SSD Write Endurance 25nm Vs 34nm

Quote:

---

Originally Posted by Tony

request an RMA...

I have run a 256GiB Vtx4 on 1.5 continually for days/weeks stress testing 100% of the drive, it got HOT, it kept going and its now sat in my macbook pro.

Vtx4's can run warm, you may have just heated that drive up and it took a dump...

Seriously though, 1 guy with an issue making it sound like the whole FW is broken...you really do need to see the big picture, I ONLY act and report issues when I see there is an issue...if we all jumped when 1 guy reports his drive took a dump you would never get beta FW never mind final

I will say this 1 last time...the VTX4 is designed to deliver fast writes BUT you have to let the drive FW work as intended, this means there is a possibility to push the drive outside of how it was designed to work....BUT end users who just use the drives never do this. In most cases the drive will just crawl...in your case something else happened but its never happened to me or any of the beta testers and not at HQ where we have been testing Vtx4 continually for months now continually.

We do NOT have a FW issue...you just think we do, if we did I would have a forum full of people with the same exact issue...and I don't, and neither does inhouse support...please see there is a massive machine watching and they jump ONLY when there is something to jump for.

If you want an RMA please PM Ryder on the ocz forum....and please stop wearing out your drive needlessly, we do not warrant nand burn remember, no one does

---

Simmer down, Tony.

Hmmm.

Quote:

---

Originally Posted by Tony

......its never happened to me or any of the beta testers and not at HQ where we have been testing Vtx4 continually for months now continually.

---

So if I follow the procedure that canthearu identifies in his post over @ OCZ you are sure I will not see the same thing? Care for a bet before I start?

Sandisk Extreme G25 - 120gb - Day 83

Drive hours: 1925
ASU GiB written: 821,335.01 GiB (802.09 TiB)
Avg MB/s: 124.95 MB/s (23.31 hours)
MD5: OK

Host GB written (F1): 826,816 GiB (807.44 TiB)
NAND writes (E9): 617,345 GiB (602.88 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 110 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 4 raw
Reported Uncorrectable Errors (BB): 0 raw

Steady as it goes for the sandisk extreme.

Attachment 129685

I wonder if the Kingston E100 is as reliable as they claim: http://www.techpowerup.com/171171/Ki...prise-SSD.html .
Weird that they would only give it a 3 year warranty with these claims...

Samsung 830 256GB Day 157

(GiB) 3,942,364
(TiB) 3,849
(PiB) 3.78


(Avg) 297.58 MB/s over the past 1200 hours

(B1) Wear Leveling Count: 17,063

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864
(POH) 3887


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

I've been out of town, so I've haven't had the chance to post any updates lately, but TBH not much has happened... just a lot more writes. We're getting close to the 4 million GiB mark though.

'Twill be a momentous occasion. But then again, the M4 is at almost 1.5 PiB.

You didn't know but this Samsung is in fact a demigod, it can throw the data you send to him to the heavens so he's not hurt :P

Haven't checked on this thread in a long while. That M4 is just insane, 1.5BB on a 64GB drive is just ridiculous. http://smiliesftw.com/x/eek5boink.gif

Almost 4PB....that's damn impressive. Would love to see more 256GB drives get put to the test. It'd also be interesting to see how the 128GB 830 fairs to see just how much of an advantage the increased nand amount actually yields...or close to it maybe.

The M4 and the 830 are the 6gbps drives you want if you like writing a lot of data. Interestingly, the 64GB M4 and the 256GB 830 are about identical in terms of avg speed/capacity. Coincidence? Perhaps, but I don't think you'd have an issue getting 1.5PB on a 64GB 830 either.

Yes, Coincidence. They're just well built drives, I doubt Samsung and Micron would be sharing anything with each other.

Quote:

---

Originally Posted by Kain665

Yes, Coincidence. They're just well built drives, I doubt Samsung and Micron would be sharing anything with each other.

---

No, I mean, if the 830 is 4x larger and 4x faster than the m4, perhaps they have equal amounts of recovery time and this is a large part of their success. It could just be their proprietary nand management techniques. It could be lots of things, but if you own a fab and have a proprietary controller (the M4's Marvell is pretty much their own controller in the end), you should be able to make a drive that lasts a long, long time.

Been a while since my last update. My holyday is over and I get more time to have fun here :)

Here are todays update:

m4
1487.5670 TiB
6148 hours
Avg speed 74.24 MiB/s.
AD 59 to 17
P/E 25552.
C3 4647 to 5683
CE 58
MD5 OK.
Reallocated sectors : 00
http://www.diskusjon.no/index.php?ap...tach_id=501574

Intel X25-M G1 80GB .
882.42 TiB
25434 hours
Reallocated sectors : 107 to 255 to 74
Available Reserved space: 0 to 59 to 31
MWI= 38 to 255 to 233
MD5 =OK
21.65 MiB/s on avg
http://www.diskusjon.no/index.php?ap...tach_id=501575

Intel X25-E 64GB
830.36 TiB
2615-30=2585 hours
Reallocated sectors : 48 to 148
Available Reserved space: 98 to 95
MWI= 95 to 94
MD5 =OK
90.48 MiB/s on avg
http://www.diskusjon.no/index.php?ap...tach_id=501573

Mtron Pro 7025 32GB
Paused until new firmware is updated.

The G1 decided that it had some more spare area hidden so the counter went from 0 to 59 and started to count down again. Very strange behaviour.
The m4 just keeps going and going and going.... Not much change just a little more cumulative corrected ecc.

What does concern me is the number of bad sectors on that SLC drive ....

That suprised me too. It burns through sectors at the same rate the G1 did in the begining.

Quote:

---

Originally Posted by Christopher

No, I mean, if the 830 is 4x larger and 4x faster than the m4, perhaps they have equal amounts of recovery time and this is a large part of their success. It could just be their proprietary nand management techniques. It could be lots of things, but if you own a fab and have a proprietary controller (the M4's Marvell is pretty much their own controller in the end), you should be able to make a drive that lasts a long, long time.

---

Agreed.

Samsung 830 256GB Day 159

(GiB) 3,989,792
(TiB) 3,896
(PiB) 3.83


(Avg) 297.73 MB/s over the past 1200+ hours

(B1) Wear Leveling Count: 17,267

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864

Attachment 129744
(POH) 3932


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

I'm starting to get the feeling that this 830 might not ever die. Perhaps there is a critical point of wear beyond which the drive starts degrading rapidly? I can only hope.

Maybe it is the error correction that makes the difference. Drives with weak error correction implementation will show a gradual increase of bad sectors, just like Intel G1 while those with a strong one will have a fast degradation at the end. Probably the lack of a complex ECC in X25-E 64GB (at least not as complex as in the MLC drives) is forcing it to mark bad blocks faster.

Sandisk Extreme G25 - 120gb - Day 86

Drive hours: 2002
ASU GiB written: 855,241.02 GiB (835.20 TiB)
Avg MB/s: 125.53 MB/s (100.03 hours)
MD5: OK

Host GB written (F1): 860,890 GiB (840.71 TiB)
NAND writes (E9): 642,808 GiB (627.74 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 115 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 4 raw
Reported Uncorrectable Errors (BB): 0 raw

Samsung 830 256GB Day 160

(GiB) 4,012,542
(TiB) 3,918
(PiB) 3.85


(Avg) 297.75 MB/s over the past 1200+ hours

(B1) Wear Leveling Count: 17,365

(B6) Erase Fail Count: 9
(05) Reallocated Sectors: 36864

(POH) 3953


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

Hey guys,

Think we can get a graph done with all drives that have been tested and at what point they died? and the fact I want to see the Samsung 830 at 4PB :)

The first page of the thread has most of what you're looking for. Anvil will occasionally post a current report for drives currently being run thru the ringer.

Here are todays update:

m4
1502.8526 TiB
6209 hours
Avg speed 74.13 MiB/s.
AD 17 to 8
P/E 25818.
C3 5683 to 6349
CE 58
MD5 OK.
Reallocated sectors : 00
http://www.diskusjon.no/index.php?ap...tach_id=501794

Intel X25-M G1 80GB
883.15 TiB
25495 hours
Reallocated sectors : 74 to 202
Available Reserved space: 31 to 24
MWI= 233 to 1
MD5 =OK
20.25 MiB/s on avg
http://www.diskusjon.no/index.php?ap...tach_id=501795

Intel X25-E 64GB
848.63 TiB
2676-30=2646 hours
Reallocated sectors : 148 to 172
Available Reserved space: 95 to 94
MWI= 94
MD5 =OK
90.18 MiB/s on avg
http://www.diskusjon.no/index.php?ap...tach_id=501793

Mtron Pro 7025 32GB
Paused until new firmware is updated.

When I took the screendump of the ASU G1 the program was slow and felt non-responsive. I had to wait 4-5 sec before I could save the screendump. I think something might happen this weekend so I'll check up on it more often then normal.... Exciting :D

Sandisk Extreme G25 - 120gb - Day 88

Drive hours: 2049
ASU GiB written: 876,012.60 GiB (855.48 TiB)
Avg MB/s: 125.50 MB/s (147.13 hours)
MD5: OK

Host GB written (F1): 881,765 GiB (861.10 TiB)
NAND writes (E9): 658,408 GiB (642.98 TiB)

Retired Block Count (05): 4 raw
Failure count (AB, AC): 1 program, 1 erase
Raw Error Rate (01): 117 normalized
Media Wearout Indicator (E7): 10 normalized
Wear Range Delta (B1): 4 raw
Reported Uncorrectable Errors (BB): 0 raw