3rd Vertex2 gone bad

[deeppow]

Got a new 35nm Vertex2 from OCZ after an RMA of a 25nm Vertex2 to use on my business laptop. It lasted about 3 weeks and has died. The laptop is a DIY OCZ 17" one. My guess is that the laptop is killing the SSD because the chances of getting 3 bad SSDs from OCZ are so slim. OCZ doesn't know anything about a potential problem.

Oh well, what a pain. Think I'll just go back to a HD.

[aphellyon]

My hat's off to you sir... many would just rageflame OCZ. How did you figure out you had a 25nm for the first one you RMA'd?

[EarthDog]

Its in the model #...

Yeah damn, +1 to you for not going ape****. Weird situation you have there...

[CompuTamer]

I'd have to say it probably IS OCZ. It took me 2 Vertexes to find one that worked... first one died in a few days, seconded on in a few months. 3rd one seems to work though.


I vote go Intel or someone else.

[deeppow]

My hat's off to you sir... many would just rageflame OCZ. How did you figure out you had a 25nm for the first one you RMA'd?

The model # does typically provide the info but some have indicated they have the 25nm NAMDs when the model # doesn't say it ..... I can't validate that fact. The OCZ forum has indicated if the available storage space is less than 93% of the size sold, it usess the 25nm NANDs. That is how I found it and the OCZ support folks accepted that as validation of the fact.


I'd have to say it probably IS OCZ. It took me 2 Vertexes to find one that worked... first one died in a few days, seconded on in a few months. 3rd one seems to work though.


I vote go Intel or someone else.

It is possible all 3 have been bad but it is not probable.

[creecher]

ive had 2 go bad, Idk why. might be my mother board, but my hhd has been fine

[King107s]

So when you say they went 'bad' what exactly happened? Turn on the computer and system couldn't find the drive?

[deeppow]

So when you say they went 'bad' what exactly happened? Turn on the computer and system couldn't find the drive?

Computer hangs in bios boot till V2 is removed. With V2 removed, computer boots fine from DVD to Ubuntu in a live session and can see and use the HD that has all applications and my user data. No OS on the HD so I can't boot from that.

[Bobnova]

I had a laptop that would do that with dvd drives. The drives worked perfectly, it was the controller that had an issue.

[deeppow]

So when you say they went 'bad' what exactly happened? Turn on the computer and system couldn't find the drive?

Computer hangs in bios boot till V2 is removed. With V2 removed, computer boots fine from DVD to Ubuntu in a live session and can see and use the HD that has all applications and my user data. No OS on the HD so I can't boot from that.


Regarding the likelyhood of getting one or more OCZ SSDs that fail, you can use the info provided by Mr Alpha in this thread (http://www.overclockers.com/forums/showthread.php?t=670592). As he notes the failure rates only reflect the failure returned through the seller and in the first year.
1 failure: 3 out of 100 (3%)
2 failures: 9 out of 10 thousand (.09%)
3 failures: 27 out of million (.0027%)
4 failures: 81 out of 100 million

Estimates are only approximate nd might be considered the minimal likelyhood because all failures (such as those going to OCZ) aren't known.

[aphellyon]

The model # does typically provide the info but some have indicated they have the 25nm NAMDs when the model # doesn't say it ..... I can't validate that fact. The OCZ forum has indicated if the available storage space is less than 93% of the size sold, it usess the 25nm NANDs. That is how I found it and the OCZ support folks accepted that as validation of the fact.

Yea, I've heard not all drives with "E" model numbers are 25nm and not all drives without an "E" aren't. Thanks for the storage capacity tip. I knew it would take a hit due to extra over provisioning, but I've only seen numbers on the 120Gb drives... i wasn't sure if the percentages scaled with their larger drives. It would at least be a little nice if their "toolbox" would list the die size. At this point, I don't think I'll try to RMA my 180s if they're 25nm. I'm not happy with OCZ's bait'n switch business tactics, but so far the drives themselves seem to work well.

[deeppow]

Yea, I've heard not all drives with "E" model numbers are 25nm and not all drives without an "E" aren't. Thanks for the storage capacity tip. I knew it would take a hit due to extra over provisioning, but I've only seen numbers on the 120Gb drives... i wasn't sure if the percentages scaled with their larger drives. It would at least be a little nice if their "toolbox" would list the die size. At this point, I don't think I'll try to RMA my 180s if they're 25nm. I'm not happy with OCZ's bait'n switch business tactics, but so far the drives themselves seem to work well.

Somewhere I've read that there are 3 effects associated with the 25nm NANDs in the OCZ SSDs.
- less usable storage space
- slower performance
- shorter life of the 25nm NANDs.

If you are happy with what you have, great. I find RMAs to be a pain in the a$$ and time consuming to then repair/replace the apps and data.

[5er driver]

there are 3 effects associated with the 25nm NANDs in the OCZ SSDs.
- less usable storage space
- slower performance
- shorter life of the 25nm NANDs.


That's why I went Mushkin this time instead of Vertex 2......no surprises. But these issues are not limited to only OCZ SSDs, they apply to all 25nm nand drives.

34nm nand has 5000 write cycles, 25 nm nand has 3000.

I'm still using a Vertex 1st gen, 50nm nand, 10,000 write cycles.

[deeppow]

The reason the substation has ******ed most folks off is we weren't told up front. I'm don't accept the fact that "the model number changed." What am I suppose to do, memorize model #s so I can figure out something is different?

One thing about the upcoming next generation of SSDs, I see no way any manufacturer can avoid using 25nm NANDs. Of course the performance hits may be hidden by the SATA III connections and extra chips could be used to make up the lost storage space. The lose of life time probably can't be avoided. It'll be interesting to see if they reduce their warrenty time.

[Mr Alpha]

That's why I went Mushkin this time instead of Vertex 2......no surprises. But these issues are not limited to only OCZ SSDs, they apply to all 25nm nand drives.

34nm nand has 5000 write cycles, 25 nm nand has 3000.

I'm still using a Vertex 1st gen, 50nm nand, 10,000 write cycles.That is overstating it somewhat. Not all of these issues are universal to 25nm NAND.

The loss of space only applies to some SandForce drives in the 50 to 200GB range. The reason you loose space is SandForce's RAISE (Redundant Array of Independent Silicone Elements) technology. What it does is basically RAID5 where you loose one element to parity data. So on a 34nm drive with 16 NAND chips you loose 1/16th of the storage space to parity. When they moved to 25nm NAND they went from 16 to 8 NAND chips in drives smaller than 200GB and with only 9 chips you loose 1/8th of the storage space to parity.

The slower performance is also a result of going from 16 to 8 NAND chips. The channels on the SandForce controller are 16 bit wide and your typical MLC NAND chip is only 8 bit wide so you could hook two NAND chips up to the same channel, which gave a performance improvement. But when they went from 16 to 8 NAND chips this improvement was lost.

Neither of these issues affect the really big SandForce drive because they still use 16 chips even with 25nm NAND. It is also possible to use 16 smaller 25nm chips instead of 8 bigger ones, which would solve the previous two problems, but they cost more per GB. As for performance of the NAND chips themselves, I believe they usually are a bit faster than 34nm chips.

The issues of reduced write cycles is the one issue that is mostly universal to 25nm NAND. I say mostly because there is premium 25nm NAND rated for 5000 write cycles. Comes with a price premium though. Best way to deal with the reduced write cycles is to buy a bigger drive.

[deeppow]

Very informative Mr Alpha. Thanks.

Based on what you say I would conclude the next generation of SSDs should be able to accommodate some of the differences via their designs. Length of chip life would require $s.

[5er driver]

The slower performance is also a result of going from 16 to 8 NAND chips.

As for performance of the NAND chips themselves, I believe they usually are a bit faster than 34nm chips.

Slower performance on the 25nm nand will also occur as a result of the controller invoking the LTWT (LifeTime Write Throttle) to help increase the nand life which happens if 10s of GBs are written to the drive in a short time or after every cell is written to, whichever occurs first.

Performance/Spec wise, the 25nm chips don't appear to be faster than 34nm in comparing just these two drives. Both the 34nm Mushkin Callisto Deluxe 240GB and the 25nm Vertex 2 240GB are specd at "Read: up to 285MB/s, Write: up to 275MB/s". On my system the Mushkin is running reads at 280MB/s and writes at 270MB/s. I'll post the ATTO when I get back to the laptop. I don't know what users are actually getting with the Vertex 2. Both Mushkin and OCZ spec there drives using ATTO.

EDIT: Here's the ATTO bench.


http://i152.photobucket.com/albums/s199/5erdriver/PC%20Stuff/ATTOMushkinCallistoDeluxe240GBwithW764.png

[dropadrop]

I got a Vertex LE for my iMac, but decided to run it in my laptop for a while before swapping it over (since it's not a very straight forward install). It died the first time the laptop went to sleep...

I took a refund from the shop, and will probably end up getting something with a different chipset. I do note from users experiences that OCZ seems to have more problems then anyone else, but still it feels absurd.

[Mr Alpha]

Performance/Spec wise, the 25nm chips don't appear to be faster than 34nm in comparing just these two drives. Both the 34nm Mushkin Callisto Deluxe 240GB and the 25nm Vertex 2 240GB are specd at "Read: up to 285MB/s, Write: up to 275MB/s". On my system the Mushkin is running reads at 280MB/s and writes at 270MB/s. I'll post the ATTO when I get back to the laptop. I don't know what users are actually getting with the Vertex 2. Both Mushkin and OCZ spec there drives using ATTO.That is hardly an informative test since both drives are limited by the SATAII interface. Nor is ATTO a good test becaue it uses a queue depth of 4, which is nice in that it is the type of queue depth you might see on a desktop, but since it lest the controller doing its thing it is useless for measuring the speed of the NAND.

[5er driver]

That is hardly an informative test...
That's right, it isn't. But that's how those two and some of the other manufacturers spec their drives for marketing purposes. So, to see if your drive is "up to spec" (their specs), you would run ATTO. It would be nice if they would all agree to use one bench that would show true figures for day-to-day use, server use, different scenarios, etc.

[King107s]

That is hardly an informative test since both drives are limited by the SATAII interface. Nor is ATTO a good test becaue it uses a queue depth of 4, which is nice in that it is the type of queue depth you might see on a desktop, but since it lest the controller doing its thing it is useless for measuring the speed of the NAND.

That's right, it isn't. But that's how those two and some of the other manufacturers spec their drives for marketing purposes. So, to see if your drive is "up to spec" (their specs), you would run ATTO. It would be nice if they would all agree to use one bench that would show true figures for day-to-day use, server use, different scenarios, etc.

I agree with both of you here.... Certainly benchmarks like ASSD will show you when you have a 'problem' when ATTO will not.

[Marshmallow64]

I have had 2 vertex 1s go bad on me already. One lasted about a year, other under 2 days. This one is running strong so far though. *knock on wood*

[5er driver]

The first Vertex I got 2 years ago was DOA out of the box.

[deeppow]

I think this thread has had some good discussion but it might be time to let it die and move on to other threads. Many thanks for your thoughts, suggestions and input.

I believe Mr Alpha is writing a front page article on SSDs (don't think it has yet been published) so you might keep and eye open for it. :thup: