SF2281 TRIM support firmware by Sandisk that actually works? When?

Internal SSD All Other Internal SSD
41

My G-Skill Falcon was nothing but trouble,replaced 3 times.

I had a Crucial C300 that was hosed by a firmware update.

A few Intels were bummers also.

I have absolutely no issues with all the OCZ SSDs I own/owned in fact a Vertex 2(34nm) has been performing flawlessly for 28 months.

My 2 Vetex 3 MIs have also been working without issues for over a year.

My 3 240GB SanDisk Extremes are all working well in a Raid array and as a single drive but I too would like a good 5.0.4 based firmware.I donā€™t want a half assed job so they need to take the time and get it right.

Many issues with SSDs can be laid at the feet of the users and their bad systems.

42

@nightcap wrote:

Sandisk is remaining silent. This is felt by many end-users here as unacceptable. Unless corporate policy changes Sandisk is likely to lose future SSD customers. What for? What is the point?

To do extensive testing to insure it is right before they release it. They would p|ss off many more customers if it were buggy and they released it prematurely just to satisfy a few.

43

Tapeworm,

I do not get your logic that they are extensively testing???

They were a month later than most other Drive Makers releasing current FirmwareĀ  with 5.0.1 that is totally defective.

What make you think they do any extensive testing if they did not realize that TRIM was broken? GET REAL!!

This is a Company that just does not give a ā– ā– ā– ā–  about the Customers who use their devices and is just proving that they do notĀ haveĀ the resources to perform the needed validation in an acceptable time.

Quit making excuses for a Company that is obviously a collection of Slackers!!!

44

Letā€™s stick with facts.

#1) The problem exists and was identified two months ago (August 1, 2012).
#2) Sandisk does not publicly acknowledge any problem, there is no official communication to indicate when, if ever this will be fixed.

If Sandisk had done extensive testing to start with, they would have known TRIM was broken.  So, either they missed it and didnā€™t test extensively to begin with, or they knew it was there and released a bug hoping nobody would notice.  Either answer costs Sandisk credibility.

At this point I would like Sandisk to publicly admit there is a problem, let us know if they are working on a fix for it, and if they are working on it, tell us their best estimate when the fix will be available.  Doing nothing shows great disrespect to all retail customers. 

A bad reputation among IT professionals and hobbyists can take years to fix, if itā€™s even possible.  It costs much more for a company to gain a new customer than keep an existing one.  At this point some of Sandiskā€™s former SSD customers are not only taking their business elsewhere, theyā€™re convincing everyone else they know in the market to do the same.

@tapeworm wrote:

To do extensive testing to insure it is right before they release it. They would p|ss off many more customers if it were buggy and they released it prematurely just to satisfy a few.

45

If they release the fix as v5.0.5 today that is not premature.Ā  That is late.Ā  And if there are ANY bugs, issues or performance problems with a release today that clearly demonstrates incompetence at SanDisk.Ā  Due care has long passed its use by date.

46

BeerPhart wrote:

At this point I would like Sandisk to publicly admit there is a problem, let us know if they are working on a fix for it, and if they are working on it, tell us their best estimate when the fix will be available.Ā  Doing nothing shows great disrespect to all retail customers.Ā 

Agreed. In this day and age it does not seem unreasonable to expect a large, well-established company to communicate basic technical information to its customer base. For that company not to communicate at all (given various rumours and inferences both inside this forum and elsewhere) appears inept, to say the least. How can most other LSI-SF-controller supplier companies calmly do so, and Sandisk cannot, or will not? What is it that Sandisk is fearful of communicating to us?

47

BeerPhart wrote:

At this point I would like Sandisk to publicly admit there is a problem, let us know if they are working on a fix for it, and if they are working on it, tell us their best estimate when the fix will be available.Ā  Doing nothing shows great disrespect to all retail customers.Ā 

Agreed. In this day and age it does not seem unreasonable to expect a large, well-established company to communicate basic technical information to its customer base. For that company not to communicate at all (given various rumors and inferences both inside this forum and elsewhere) appears inept, to say the least. How can most other LSI-SF-controller supplier companies camly do so, and Sandisk cannot, or will not? What is it that Sandisk is fearful of communicating to us?

Gentlemen,

Here are the obvious Facts:

Sandisk does not communicate with RetailĀ Customers and itā€™s seems they actually have disdain for them by their rudeĀ  & arrogant silence on this issue.

If theyĀ actually test the Firmware how could they miss the TRIM being brokenĀ the 5.0.1.

If this was an Enterprise Product the SKU would be pulled from Distribution @ Sandiskā€™s cost and also absorb any Field Cost for updating, because Enterprise User communicate amongst each other @ Shows and Conferences and defective product news travels fast though the Community.

We are the bottom of the food chain in Sandiskā€™s view and are being treated as such!

Sandis does not feel compelled to deliver to the Consumer a fully functional product since there is little consequence to suffer from Individual Users.

48

Basic questions for Sandisk management:

  1. What is the purpose of hosting user forums when Sandisk does not provide helpful, timely technical information?

  2. What is the purpose of witholding timely firmware updates?

  3. What does Sandisk hope to achieve by being uncommunicative and unhelpful?

  4. Does Sandisk envisage holding or building SSD market share? If so does Sandisk intend offering bonus $100 bills with every SSD in order to convince reticent SSD buyers?

  5. Are performance bonuses paid to Sandisk managers for remaining secretive?

  6. Are Sandisk legal counsel always present at (firmware validation) team meetings?

49
  1. Are Sandisk legal counsel always present at (firmware validation) team meetings?

I read a post in another Forum where one of the posters was contacted by Class Action Law Firm.

@ this point they probably need the Lawyers.

There are also links to the Local BBB & FTC Consumer Protection Agency to file Complaints.

The amazing thing is that Sandisk continues their silence & thus aggravating the situation even more.

Who ever runs this Company need an education in Customer Relations or they should find a different job.

Perhaps a few post concerning all these defective products being sold by Sandisk should hit some NASDAQ Bulletin boards.

Nothing works better than telling the truth.

Maybe that may wake them up?

We have a saying @ work that if you punish a Companyā€™s Stock Price you are sending a wake up call to Management.

Could be a good time to short this Company because nothing good is going to come out of this situation as it becomes more public

and Investors see the gross incompetence being demonstrated by Sandisk Management.

Consumer Complaints & Class Action Law Suites are never good for a Companyā€™s Stock Price!!!

50

Has anybody else thought that SanDiak may have deemed the SF2281 end-of-life.Ā  They may just have a new controller from another source that they are working on.Ā  This would explain their long term no comment and lack of action.Ā  Just a thought.

51

Thereā€™s obviously something wrong. The continuing silence from SanDisk is baffling. Maybe the problem is potential legal trouble between SanDisk and Sandforce over the responsibility for this debacle. The fallout of that could be that SanDisk and Sandforce canā€™t agree on how to move forward together to fix this, so SanDisk has written off the product and is working on a new design with a new controller. Have they written off the customers too? Possibly - or maybe theyā€™re thinking of trying to rescue their brand name by offering a trade-in program.

52

I think a bigger problem is growing for Sandisk.Ā  Discussion of broken TRIM is winding down here; not because the issue has been resolved, people are giving up and moving on.

This forum ought to be Sandiskā€™s canary in a coal mine.Ā  When the canary dies, Sandiskā€™s SSD business will soon follow.

Granted if you look at Sandiskā€™s stock ticker on finance.google.com their stock isnā€™t really hurting at all, probably because they have a larger portfolio to offer than solid state drives.

http://www.google.com/finance?q=NASDAQ%3ASNDK

Now letā€™s compare them to OCZ

http://www.google.com/finance?q=NASDAQ%3AOCZ

OCZ specialize in solid state drives and power supplies and have been fighting uphill against a very bad reputation for their poor/lack of support track record.Ā  I would hazard a guess to say OCZ and Sandiskā€™s record as far as SSDā€™s are equal these days in light of Sandiskā€™s lack of customer support.

With OCZ having 708 employees and Sandisk having 3,939 employees, I think taking in consideration Sandiskā€™s economies of scale they could have as many as 400 employees working on SSD drives internally.Ā  If Sandisk completely fails with their SSD strategy, this could cost them up to 10% of their overall company in labour.Ā  Letā€™s not forget the bleedover effect a bad reputation will have in their other divisions, and the loss of morale other departments will feel when people in the SSD division get reassigned or laid off.

Sandisk is fortunate in being quite vertically integrated, but they are not so vertically integrated that they could make their own SSD controller.Ā  Maybe they have internally learned from this debacle and are concentrating on making their own controller so they are not at the mercy of an outsourced contractor again, but itā€™s really too late at this point.Ā  Unless quick action is taken by their SSD division to appease customers, their SSD line is going to suffer the death of a thousand cuts the same as OCZ, and their poor reputation will ensure their SSDs can only compete in the value market and not the enterprise business market where the real profit is.Ā  I would say companies like Intel and Samsung are very well poised to become the market leaders in enterprise SSD, and by default the brand of choice of enthusiasts who have expectations in reliability, support and quality.

Maybe Iā€™m wrong here, I hope I am.Ā  I originally chose Sandiskā€™s SSD over their competitors because I considered their vertical integration a good reason to trust their products, and had good experiences with their USB flash drives and solid state MP3 player offerings.Ā  As a result of Sandiskā€™s lack of communication on this, I doubt I am going to buy any of their products in the future.Ā  It will most certainlyĀ  influence who I consider giving invitation to tender for any IT capital expenditure projects.

53

Out of the great many companies that use the Sandforce 2281/2282 controller only Muskin has released new firmware based on reference firmware 5.0.4

OCZ didnā€™t use 5.01 in their 2281 lineup so their drives are not effected by the trim issue but they said they will release new firmware possibly by this weekend.

BTW,all makers are in all probability working on drives with unreleased controllers like the Barefoot 3 and Sandforce 3rd generation or even the released Marvel 187.

Probably itā€™s not good to be 1st or last with firmware releases based on reference Sandforce firmware.

Again,whatā€™s the rush folks?

A rushed cure may be far worse than the disease.

54

SmogHog,

Do some research.

Transcend released 5.0.4 over three weeks ago.

Ā I flashed two Drives with it and they had dramatic increase performance.

Here is how you test TRIMs performance.

The test:

Fresh format/secure erase/ Install Win 7
Run CrystalDiskMark ā€“ you should get 550MB/s | 520MB/s read/write
Create Large Dummy Data Files (note that it has to be used, instead of just allocated)

  • see middle of http://www.windows-commandline.coā€¦-file.html
    Re-Run CrystalDiskMark, benchmark should be lower
    Delete Dummy Data Files
    Re-Run CrystalDiskMark, benchmarks should be higher (close to 550/520), but supposedly since TRIM isnā€™t working correctly, it will be much lower. This was the case before we flashed the Drive.

After the Flash the read/write Speed came back to near Maximum immediately.

Try this on your Sandisk Drive you will find the dismal low read /writes.

You need to leave the Sandisk Drive idle for hours(it took about 48 hours on our Test)Ā before the read/writes get anywhere near normal.

This is the Garbage collection trying to subsitute for TRIM

Cause broken TRIM!!!

55

OK,I bow down before you all knowing one whoā€™s the research king with a bad attatude toward other members not as enlightened as he isā€¦

The Transcend site says September 21 for release of 504.

Now thereā€™s 2 "Ā I" know of.

You could have just said I forgot one ā€œyouā€ know of but you needed to vent some frustration.

BTW guys,have you taken advantage of the SSD giveaway contests on thessdreview forum.

Just join and youā€™re eligable to win.

None of the Facebook and Twitter ā– ā– ā– ā–  requirements like you see on most siteā€™s giveaways.Alls you need is to be a member.

56

Consider the most incredible thing about this situation: LSI/SandForce, the designers and creators of the 2281 controller, and its essential firmware, released the 5.0 firmware to its customers (SanDisk, Corsair, etc) with malfunctioning TRIM. No one knows more about this controller and firmware than they do, and (apparently) this bug got past them! If SandForce could not find it, what chance does a company that knows far less about it have of finding it?

OTOH, we then have a reviewer, with a few programs and data files, that finds this problem quite simply. Most of us reading this forum, using the same programs, could also do that. The leap from this relatively simplistic method of finding this issue, back to SandForce and all their knowledge, is huge. IMO, something is missing in the middle, something we are not aware of that would explain at least some of the big question, how did this happen?

It is also strange how little about this has been reported in the PC hardware world. Normally, a news story is published by one source, and all the others immediately pick it up and do the same, they canā€™t be left out or miss it, theyā€™ll look bad. We donā€™t see that here. Why not?

Regarding the release of new firmware, if the source (SandForce) releases it to their customers, and then one of those customers releases it a week or two later, they barely tested it, if at all. All they tested was their firmware update program(s). A month delay is short too. IMO, three months minimum for a decent job. When Intel validated the SandForce firmware when they decided to use the 2281 controller, it took a year to do so. I also bet they had access to code no one else has, and modified it, since many reviewers consider the Intel firmware to be unique. Of course, they did not sell products using it before their validation.

What happens if in testing an issue is found? Verify the issue, fix it if it exists, and test again. Oops, starting over again. Can you put that on a schedule? Can you publish a release date under those circumstances?

57

So following the discussion above.Ā  Intel supposedly found and fixed the problem with the SandForce version 5 firmware sometime ago.Ā  I would assume SandForce had/has the ability to review and understand the changes made by Intel, and consequently know exactly how to fix this problem.Ā  So why the holdup in SanDisk getting this info from SandForce.Ā  Doesnā€™t make sense to me.

58

SanDisk isnā€™t actually fixing the firmware. They buy the controller chip and the core firmware from Sandforce. Their role is limited to testing (which they obviously didnā€™t do very well) and minor customizing. Sandforce remains behind the scenes as a chip supplier to SSD manufacturers, and they arenā€™t talking about this problem. The story is supposedly that they have already fixed this problem in their firmware, and some other SSD manufacturers have already deployed it in their firmware updates to customers, but SanDisk hasnā€™t because they are waiting on some other problems to be fixed. The problem is, that story doesnā€™t make sense. Why is SanDisk remaining completely silent and taking no action while their reputation is dragged through the mud and their brand name is permanently damaged?

59

One is prompted to ask, has there been a colossal falling-out between LSI-Sandforce and Sandisk over this matter?

If so, what hope have we end-users got of ever receiving effective, timely product support well into the future?

If there remain additional technical issues in SF2281 firmware (of which we end-users can only guess) how serious could they be, given Sandiskā€™s continuing secrecy?

If LSI-Sandforce and Intel between them could not perfect TRIM over a long, collaborative development and testing period, what does that say about SSD technology and IT storage industry overall?

60

@rifter wrote:

So following the discussion above.  Intel supposedly found and fixed the problem with the SandForce version 5 firmware sometime ago.  I would assume SandForce had/has the ability to review and understand the changes made by Intel, and consequently know exactly how to fix this problem.  So why the holdup in SanDisk getting this info from SandForce.  Doesnā€™t make sense to me.

Intel did NOT use the version 5 firmware, that was released after the Intel SSDs came to market. Intel uses the version 3 firmware. Version 5 firmware is, according to what Iā€™ve read, a rewrite of the firmware. Also, the newer versions of that firmware, 5.0.3, is using code from version 3 firmware. Version 5.0 is supposedly faster than 3.0, and 5.0.3 is back to 3.0 performance. Firmware 5.0.4 may be a fixed version of pure 5.0 firmware, which is what SanDisk may be wanting to use, thus the further delay. The TweakTown articles on this topic are the source of this information.