SSD Write Endurance 25nm Vs 34nm

[B.A.T]

Todays update:
m4
758.0663 TiB
2714 hours
Avg speed 87.82 MiB/s.
AD gone from 179 to 174.
P/E 13152.
Value 01 (raw read error rate) has changed from 23 to 30.
MD5 OK.
Reallocated sectors : 00


Kingston V+100:
179.6779 TiB
710 hours
Avg speed 77.27 MiB/s
AD gone from 20 to 17.
P/E?
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
11,1354 TiB Host writes
18674hours
Reallocated sectors : 00
MWI=92 to 86
MD5 OK
41.31MiB/s on avg


The Mtron is a strange animal. I've tried both my internal SATA ports on 2 different motherboards, USB and esata on the same (M4A89TD Pro/Sabertooth 990FX) and USB and esata on my work laptop with no luck. I can get it to show in disk management but when I click on it the program hangs. I've also tried SATA 3Gb/s and 6Gb/s with no luck.

[Christopher]

The mtron is either just really picky, or really broken.

[B.A.T]

I'm going to try it in my folding rig during the weekend and see if that makes any difference. I can change between SATA 3 Gb/s and 1,5 Gb/s on that motherboard.

[Christopher]

Curious!

Strange that it did not work with USB either.

I bet it works with the folding rig, though.

[B.A.T]

Lets hope so. If not then I might need to buy an old pc just to make it work.

[Christopher]

That would be the only reason to buy a Pentium IV 478 system.

[B.A.T]

Talk about downgrading

[Christopher]

My Socket 478 Intel PERL865 had SATA 1.5GBPS and 0 + 1. My P4 2.6C was probably better in IPC than Bulldozer... [cheap shot at Bulldozer]

That motherboard was great. I always regretted not getting one of the other brand boards (because you couldn't really overclock it), but it had optical audio outputs, Intel GbE, and was working nice with my original WD Raptor 36GB drive... 36GB is still a lot to me.

[B.A.T]

It's always fun to have system that you remember for years after I've been going the amd way since 2003 and I'm not still there yet. Maybe I need that SB rig sooner than I thougt.

[bulanula]

What happened to Vapor ? What happened to the C300. Thanks !

[Anvil]

Kingston SSDNow 40GB (X25-V)

426.12GB Host writes
Reallocated sectors : 12

MD5 OK

33.48MiB/s on avg (~74 hours)

--

Corsair Force 3 120GB

01 90/50 (Raw read error rate)
05 2 (Retired Block count)
B1 84 (Wear range delta)
E6 100 (Life curve status)
E7 35 (SSD Life left)
E9 245727 (Raw writes)
F1 327164 (Host writes)

MD5 OK

102.56MiB/s on avg (~74 hours)

power on hours : 979

B1 (WRD) is down again, from 85 to 84.

[Anvil]

Todays update:
...
Intel X25-M G1 80GB
11,1354 TiB Host writes
18674hours
Reallocated sectors : 00
MWI=92 to 86
MD5 OK
41.31MiB/s on avg
...

So, it's slowing down, a lot!

I guessed 40-50MiB/s, hopefully it will stabilize > 40MiB/s.
Let's see how it develops the next few days...

[Christopher]

What happened to Vapor ? What happened to the C300. Thanks !

We're going to have to form a search party. Perhaps he's fallen down a well, or trapped at the airport.

Posse up...

[Christopher]

Mushkin Chronos Deluxe 60 Update, Day 35
3.3.2FW

05 4
Retired Block Count

B1 11
Wear Range Delta

F1 334738
Host Writes

E9 258195
NAND Writes

E6 100
Life Curve

E7 10
Life Left

Average 127.72MB/s Avg
Intel RST drivers, Celeron G530 Biostar TH67+

804 Hours Work (26hrs since the last update)
Time 1 month 3 days 12 hours

12 GiB Minimum Free Space

[B.A.T]

So, it's slowing down, a lot!

I guessed 40-50MiB/s, hopefully it will stabilize > 40MiB/s.
Let's see how it develops the next few days...

At the moment it's around 36.5 MiB/s. Maybe I need to to SE regulary or the test will take forever.

[Christopher]

If the G1 is only going to write as fast as an X25-V, it will take something like 2 years to kill it.

[B.A.T]

2 years is a long time to test
I need to develop some routine for SE. To bad intel closed off the toolbox for G1, because then it would have been real easy.

[Christopher]

That's kind of a tough problem to solve. If you assume that the flash in the G1s lasts twice as long as the 34nm flavor, and you have twice as much of it as a X25-V, then factor in the sustained write speed (very similar to the X25-V)...

And you have a problem.



One thing you (potentially) could do to keep speed higher for longer is over provision the drive, but that's just going to make the drive last even longer...

Stopping to secure erase the drive ever few days: takes a lot more time, and more direct supervision.



B.A.T., you have your work cut out for you...


Of course, the M4 might last for another couple years too. Who knows? I would hate to bet against it at this point. And so far every drive has been a little surprising, but none more so than the M4. It's too bad the drives don't explode when they're done. That would be way more exciting.

[B.A.T]

I'll just have to find a way. SE every 2 day takes 5 min so it might be possible. It will be less work when the m4 is finished, but that might take weeks. It still very impressive even if the ecc has started to work.

[Christopher]

All of the drives have been impressive. Maybe not the F40-A. But the M4 and both Intels are certainly impressive. The M4 is just getting "broken in".

[Anvil]

Kingston SSDNow 40GB (X25-V)

427.32GB Host writes
Reallocated sectors : 12

MD5 OK

33.39MiB/s on avg (~84 hours)

--

Corsair Force 3 120GB

01 92/50 (Raw read error rate)
05 2 (Retired Block count)
B1 83 (Wear range delta)
E6 100 (Life curve status)
E7 35 (SSD Life left)
E9 248616 (Raw writes)
F1 331003 (Host writes)

MD5 OK

102.55MiB/s on avg (~84 hours)

power on hours : 990

B1 (WRD) is down again, from 84 to 83.

[Anvil]

I'll just have to find a way. SE every 2 day takes 5 min so it might be possible. It will be less work when the m4 is finished, but that might take weeks. It still very impressive even if the ecc has started to work.

There is something you could try and that is simply filling the drive with large block sequential writes and then delete the file.
It won't be the same as TRIM of course but that was the "old" way of doing it, it can be done from ASU but the options is not available in the "open" beta.

[B.A.T]

Okay, my first ssd was the G2 so I've never tried those things before. What is the most easiest approach?

[Christopher]

I've copied acronis true image backup files to a striped raid array before, as its just one huge file (like 20 GB). Then you just delete it. I'm not sure how well it worked as I was trying to prevent performance degradation, so it wasn't terribly slow yet. Plus, that was with OPed G2 X25-Vs. The G2s were more resilient to begin with, so you may want to experiment with something like HD tach to take a performance baseline, then copy some huge .ISO/.tib/.whatever files and determine efficacy.

[B.A.T]

I can try something like that, it can actually just be a large zip file too. I'll test this tonight.