TweakTown article: "LSI SandForce 5 Series SSD Firmware - TRIM Lost and Found"

I’m a little surprised that nobody here seems to be commenting on the article at

http://www.tweaktown.com/articles/4870/lsi_sandforce_5_series_ssd_firmware_trim_lost_and_found_performance_investigated/index.html.  It talks about how the SanDisk Extreme 240GB and other SSDs using Biwin Elite 5.0.1 firmware is NOT passing TRIM commands and is suffering greatly for that omission.

I have not had time yet to investigate, but my SanDisk Extreme 120GB after a few months of use feels *well* slower to the point that I’d swear I had a mechanical drive again!  Now, I’ll admit I’ve also had to take my drive off the 6 GB/s SATA bus as well as switch back to IDE mode to reduce the number of BSODs I’m getting on Win7 SP1 x64.  I’m not sure yet if these things are related, though clearly changing the mode back from AHCI to IDE will affect performance.

So that would be why my writes have dropped off significantly in the MONTH I’ve had this drive?

Would that also be why in searching the forum for a SSD secure erase utility (Sandisk or third party) I’ve read about hostility regarding Firmware updates and such?

From the link:

“The second is absolutely in-freaking-conceivable; you were lied to in a cover up of the facts. We’ve talked about the TRIM issue for at least the last three months and just about everyone who reads TweakTown knows TRIM is not working on newer SandForce drives. The issue is discussed in various forums, on LinkedIn and other places where SSD reviewers openly discuss SSDs. As we said, this issue isn’t difficult to find, anyone who spends more than 20 minutes testing a drive with broken TRIM and compares the data to SandForce based drives with working TRIM knows that something is not right! You really can’t miss it; it’s a giant red blinking sign two feet away from you.”

At the moment, using AS SSD, I’m getting similar performance from the 120GB Sandisk extreme compaired to an Intel 330 60GB and the 330 has a full windows 7 install on it while the Sandisk only has about 15GB of games and other programs/utilities. That does not seem right to me lol ;).

One other thing… doesn’t TRIM contribute to the over all, long term health and longevity of the SSD? If so, what’s this "broken TRIM base code going to do to the life of the drive?

So it sounds like we should “hopefully” be getting a firmware update from Sandisk in a week or two…correct? Anyone hear anything?

Personally I wasn’t willing to wait on a “hope”… Especially with the information at hand. I mean really, no matter how you cut it there’s a serious issue here. Either Sandisk didn’t know about any TRIM problems and they seriously need to work on their in-house quality control or they DID know and didn’t delay the product and/or tell their customers…

Returned the drive and got Intel. At least when they mess up they come out and say it.

Everything those guys at Tweaktown said is true, the author is an expert in SSDs and believe me, you cannot put anything past him. A new firmware should be released soon, ( just a guess but they have to fix this).

SanDisk’s silence speaks volumes…

The issue is the LSI/SandForce 5.0 firmware has a bug that does not allow TRIM to work. SSDs that use that version of the FW will have that problem, but not all do. The SanDisk Ex is one that is affected, and happens to be the SSD that Tweak Town first noticed this problem.

Actually, LSI/SandForce have not officially announced that any problem with this FW exists. They have also not denied it either, AFAIK. I’ll assume that this bug exists for this discussion, but we should keep an open mind.

It’s curious that it seems only Tweak Town is openly discussing this, you can’t find anything about it on any of the well known PC hardware review web sites. IMO, that is another big question related to why no one seems to be commenting on this. Or are they simply embarrassed they missed it? Other scenarios easily come to mind…

SSD manufactures using this FW must wait for LSI/SandForce to provide a fix. Even if a manufacture had the FW code and could fix it, they likely could not distribute it due to the licensing agreement they have with LSI/SF. It seems that Tweak Town has an updated 5.0 FW that fixes this issue, but it has not yet been released as an update to the public.

If you’ve ever worked in software development, you know it takes time to test and verify fixes. It’s not a fast process, and that is as it should be. Considering this bug got out the door in the first place is an example of how difficult verification testing really is. My point is, don’t hold your breath waiting for a FW update.

As I’ve seen in the past, corporations feel it is better to say nothing about an issue for as long as possible, since once they do it is virtually impossible to take back, and they are then responsible. As an old saying goes, it is better to remain silent and appear to be an idiot, than to open one’s mouth and confirm it.

Actually, LSI/SandForce have not officially announced that any problem with this FW exists. They have also not denied it either, AFAIK. I’ll assume that this bug exists for this discussion, but we should keep an open mind.

Watching my performance drop the way it did, I can say with little doubt there is SOME issue. On top of that, if the reviewer at Tweektown had “a fleet” of Sandisk drives all showing the same results, that says something too. Lets also not forget the reason it was found… as was pointed out in the article, most reviewers do a secure erase in between benchmarks, since sandisk doesn’t supply that feature, it was tested without a secure erase in between tests. Honestly, they could have given themselves an easy out simply by saying back up your drive and run a secure erase to restore full performance… Not ideal, but better than the current situation.

It’s curious that it seems only Tweak Town is openly discussing this, you can’t find anything about it on any of the well known PC hardware review web sites. IMO, that is another big question related to why no one seems to be commenting on this. Or are they simply embarrassed they missed it? Other scenarios easily come to mind…

Or was Tweektown the only one to not use a third party secure erase between benchmarks? Might be worth checking into… but the point is still the same.

Considering this bug got out the door in the first place is an example of how difficult verification testing really is.

This only holds water if a person believes most of the major SSD companies using sandforce ALL missed something so easily caught that a 31yo guy with no formal IT background (Me) could stumble across the end result of… Sorry, but I’m not biting on that one… And even if they did “opps, just missed it”, why on earth are they not checking the product they are buying?? Sure, Sandforce messed up… but ultimately it’s SANDISK that is responsible for the product they release and sell to the public not Sandforce. We bought SSDs from Sandisk not Sandforce, they need to pay attention to what their vendors are giving them.

I worked for a company that made downhole tooling for another company. A lot of the parts had to go out to another company for heat treating or surface treatments since we didn’t do that in-house. Ultimately if the heat treat was done improperly, WE were responsible to our customer, not the heat treat company. Our customer didn’t care where the surface/heat treatments were done, they just wanted a reliable, quality part that met their requirements. There was one point where there was several batches of parts (several thousand) that a large percentage of didn’t meet the customers requirements. We had to seriously consider either switching vendors or doing the treatments in house… perhaps Sandisk (and others) need to consider the same.

As I’ve seen in the past, corporations feel it is better to say nothing about an issue for as long as possible, since once they do it is virtually impossible to take back, and they are then responsible. As an old saying goes, it is better to remain silent and appear to be an idiot, than to open one’s mouth and confirm it.

This is simply bad business… plain and simple. I for one won’t stand for it. If that means I have to limit my computer purchases to companies that make prompt updates available and/or publicly admit when there’s a mistake, so be it.

Totally agree :smiley:

I’d also like to point out that while OCZ has had nothing but black eye after black eye with one issue after another… they apear to have started moving away from Sandforce with their new controller (Indilinx Everest 2). Admittedly I haven’t looked into these drives or that controller much, but the point I’m making is if there’s issues with the main vendor maybe Sandisk should do the same.

Honestly, I wrote off OCZ due to it not allowing windows to boot at least a few times a week. That said, if they’ve switched to a new controller, it shows me they are at least TRYING… and that makes me willing to try them out again at some point.

Indilinx 2 is Marvell controller with Indilinx/OCZ firmware, nothing related with Sanforce, it works awesome so far and it’s real fast.

@memnoch31 wrote:

Actually, LSI/SandForce have not officially announced that any problem with this FW exists. They have also not denied it either, AFAIK. I’ll assume that this bug exists for this discussion, but we should keep an open mind.

 

Watching my performance drop the way it did, I can say with little doubt there is SOME issue. On top of that, if the reviewer at Tweektown had “a fleet” of Sandisk drives all showing the same results, that says something too. Lets also not forget the reason it was found… as was pointed out in the article, most reviewers do a secure erase in between benchmarks, since sandisk doesn’t supply that feature, it was tested without a secure erase in between tests. Honestly, they could have given themselves an easy out simply by saying back up your drive and run a secure erase to restore full performance… Not ideal, but better than the current situation.

 

It’s curious that it seems only Tweak Town is openly discussing this, you can’t find anything about it on any of the well known PC hardware review web sites. IMO, that is another big question related to why no one seems to be commenting on this. Or are they simply embarrassed they missed it? Other scenarios easily come to mind…

 

Or was Tweektown the only one to not use a third party secure erase between benchmarks? Might be worth checking into… but the point is still the same.

 

Considering this bug got out the door in the first place is an example of how difficult verification testing really is.

 

This only holds water if a person believes most of the major SSD companies using sandforce ALL missed something so easily caught that a 31yo guy with no formal IT background (Me) could stumble across the end result of… Sorry, but I’m not biting on that one… And even if they did “opps, just missed it”, why on earth are they not checking the product they are buying?? Sure, Sandforce messed up… but ultimately it’s SANDISK that is responsible for the product they release and sell to the public not Sandforce. We bought SSDs from Sandisk not Sandforce, they need to pay attention to what their vendors are giving them.

 

I worked for a company that made downhole tooling for another company. A lot of the parts had to go out to another company for heat treating or surface treatments since we didn’t do that in-house. Ultimately if the heat treat was done improperly, WE were responsible to our customer, not the heat treat company. Our customer didn’t care where the surface/heat treatments were done, they just wanted a reliable, quality part that met their requirements. There was one point where there was several batches of parts (several thousand) that a large percentage of didn’t meet the customers requirements. We had to seriously consider either switching vendors or doing the treatments in house… perhaps Sandisk (and others) need to consider the same.

 

As I’ve seen in the past, corporations feel it is better to say nothing about an issue for as long as possible, since once they do it is virtually impossible to take back, and they are then responsible. As an old saying goes, it is better to remain silent and appear to be an idiot, than to open one’s mouth and confirm it.

 

This is simply bad business… plain and simple. I for one won’t stand for it. If that means I have to limit my computer purchases to companies that make prompt updates available and/or publicly admit when there’s a mistake, so be it.

 

I certainly agree with all of this, I was just trying to provide some insight into this situation. I agree that this problem does exist, although it is also true that any SSD will not perform at the same level that it did when new, after it has been used for a while.

It is incredible (and ridiculous) that SandForce could miss this TRIM bug, as well as all the SSD manufactures that are using this version of the firmware. I’d really like to know what actually happened, such as a bad version was distributed to the manufactures by mistake or (_ purely speculation!! _)  knowingly sabotaged by a SandForce employee. You would think given the bad reputation that SandForce had with the earlier versions of the firmware for this controller, that they would be extra careful now. It’s all just shocking.

This only holds water if a person believes most of the major SSD companies using sandforce ALL missed something so easily caught that a 31yo guy with no formal IT background (Me) could stumble across the end result of… Sorry, but I’m not biting on that one… And even if they did “opps, just missed it”, why on earth are they not checking the product they are buying??

 

Not to argue or disagree with you, but what do you think happened? Such as, they knew about it but said nothing and hoped no one would notice until they had a fix? It does make you wonder how much testing companies actually do, or do they assume that the product they purchased is fine and does not need much testing.

So right off I just want to say, my post comes across way more confrontational than intended lol ;). So fair enough as far as pointing things out and perspective. Nothing personal, nothing mean, ect. I actually do appreciate the comments and input :). 'Tis the joy of online forums, so much of a conversation can be lost over text!

So far as all SSDs having performance drop over time, very fair comment. That said personally I’d expect a little more than a month before seeing such a major drop. A year? 2 years? I can’t remember off the top of my head but for example if the warrenty is 3 years (I think it’s 4?) personally I’d expect it to keep up a reasonable level at least that long, depending on usage of course. Plus this is also where over provisioning comes in. For example the Intel 330 180GB SSD I replaced the sandisk drive with actually has 192GB of NAND in it. As blocks are retired, it uses some of these extra NAND blocks to maintain performance. Since SanDisk is using 120GB, 240GB and 480GB my guess is they are also using provisioning otherwise they’d *likely* be 128, 256 and 512 drives (hopefully with RAISE functionality too lol).

 I’d really like to know what actually happened, such as a bad version was distributed to the manufactures by mistake or (purely speculation!!)  knowingly sabotaged by a SandForce employee.

I hadn’t even considered this possibility! for sure, completely speculation, but an interesting scenario non-the-less! Continuing with that line of thinking, it’s a complete craps shoot on whether or not Sandforce would come out and admit to that… The public’s response could go either way.

Not to argue or disagree with you, but what do you think happened? Such as, they knew about it but said nothing and hoped no one would notice until they had a fix? It does make you wonder how much testing companies actually do, or do they assume that the product they purchased is fine and does not need much testing.

In all honesty, I have no idea. I have my theory’s but I’ll be the first to admit that’s all they are. That said, there has been a growing trend where these large companies have been using their customers as guinea pigs. And don’t get me wrong, I get why. Really, Apple has the least amount of work to do… they have set systems with limited hardware options (for the most part, but that has been changing since they switched to intel systems… just slowly) and set drivers. There are only so many hardware configurations that are even possible. Where as with “normal” PCs just about anyone and their dog can slap a system together and hope to hell everything is compatible (this is why I refuse to pick on Microsoft as trendy as that might be in some circles).

1) So did Sandforce not find the issue… I doubt it, but that is possible. It would show they’re not testing (or not testing enough) though.

2) If they did find the TRIM bug, they might have decided the cost of a recall or producing new firmware was more than the cost of expected returned drives. Lets face it, the majority of people would never find the issue because they just don’t care to look for it. This type of person is perfectly happy knowing they have a SSD and “those are crazy fast” and only taking it that far ;).

3) Maybe the drive manufacturers did know about the issue and decided to wait and see how many people were effected and in what way before working on some type of a patch? Or to pass the information on to Sandforce so they can work on it.

4) Maybe some of the manufacturers are trying to get access to the Sandforce Base Code to write their own drivers from the ground up?

Another Sandisk Extreme SSD 240GB review up on the web, another mention of poor Trim. Disappointing.

“This specific model has one downside, though. It suffers a heavy performance hit as it transitions from its out-of-box to steady state. This downside can be eliminated by firmware updates although SanDisk doesn’t seem too eager to do that. So, the SanDisk Extreme is an attractive product potentially, yet we wouldn’t recommend buying it right now. It’s wiser to wait for similar products from other brands.”

http://www.xbitlabs.com/articles/storage/display/sandisk-extreme-ssd-240gb_8.html

The lack of updates on the TRIM problem here is disapointing as well. I wonder if Sandisk are not aware of the problem or just don’t care…

Not even a full day after my previous post here I found another aricle that is relevant to the TRIM problem.

Tweaktown posted their review of Corsair Neutron series of SSDs. The interesting part that for comparison they include Sandisk Extreme SSD 240GB running beta firmware 5.0.3!  So Sandisk are quitely working on a solution to TRIM. Also the performance looks a bit better with 5.0.3 than with R201 when comparing fresh out-of-the-box performance. Hopefully we won’t have to wait for too long…

Check the performance numbers here:

http://www.tweaktown.com/articles/4897/corsair_neutron_series_solid_state_drives_review_lamd_or_wolf_in_a_7mm_package/index6.html

For those of us whose drives now feel like a creaky old mechanical drive, it’s already been TOO LONG!  And STILL SanDisk is silent.  Never had problems like this with my 3 year-old OCZ Agility!!! NEVER AGAIN, SanDisk!

http://www.tweaktown.com/reviews/4895/sandisk_extreme_240gb_fixed_trim_solid_state_drive_review/index.html

looks like a fix should be coming in about a month. 

Thanks for the good news, Dr Lucky!  I can hardly wait!

The 240GB Sandisk Extreme was my first SSD. I’ll probably just go Intel next time as well. (I got this drive on a good sale, else I would have went Intel in first place)

Hello,

I joined to catch when the new firmware is out.

Extreme 480GB here :slight_smile: