What's wrong with sandforce?

[Zxian]

Did you read the quote I replied to? OCZ still being unreliable is no myth, regardless of what controller they use.

And how many of those were based off the SF-2281 that caused so much trouble for early adopters? One. The topic at hand is about Sandforce, not OCZ. Your information is moot in this thread.

If the SSDs aren't faulty, what expectations or features do you think caused OCZ users to completely blow out return rates, but not other vendors?

It's one thing to bury your heard in the sand an ignore the stats, but it doesn't change the fact that the other SSD vendors are leagues better. Remember, these are recent figures.

I never said that the Petrol or Octane (or Vertex2 for that matter) drives weren't faulty. If you re-read your own page, the author writes the following:

The most popular ranges, namely the Vertex 3s and Agility 3s, do relatively well with returns of 1.51% and 2.03% respectively.

-Source (for the lazy)

Now who's ignoring the stats for the sake of an argument?

You're also not looking at the overall volume of sales. OCZ as a whole went down in terms of returns. The Petrol and Octane were terrible drives - we all get it. (Note, it was the SATA2 versions of the Octane drives that had such high return rates). Nobody in their right mind would recommend those drives. If you take those off the table, what do you get? You get the Vertex2 (which is unavailable now) and Vertex Plus. OCZ was one of the most aggressive with their advertising during the early days of SSDs, and as a result gained a lot of market share. This also increases the sheer number of cases you see where people post about their troubles on forums such as this one. As IGemini mentioned, OCZ got the bad end of the stick for nearly all the troubles caused by the poor early firmware for the SF-2281.



Now, to get back on topic, all of the major firmware issues related to the SF-2281 controller have been resolved. Remember, all of those issues affected ALL manufacturers that used the SF-2281 controller (Corsair Force, ADATA, Patriot, OCZ, and others). Intel came into the game after several months of testing and after most of the waters had settled, which is why you haven't heard of any serious issues related to the 520 and 330 drives (aside from the AES-256 bug, which they generously offered refunds for).


The Sandforce SF-2281 is no longer "that bad". If the OP has the ADATA drive he mentioned, all he should do is make sure that he's got the latest firmware available from ADATA installed on the drive and he should be good to go.

 

[taltamir]

Most of the negative attitude towards OCZ came from the first few months after the Vertex3 was released. Sandforce firmware was unstable on certain platforms during this time (this affected ALL SF-2281 drives - not just OCZ). All of these firmware issues have since been resolved.

No they haven't, see post 10 http://forums.anandtech.com/showpost.php?p=34376973&postcount=10
for which issues remain.

 

[John Connor]

My Adata is a S599. Not sure if it's a SF-2281 or not. I tried updating the firmware a while ago, but it wouldn't work. I think the firmware installer is for Win 7 only and i have XP 64. Like I said though, I won't be using XP 64 for much longer when I build my new machine.

 

[jwilliams4200]

So, keep on trying to knock me off my self-righteous high horse?..

Huh? It makes no difference to me if you want to be self-righteous. Knock yourself out.

All I have to say is that it is ironic to hear you complaining about off-topic posts.

 

[blastingcap]

OCZ does not have custom firmware, and neither does Intel. As a matter of fact, many of the fixes for the SandForce firmware came at the behest of Intel. However, Intel still uses the same as all SF SSDs.
These mfrs would try to have you believe otherwise, but just ask any SF rep.

"Intel's strenuous validation will eventually make SandForce's drives better for everyone, but for now the Cherryville firmware remains exclusive. Intel wouldn't go on record with details of its arrangement with SandForce, but from what I've managed to piece together the Intel Cherryville firmware is exclusive for a limited period of time. That exclusivity agreement likely expires sometime after the SF-2281 is replaced by a 3rd generation controller. There are some loopholes that allow SandForce to port bug fixes to general partner firmware but the specific terms aren't public information. The important takeaway is anything fixed in Intel's firmware isn't necessarily going to be fixed in other SF-2281 based drives in the near term. This is an important distinction because although Cherryville performs very similarly to other SF-2281 drives, it should be more reliable."

http://www.anandtech.com/show/5508/intel-ssd-520-review-cherryville-brings-reliability-to-sandforce

Given Intel's generally good rep, I don't believe they would rush something out to market if it weren't at least decent. So the only Sandforce drives I would consider buying would be Intel SSDs, at least at this time.

 

[JellyRoll]

lol, just because you quote someone and put it in bold does not make it true.

This is Intel spin, much like the "Super-DOOPER OHMG consistent performance" of their Enterprise SSDs. When in fact they are no more consistent than any other Enterprise SSD, or even consumer SSDs with a hefty bit of OP. This wouldn't be the first time this site were duped by Intel marketing razzmatazz.
I told you, mfrs are not going to be quick to admit these things.
Do you seriously think that any bugfix recommended by Intel isn't going directly into all SF gear?
And that is how it worked, Intel recommended a bunch of bugfixes, and they were ported over to SF firmwares prior to Intels release. Did you not note the timing of the BSOD fix? Right before LSI purchased SF, and also before Cherryville launch. Certain things had to be done to clean up SF's image before those two events occurred

 

[Amnesia1187]

My biggest issue with Sandforce is the performance with incompressible data. Makes full disk encryption a huge burden.

 

[Mfusick]

I've owned 15+ OCZ SSD with Sandforce and none of them have ever had an issue.

Stats or no stats I'm not changing my mind unless one of my SSD fail.

I saved about $20 on average each purchase buying the Vertex over another so by my estimation I have 15 times $20 on a replacement budget for the first that fails. I'm way ahead of game. All of these have Sandforce controllers. I'm not scared.

 

[jacktesterson]

I've owned and use numerous OCZ SSDs including a vertex 3 without issue.

 

[Mfusick]

And how many of those were based off the SF-2281 that caused so much trouble for early adopters? One. The topic at hand is about Sandforce, not OCZ. Your information is moot in this thread.



I never said that the Petrol or Octane (or Vertex2 for that matter) drives weren't faulty. If you re-read your own page, the author writes the following:


-Source (for the lazy)

Now who's ignoring the stats for the sake of an argument?

You're also not looking at the overall volume of sales. OCZ as a whole went down in terms of returns. The Petrol and Octane were terrible drives - we all get it. (Note, it was the SATA2 versions of the Octane drives that had such high return rates). Nobody in their right mind would recommend those drives. If you take those off the table, what do you get? You get the Vertex2 (which is unavailable now) and Vertex Plus. OCZ was one of the most aggressive with their advertising during the early days of SSDs, and as a result gained a lot of market share. This also increases the sheer number of cases you see where people post about their troubles on forums such as this one. As IGemini mentioned, OCZ got the bad end of the stick for nearly all the troubles caused by the poor early firmware for the SF-2281.



Now, to get back on topic, all of the major firmware issues related to the SF-2281 controller have been resolved. Remember, all of those issues affected ALL manufacturers that used the SF-2281 controller (Corsair Force, ADATA, Patriot, OCZ, and others). Intel came into the game after several months of testing and after most of the waters had settled, which is why you haven't heard of any serious issues related to the 520 and 330 drives (aside from the AES-256 bug, which they generously offered refunds for).


The Sandforce SF-2281 is no longer "that bad". If the OP has the ADATA drive he mentioned, all he should do is make sure that he's got the latest firmware available from ADATA installed on the drive and he should be good to go.

Drops down a most Giant Hammer.....


...... Then walks off into the sunset.


I laughed. This was a really good post.

I think you explained it well. The VERTEX3 and AGILITY 3 (two best selling lines that both use the SANDFORCE Controller we are talking about) are not as bad as the non Sandforce controllers.


My first post in this thread suggested I believed much of the hate on Sandforce was due to it being trendy to hate on OCZ.

 

[jwilliams4200]

I'm not sure why people post things like, "I've had X number of SSDs and have not had any problems". That sort of anecdotal information is useless. Even if an SSD has a failure rate of 25%, simple probability dictates that there will be a significant fraction of people with one or more SSDs that do not fail. So we know a priori that there will be quite a few people with no failures. Having someone post that they have had no failures does not add any useful information to our knowledge.

To take a more common situation, even if an SSD has a failure rate of 5%, then a majority of people, even those owning several of them, will not see any failures. That does not change the fact that the failure rate is 5%, which is more than five times greater than that of the best SSDs you can choose instead.

If you want to know about quality or reliability, you need to look at a decent statistical study with valid methodology and a sufficient sample size to determine statistically significant differences between SSDs.

 

[Ao1]

I don't know the exact agreement between SF and vendors using their controllers, but I believe there was an onus on vendor companies to undertake validation testing. That would make sense as SF were at the time a small outfit.

OCZ (by far the largest seller of SF based drives) did not however bother with any serious attempt at validation, they simply wanted to get the product out to market as quickly as possible to maximise sales and profit.

Is that SF's fault or OCZ's fault?

Intel on the other hand spent a lot of time validating and found a number of bugs (nearly one year after OCZ first released SF-2281 based drives). I believe the bug fixes were exclusive to Intel for a period of time due to the validation process they went through to identify them. (Why should vendors that did not bother with validation benefit from Intel's validation work?)

Anyway besides the bug issues SF drives don't perform that well against other SSD's as the compression technology was geared to enterprise workloads and not desktop workloads. All that pain never gave any gain, outside of useless 0 fill compression benchmarks.

 

[blastingcap]

If you do not respect this website and call Anand "someone," then get the hell off of this website. Instead of casting aspersions and speculating that Anand's sources are worse than yours, you might want to find supporting evidence for your own views. Anand did not state things concretely but gave background information and there could have been some contractual jousting.

Btw, Anand is someone, all right. You, on the other hand, are a nobody.

lol, just because you quote someone and put it in bold does not make it true.

 

[VirtualLarry]

My biggest issue with Sandforce is the performance with incompressible data. Makes full disk encryption a huge burden.

This.

 

[JellyRoll]

You have no idea who I am, friend
I have much closer knowledge of this than you would believe, and it is not through marketing reps.
No need for me to prove it to you. Really I do not care how you feel about it, I can only give water to the horse.
Remeber, everything that you read is TRUE on the internet. Much like the articles that you digest, such as the Intel enterprise SSD fiasco. I find it entertaining that a follow up article on consistent performance showed the fault in the Intel marketing, yet it is still listed with a glowing review.
Open your mind.

 

[blastingcap]

You have no idea who I am, friend
I have much closer knowledge of this than you would believe, and it is not through marketing reps.
No need for me to prove it to you. Really I do not care how you feel about it, I can only give water to the horse.
Remeber, everything that you read is TRUE on the internet. Much like the articles that you digest, such as the Intel enterprise SSD fiasco. I find it entertaining that a follow up article on consistent performance showed the fault in the Intel marketing, yet it is still listed with a glowing review.
Open your mind.

Yeah, I totally trust anonymous people who claim to have knowledge and don't know how to spell "Remember."

Don't worry, I don't believe everything I read on the internet--including posts by anonymous forum members with almost no post history and no indicators of credibility.

Is it possible you are an insider? Sure. Is it likely? I haven't seen anything that would lead me to believe so. Smug and vague claims are all I see.

 

[jwilliams4200]

I don't have any inside information about Intel or Sandforce. But from my understanding about how those companies operate, I think it is very likely that Intel does not have any Intel employees modifying the Sandforce firmware. Intel may have spent many months testing their SSDs with the Sandforce firmware, but it seems likely they just reported issues to Sandforce engineers who then tried to fix them and then Intel did more testing on the modified firmware, found more issues, reported them, etc....

As for whether Intel has an exclusive license for some special firmware that Sandforce made for Intel, if Intel representatives say that is true, then I suppose it is true.

But that does not mean that Sandforce could not also modify their standard firmware in a similar (or even better) way to fix issues that Intel may have helped uncover. It is not like Intel would have oversight on Sandforce's standard firmware, looking over their shoulder and saying, "Oh! Oh! You cannot fix that bug there, we reported that to you and the fix is exclusive to our firmware, nyah nyah nyah!" :sneaky: I don't think so.

 

[JellyRoll]

Yeah, I totally trust anonymous people who claim to have knowledge and don't know how to spell "Remember."

Typical flamewar behavior. Call out an obvious error and attempt to use that to discredit the poster in an attempt to avoid actually providing some thoughtful, intelligent response. Quaint

You notice that Intel would not 'go on the record' with the information? This alone should tell you that there is a huge dose of doo-doo in the statement.
Also, as John notes, it is preposterous to imagine that a company would not share, or allow, another manufacturer to post bugfixes. This is also an obvious sign that this is not a true statement. Think about it. Please do not be a sheep and believe the hype.

Much like the Intel roundtable video that Anand attended. We have an attractive engineer wrinkling her nose and saying "its all about consistent sustainable performance" in an obvious ploy to be attractive and 'cute'.
Jesus.
These people are engineers who work with SSDs every single day, including enterprise class SSDs. They know full well that they are blowing smoke, and a tremendous amount of it, right up his wazoo. These people set around that table and sold a product, making every attempt to make this SSD the "gold standard in enterprise SSDs" even though they knew precisely well that it is not.
There is nothing remarkable about the performance of that SSD, other than the fact that it is remarkably slow compared to other offerings in the enterprise space.
Anand is not familiar with this type of testing and device, so he is not familiar of just how unremarkable the SSD is.
John4200 comes along and enlightens him a bit, then Anand posts an article that inadvertently shows the elephant in the room that there is nothing special about that SSD at all. Other than it is relatively cheap.
Why the big long story? because you can watch him take the bait...hook, line and sinker.
Exactly like the Intel 'off-the-record' (marketing speak for not entirely true) statements about the 520 firmware.
I like and respect Anand, but he probably needs to make a bigger effort to take a more global look at SSDs, instead of strictly reading from the marketing material.

 

[JellyRoll]

And finally as a bit of an interesting mention on this issue, I will quote Anand himself in his 'TRIM testing' of the 520.

Based on these results I'd be willing to bet that Intel doesn't have source code access to the SF-2281 firmware otherwise it would've worked on a solution to this corner case. Performance in this worst case scenario isn't terrible but the fact that it's irrecoverable even after a TRIM is what's most troubling.

http://www.anandtech.com/show/5508/...cherryville-brings-reliability-to-sandforce/7

In order to have their own version of the firmware, they would need to have access to the source.

Its also obvious that TRIM is not working in that test. Just like every other SF SSD for several firmware versions at the time.
Do you see a pattern here? Intel didn't notice that TRIM wasn't working for the love of christ? NO, it was just SF firmware the same as every other SF.

 

[bradley]

Sandforce is only two or three notches above JMicron's JMF601/602 controllers, which literally scared buyers away in droves from SSD.

What's wrong with Sandforce? Their controllers have been problems looking for solutions. I only want to buy from a semiconductor company positioning themselves in stability/integrity not trickery.

 

[blastingcap]

Since you are attempting to discredit Anand, you are the one who has to back that up.

You are saying Anand got hoodwinked and took issue with my citing to it as evidence that Intel might have had a reliability advantage vs other Sandforce clients due to their testing and working with SF/LSI. How long that advantage lasted, I don't know; but apparently there was some time in which at least one error didn't affect Intel SSDs that affected others' SSDs. You seem to take a more black and white approach to things but it appears to be a gray area where Intel could have stalled in its cooperation with SF/LSI bugfixing or something else; Anand wasn't specifically arguing that that was what happened, but you are jumping on one of his speculations as if that were all he said.

Your sloppiness with words (both yours and others) and your location ("Your Mother's House") and your argumentativeness set off alarm bells that you are likely not terribly mature, and thus unlikely to be a senior engineer with direct knowledge of what you speak.

If you have concrete info, feel free to share. But if I'm going to listen to hearsay, I'd rather hear it from Anand, who at least isn't anonymous and has a history of sharing his thoughts, e.g., his numerous articles detailing his conversations with Ryan Petersen.

Typical flamewar behavior. Call out an obvious error and attempt to use that to discredit the poster in an attempt to avoid actually providing some thoughtful, intelligent response. Quaint

You notice that Intel would not 'go on the record' with the information? This alone should tell you that there is a huge dose of doo-doo in the statement.
Also, as John notes, it is preposterous to imagine that a company would not share, or allow, another manufacturer to post bugfixes. This is also an obvious sign that this is not a true statement. Think about it. Please do not be a sheep and believe the hype.

Much like the Intel roundtable video that Anand attended. We have an attractive engineer wrinkling her nose and saying "its all about consistent sustainable performance" in an obvious ploy to be attractive and 'cute'.
Jesus.
These people are engineers who work with SSDs every single day, including enterprise class SSDs. They know full well that they are blowing smoke, and a tremendous amount of it, right up his wazoo. These people set around that table and sold a product, making every attempt to make this SSD the "gold standard in enterprise SSDs" even though they knew precisely well that it is not.
There is nothing remarkable about the performance of that SSD, other than the fact that it is remarkably slow compared to other offerings in the enterprise space.
Anand is not familiar with this type of testing and device, so he is not familiar of just how unremarkable the SSD is.
John4200 comes along and enlightens him a bit, then Anand posts an article that inadvertently shows the elephant in the room that there is nothing special about that SSD at all. Other than it is relatively cheap.
Why the big long story? because you can watch him take the bait...hook, line and sinker.
Exactly like the Intel 'off-the-record' (marketing speak for not entirely true) statements about the 520 firmware.
I like and respect Anand, but he probably needs to make a bigger effort to take a more global look at SSDs, instead of strictly reading from the marketing material.

 

[JellyRoll]

Your sloppiness with words (both yours and others) and your location ("Your Mother's House") and your argumentativeness set off alarm bells that you are likely not terribly mature, and thus unlikely to be a senior engineer with direct knowledge of what you speak.

I see that the insane hilarity of my presence at your mothers house is lost upon you. I find that terribly amusing.
The hottest product in silicon valley right now is a 24 year old engineer fresh out of college. I am far from that age unfortunately, but the point is that it wouldn't take a senior engineer to have inside knowledge. The interns have a far deeper knowledge of the inner works than most in this forum. (Not to discredit this forum, I do enjoy myself here).
Find an intelligent person that does not possess dogged determination (usually in being right, or 'argumentative' as you call it) and you will see one that lacks passion.
Many engineers do lack passion unfortunately, and that is why they do not bother to post in forums and have fun. They view SSDs as their job, and when at home they pursue their real passions. They have hobbies as well. It just doesn't happen to be the same as ours.
My language is colorful and fun. Laugh and have fun as I do and your life will be much more enjoyable. Shove that in the 'ol wazoo.

You do not find it curious that TRIM was obviously not working in Anands review, just as it was not working for every other SF SSD on the planet for several firmware revisions? Does the thought occur to you that Intel, in all of their infinite wisdom, would have fixed this issue if they were able?

 

[WhoBeDaPlaya]

Nothing particularly wrong. I have several 120GB Vertex 2s, 240GB Agility 3s and 240GB SanDisk Extremes running with nary a problem.

Then again, I do perform occasional secure erases on them.

 

[Mfusick]

I just don't understand how you can link an OCZ return data on a crappy failed line of SSD drives that don't even have Sandforce controllers in them as evidence Sandforce is unreliable.

That just makes no sense.

It only proves its Trendy to hate on Sandforce because its been trendy to hate on OCZ who rushed products to market with sandforce controllers a while back and experienced done growing pains.

Today I think that's all in the past. It's unreasonable to think the circumstances and results from 12 months ago would translate directly into a purchase Somone made today.

 

[Makaveli]

Funny. I have no touch screen. It's a Dell IPS panal that I do alot of photo editing on. Be warned - do not touch my screen. No fingerprints or dust allowed.
I've been on Win 8 for about a month now. With "Start-8". If you where to look at my desktop you could not tell the differance at all. OK.... The start orb is a win8 icon .... that would surely give it away.
What I am saying is that for $30.00 right now for Win8 Pro you would be crazy to not get it at this price.
Start8 will set you back another $4.99

I would surely be able to tell the difference in about 1 sec with no Aero on your desktop!