Had enough of OCZ

[Arachnotronic]

I bet most of OCZ's problems were due to the SandForce controller. Every Vertex 4/Agility 4 I've used has been fantastic. Marvell controllers are quality, and OCZ's custom firmware is quite good.

 

[StarTech]

And sorry, I don't buy the "I'm up-to-date" one bit. You had to have been living under a rock these past couple of years to not have heard without Anand publishing a rant for you.

I agree. Anyone using SSDs that is "up-to-date" is aware of all the problem history and anyone that is not aware of the problem history cannot claim that is up to date because is not.

 

[VirtualLarry]

I bet most of OCZ's problems were due to the SandForce controller. Every Vertex 4/Agility 4 I've used has been fantastic. Marvell controllers are quality, and OCZ's custom firmware is quite good.

Vertex Plus also had a lot of issues when it first came out. The now-named "Arowana" FTL layer firmware was also delayed for the older Barefoot controller-based SSDs for months, apparently due to bugs. (The Vertex Plus was their first Barefoot-based SSD that shipped with that "Arowana" firmware from the factory.)

I've been using some 30GB OCZ Agility (Indilinx Barefoot) SSDs with firmware 1.7 for four months or so, no problems to report.

Also just recently purchased three Vertex Plus "R2" 120GB SSDs, which use the Barefoot "2" controller, and also some modified Arowana firmware.

So far, for the last two weeks or so, they've been stable too.

Purchased a pair of OCZ Agility 2 90GB SSDs (1st-gen SandForce). Since 1st-gen SandForce still has issues with sleep and hibernate, I chose to put those SSDs into always-on desktop machines only. They are definitely faster than the 30GB Agility (Barefoot) SSDs.

However, my desktop once showed at the boot stage in the morning. Unsure if it was a power glitch (the other desktop was fine), or whether the desktop blue-screened, possibly due to the Agility 2 SSD.

 

[Fx1]

That's like saying a car's claim to safety with a bazillion airbags gives people a false sense of security because you lost a family member when their car burned to the ground.

Airbags protect from impacts, not fire.

No moving parts protect from shock (which can be a factor with notebooks), not beta firmware.

Understand what is being claimed and don't let your mind fill in the blanks, and you will have better understanding.

FWIW HDDs are not immune to failure and data loss either, nor are they immune to firmware issues. Just ask anyone who had a Seagate 7200.11 drive brick itself a few years back. Yup, mechanically sound, but faulty firmware. Also, check HDD end user reviews and you'll see numerous 1/5 ratings. This Seagate drive has 50% 1/5 ratings. This Hitachi drive has 38% 1/5 ratings. This WD drive has 33% 1/5 ratings. Just go read some of those end user reviews and you'll hear about DOAs, drives that died within minutes, drives that didn't last a month, several in a row that died, etc.

Speaking of data loss, it could happen regardless of which brand drive you have, or even if you have a HDD instead of an SSD.



That is all he can base it off. If he were to test for reliability, he would need to have a big batch of them deployed in a number of systems for a long duration.

For instance car magazines often have long term tests where they would keep a car for a whole year, and document monthly how it was doing for reliability, maintenance, cost of ownership and any user notes (comfort, quirks, etc.). It makes sense for a car because you might have your eye on one to buy used down the road.

For a computer part where the whole industry is constantly sprinting forward in releasing new and faster products, long term testing is a horrible idea because then you would see a new article posted about the Intel G2 or Crucial RealSSD/C300, when you can no longer buy one new at most retailers.

So, Anand does what he can do to hammer the drive (and he explains all that he puts it through) to see if he can produce bad behavior from it, without taking months/years to accomplish.



Why would you expect an "upgrade" as a replacement? That's like expecting Ford to give you a Fusion to replace your Focus that was a lemon.

You purchased a product. If it fails under warranty, you have the expectation that whoever provides the warranty will "make good" on the product. That means, they will either fix it, or replace it with an otherwise identical product. If an otherwise identical product is not available, then they may opt to give you a BETTER product.

Are you saying that an Agility 3 is not a better product than a Vertex 2? It is a newer generation in both product and the Sandforce controller. It has a faster interface. It benchmarks faster. It is, in all measurements that matter, an actual upgrade over your Vertex 2. So what is your problem with it?

Other websites published the stories so lets not go down that age old american excuse "we will get sued"

AT gets top level access at OCZ at the CEO level. This could have been brought up very easily. We have all read about it.

Put it this way. If Nvidia or Intel had failure rates of GPU and CPU's as bad as these Sandforce drives there would be much more noise made about it.

2 or 3 years ago AT would have had the balls to call out the OCZ CEO in relation to the poor image of the company and terrible failure rate. Heck the attitude towards OCZ on AT's forum speaks volumes.

I cant help but feel like a more commercial approach has been taken over the last couple of years.

 

[Fx1]

I agree. Anyone using SSDs that is "up-to-date" is aware of all the problem history and anyone that is not aware of the problem history cannot claim that is up to date because is not.

didnt i write above that i only kept up to date on SSD's by reading the AT stories? i admitted that i was gleefully unaware that these drives had such bad reputations.

Hence my frustration at having to find out just how bad they were myself.

Not enough has been reported on AT which people once considered the go to place for all things "SSD"

 

[Railgun]

As opposed to their own support forums....

 

[Hellhammer]

I'll try to clarify a few things:

1. Most of the drives we test are brand new, which means there are no NewEgg reviews or other user experiences with the drive. Or if there are, the sample size and time period are way too small to make any realistic conclusions. Using older drives as a basis is not exactly accurate either because we may be dealing with a totally different product (new controller, firmware, NAND etc).

2. There are very few confirmed statistics of SSD reliability. NewEgg reviews can give you a good idea of what's going on, but there are too many variables to make them accurate. For example, people give bad reviews because of their own mistakes (e.g. not reaching advertised speeds because of SATA 3Gbps motherboard). Moreover, is there any way to make sure all satisfied customers leave a positive review? So while there may be 20% bad reviews, the real percentage of bad reviews may easily be below 10%. I've started to use NewEgg stats more in our recent reviews, but their usefulness is fairly limited.

3. When we post something, we have to be sure it's accurate. Hence the only way for us to post about an issue is to recreate it ourselves and make sure there really is an issue. Another way is if the manufacturer admits that there is an issue, but that is rather rare. This isn't limited to just SSDs, it applies to all products we test.

If we just trust someone else's word in the forums and the data ends up being inaccurate, we would get an egg on our face. It might be acceptable once or twice, but sooner than later manufacturers would start questioning our reliability. The problem is that there is lots of untrustworthy info in forums. I'm sure there are plenty of users here who hate OCZ even though they have never owned a product from them. Is it valid to count them as dissatisfied customers then? We have no way of making sure that what is said is the truth. People can be very aggressive when it comes to brands (think about an Apple fan for instance...).

4. For months, our recommendation has been Samsung's 830. Based on our own testing, it's fast and has also stood up as a stable drive (Anand has been using it as a boot drive for as long as I can remember).

I'm not trying to defend OCZ or any other brand. However, you must be a lot more careful when the content you're publishing is read by millions of people. You can't just go and publish something based on word of mouth or other unreliable sources.

 

[jwilliams4200]

Short version:

Quality testing is hard. Very hard. Too hard.

 

[Fx1]

I'll try to clarify a few things:

1. Most of the drives we test are brand new, which means there are no NewEgg reviews or other user experiences with the drive. Or if there are, the sample size and time period are way too small to make any realistic conclusions. Using older drives as a basis is not exactly accurate either because we may be dealing with a totally different product (new controller, firmware, NAND etc).

2. There are very few confirmed statistics of SSD reliability. NewEgg reviews can give you a good idea of what's going on, but there are too many variables to make them accurate. For example, people give bad reviews because of their own mistakes (e.g. not reaching advertised speeds because of SATA 3Gbps motherboard). Moreover, is there any way to make sure all satisfied customers leave a positive review? So while there may be 20% bad reviews, the real percentage of bad reviews may easily be below 10%. I've started to use NewEgg stats more in our recent reviews, but their usefulness is fairly limited.

3. When we post something, we have to be sure it's accurate. Hence the only way for us to post about an issue is to recreate it ourselves and make sure there really is an issue. Another way is if the manufacturer admits that there is an issue, but that is rather rare. This isn't limited to just SSDs, it applies to all products we test.

If we just trust someone else's word in the forums and the data ends up being inaccurate, we would get an egg on our face. It might be acceptable once or twice, but sooner than later manufacturers would start questioning our reliability. The problem is that there is lots of untrustworthy info in forums. I'm sure there are plenty of users here who hate OCZ even though they have never owned a product from them. Is it valid to count them as dissatisfied customers then? We have no way of making sure that what is said is the truth. People can be very aggressive when it comes to brands (think about an Apple fan for instance...).

4. For months, our recommendation has been Samsung's 830. Based on our own testing, it's fast and has also stood up as a stable drive (Anand has been using it as a boot drive for as long as I can remember).

I'm not trying to defend OCZ or any other brand. However, you must be a lot more careful when the content you're publishing is read by millions of people. You can't just go and publish something based on word of mouth or other unreliable sources.

Thanks for clearing that up.

I think ill either go intel or Samsung for my next drive. Probably Samsung given the price advantage.

 

[bryanW1995]

Other websites published the stories so lets not go down that age old american excuse "we will get sued"

AT gets top level access at OCZ at the CEO level. This could have been brought up very easily. We have all read about it.

Put it this way. If Nvidia or Intel had failure rates of GPU and CPU's as bad as these Sandforce drives there would be much more noise made about it.

2 or 3 years ago AT would have had the balls to call out the OCZ CEO in relation to the poor image of the company and terrible failure rate. Heck the attitude towards OCZ on AT's forum speaks volumes.

I cant help but feel like a more commercial approach has been taken over the last couple of years.

That's because Nvidia and intel don't have the worst reliability in the industry. Everybody just assumes that if you bought OCZ then you did it because it was cheaper. Their sf-2281 drives are a great case in point. Intel isn't having the same issues that OCZ did, and I'm confident that neither Samsung nor Crucial would have, either. Why? Because they wouldn't have released a product on beta firmware. OCZ has the benefit of having a terrible quality reputation, anyway, so they had nothing to lose by pushing those drives out before they were ready.

 

[bryanW1995]

snip

3. When we post something, we have to be sure it's accurate. Hence the only way for us to post about an issue is to recreate it ourselves and make sure there really is an issue. Another way is if the manufacturer admits that there is an issue, but that is rather rare. This isn't limited to just SSDs, it applies to all products we test.

If we just trust someone else's word in the forums and the data ends up being inaccurate, we would get an egg on our face. It might be acceptable once or twice, but sooner than later manufacturers would start questioning our reliability. The problem is that there is lots of untrustworthy info in forums. I'm sure there are plenty of users here who hate OCZ even though they have never owned a product from them. Is it valid to count them as dissatisfied customers then? We have no way of making sure that what is said is the truth. People can be very aggressive when it comes to brands (think about an Apple fan for instance...).

snip

I cannot count the number of times that I've been on other tech websites and seen them reference information from Anandtech, and they typically are doing it to prove the point they are trying to make. To wit: "According to Anandtech, XFX builds quality PSU's" or "Even Ryan Smith uses overclocked gpus from competitors for comparison purposes when new gpus come out". You guys do a great job, keep up the good work!

 

[jwilliams4200]

So while there may be 20% bad reviews, the real percentage of bad reviews may easily be below 10%. I've started to use NewEgg stats more in our recent reviews, but their usefulness is fairly limited.

Their usefulness is even more limited if people misuse the data as Zap did in post #23. Two of Zap's three examples had only 8 reviews.

For these purposes, you really need a bare minimum of 100 reviews to have any chance of having a decent margin of error. Of course, a lot more than 100 would be even better. Sometimes you need to combine statistics from different capacities to get enough reviews, or even combine all the models from a manufacturer if there are not enough reviews of a given model.

 

[sm625]

Part of the problem with OCZ is that they went out of their way to cover up problems listed even on their own forums. They should be out of business imo. I cant imagine why anyone would buy anything from them. Unless they release an SSD that beats the competition by 10x, they are dead to me.

 

[jwilliams4200]

Part of the problem with OCZ is that they went out of their way to cover up problems listed even on their own forums. They should be out of business imo.

Dishonesty, combined with poor quality, is a deadly combination. Or at least it should be. OCZ manages to hang on. Their CEO is a master of marketing.

 

[bryanW1995]

OCZ has actually done a lot to push SSD's into mainstream consumers' buying consideration. Most reasonably knowledgeable buyers would spend the extra money for something else, but there will always be those who buy the cheapest no matter what.

 

[Zap]

Other websites published the stories so lets not go down that age old american excuse "we will get sued"

Did I say anything about lawsuits? You're letting your mind fill in the blanks again.

Put it this way. If Nvidia or Intel had failure rates of GPU and CPU's as bad as these Sandforce drives there would be much more noise made about it.

Nvidia had its "Bumpgate."

Their usefulness is even more limited if people misuse the data as Zap did in post #23. Two of Zap's three examples had only 8 reviews.

That's probably because they are new models. How about this one?

Seagate Barracuda 7200.11 1.5TB

Around 2600 end user reviews with 29% 1/5. Is that a good enough example?

 

[steve wilson]

I have read a lot of SSD's reviews on this site and have seen Anand continually mention the problems OCD have had. In every single review of new technology with no track record he always says something like "As always with new tech there is a risk on reliability, buy with caution, or wait a few months to see if there are problems for other people". It's not anand's fault at all, it's OCZ's and maybe a little time waiting and research by yourself on these forums/later reviews you would of found out. That's what I did and that's why I avoided the OCZ drives like the plague.

 

[thelastjuju]

If you buy an OCZ drive, or any other product that gets disproportionately negative reviews, you only have yourself to blame for being too lazy to do even minimal research on an integral part of your computer.

OF COURSE you are more likely to review a bad product than a working one.. SO, when a product has over hundreds to thousand of individual reviews, and it DOESN'T receive an overwhelming amount of negative feedback on them, it can safely be interpreted as a GOOD thing and a SOLID product. :thumbsup:

 

[jwilliams4200]

Seagate Barracuda 7200.11 1.5TB

Around 2600 end user reviews with 29% 1/5. Is that a good enough example?

Sure, 2600 reviews is great for having a lower margin of error.