SSD Write Endurance 25nm Vs 34nm

[canthearu]

The 520 should take another 45 days to hit MWI 0.

Maybe a little less, I turned off the endurance test to turn AHCI back on and do a couple of benches, so about half an hour of wasted time.

What happened was that just after I did the benches for the 520, I changed the CPU in my HTPC endurance box to a 2.7ghz core 2 E5400. At this stage I reset the bios and that caused the system to go back to IDE enhanced mode rather then AHCI.

Didn't notice until yesterday that NCQ wasn't working.

[Christopher]

Vertex Turbo 64 Day 7

48514.91GiB
82.65MB/s

869 Average Erase Count

83 MWI

0/0/0 Program/Erase/Read Failures

[canthearu]

Patriot Torqx-2 64GB - Day 19

Drive hours: 454
GiB written: 26,063.37 GiB (25.4526 TiB)
Avg MB/s: 19.44 MB/s

Bad blocks: 2/0/83
Wear cycle counter: 0/1678/2542

Looks like the bad blocks counter variable stores more then only bad blocks, but I don't really know what it means.

Intel 520 60GB - Day 10

Drive hours: 262
Avg MB/s: 92.57 MB/s

Host GB written (F1): 71,128.81 GiB (69.46 TiB, 2276122 raw)
NAND writes (F9): 50,346 GiB (49.17 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 86

[bulanula]

Anyone got an idea when the next NAND die shrink is going to happen ?

I don't think 25nm is the limit. Maybe 22nm NAND is up next ?

Thanks !

[Ao1]

"Intel and Micron noted that the December production ramp of their 20nm 64Gb NAND flash product will enable a rapid transition to the 128Gb device in 2012. Samples of the 128Gb device will be available in January, closely followed by mass production in the first half of 2012".

http://hexus.net/tech/items/storage/...4gb-20nm-nand/

[Anvil]

Kingston SSDNow 40GB (X25-V)

767.88TB Host writes
Reallocated sectors : 05 27
Available Reserved Space : E8 99
POH 6898
MD5 OK

33.21MiB/s on avg (~450 hours)

[Ao1]

Hard to believe the X25-V is still plugging along. Not only has it written more than any other drive it is also one of the smallest. TBH I was surprised that it restarted after the last “rest” period.

I wonder how much of that is down to the write speed.

[Anvil]

34nm does seem to make a difference, still not able to convince myself about putting one of my C300 64GB to test.

Vapor looks to be out for some time so we might not get the final results of those drives, unless he posts an update within March I'll just move the drive to the other chart.

[Christopher]

I believe that the X25-V benefits from a low average speed, good NAND, and good firmware. Some of the same things that are going to keep the X25-M G1 working for quite some time.

One possible issue with the 25nm NAND is the fact that Intel probably has a better binning method now, one which sees the better bins of MLC going to more uses than were previously required (the best MLC becomes HET-MLC for example). So less of the highest quality MLC makes it into the consumer drives which use 25nm Intel MLC NAND.

So with the 320 series, Intel was either wondering what to do with those lower grade bins (which were still good enough for SSD use) or was concerned about quality on a new process (more inconsistent NAND), so they added redundancy to the G3s.

But unquestionably, the 34nm Intel MLC is finer than the 25nm, much more so than the difference from 50nm to 34nm.

[B.A.T]

Todays update:
The 24h retention test is over and everything is OK.
I'll post an update tomorrow after both ssd have run some more.

[Anvil]

Hard to believe the X25-V is still plugging along. Not only has it written more than any other drive it is also one of the smallest. TBH I was surprised that it restarted after the last “rest” period.

I wonder how much of that is down to the write speed.

How long will this one last?
(it's sitting as a boot drive in a server)

X25V-server.PNG

[Christopher]

I have a new, unopened X25-V in a glass case. There is a hammer attached to a chain, and a little plaque which reads "In case of emergency, break glass."

[Christopher]

Vertex Turbo 64 Day 8

56089.05GiB
82.88MB/s

1005 Average Erase Count

80 MWI
192 hours

0/0/0 Program/Erase/Read Failures

[B.A.T]

Todays update:
Everything is back running and here are some numbers

Intel X25-M G1 80GB
397,3376 TiB
21399hours
Reallocated sectors : 02
MWI=92 to 91
MD5 =OK
41.00 MiB/s on avg


I missed the first reallocated sector and now I got the second one. The first one came betwheen 28 of January and 2 of february.

m4
514.7639 TiB
2063 hours
Avg speed 65.85 MiB/s.
AD gone from 60 to 56.
P/E 9024.
MD5 OK.
Reallocated sectors : 00

[Christopher]

21000+ hours and 2 reallocations is pretty good in my book. That G1 is sweet.

[canthearu]

Patriot Torqx-2 64GB - Day 20

Drive hours: 479
GiB written: 27,695.53 GiB (27.0465 TiB)
Avg MB/s: 19.32 MB/s

Bad blocks: 2/0/83
Wear cycle counter: 0/1782/2706

Intel 520 60GB - Day 11

Drive hours: 287
Avg MB/s: 92.76 MB/s

Host GB written (F1): 79,046.66 GiB (77.19 TiB, 2529493 raw)
NAND writes (F9): 55,957 GiB (54.65 TiB)

Reallocated sectors (05): 0
Failure count (AB, AC): 0 program, 0 erase
Media Wearout Indicator (E9): 84

[bulanula]

What we really need is a C300 drive here.

Will probably own the M4.

The main things that affect endurance right now :

-capacity of drive
-SLC or MLC based
-lithographic process of NAND
-WA of the controller

[bluestang]

WTF..Would you give the C300 requests a rest already!

If you want to see another one tested...then go buy one and test it yourself.

[Christopher]

Bluestang,

When did the M225 get it's first reallocations? It was pretty far along if I recall. I could go back and look, but it's been so long that it's probably archived on papyrus.

[bluestang]

@Christopher,

I threw my 1st Program Fail @ MWI 92 and another @ MWI 89. It stayed at 2 until it died.
Didn't throw an Erase Fail until MWI 49, then another @ MWI 35. No more until after MWI 0. Died with 25.
No Read Fails until 908TB. Died with 43.

I was on XP to around MWI 59, then I switched to Win 7.

I believe you are ahead of the game since running on Win 7 from the start and on FW 1.7...looking good!

[Christopher]

Thanks, Bluestang. This drive looks pretty good so far. This VT64 version 2 is really more for vanity -- I want to get some crazy high numbers out of it.

[kensiko]

Anyone to test Everest ?

By the way, there have been benches on the OCZ forum for Everest 2, pretty good performance.

[canthearu]

Anyone to test Everest ?

You are welcome to become the first. An OCZ Petrol 64gig isn't terribly expensive.

[Christopher]

I thought about the petrol... But then I got another Turbo. Besides, Everest was probably just a bridge product to get to Everest 2. I am holding back some drives to test if the Arrowana FW ever gets released for the original Indilinx................

[Christopher]

Vertex Turbo 64 Day 9

63094.42
83.00MB/s

1130 Average Erase Count

78 MWI
216 hours

0/0/0 Program/Erase/Read Failures