]

But keep in mind the controller doesn't support trim so the WA will be higher than X25-V/M G2 during normal endurance testing.

[Christopher]

But keep in mind the controller doesn't support trim so the WA will be higher than X25-V/M G2 during normal endurance testing.

True, but the OP really helps. I don't think the E's suffer much from the lack of TRIM. Maybe I should do some testing.

[B.A.T]

The man who decides to endurance test an X25-E would officially be known as the bravest man to walk the Earth.

I think I'll quote you on that and one, frame it and hang it on my office wall

[Christopher]

I think I'll quote you on that and one, frame it and hang it on my office wall

B.A.T.: The Man. The Myth. The Legend.

[devsk]

Lot of fun around here....

We gonna need to change the graph beyond 1400 soonish. The Samsung is looking to go where no drive has gone before and will ever go: 2PiB and beyond. I see no indications of any kind of distress.

[B.A.T]

B.A.T.: The Man. The Myth. The Legend.

Sounds like I need some kind of cape