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

SSD (Solid State Drives) All Other SanDisk SSD
81

It is quite funny that some people are still not happy.

Quite honestly, none of the SSD makers (or hard drive makers) would live up to the expectations some of you have. Sandisk could have been better with communication, but their implementations for ssd updates are the best/simplest I’ve seen.

Given they are also suppling updates specificly to resolve Mac Issues (poor negotiation between nvidia SATA and sandforce controller), it is clear that sandisk are not leaving their users in the cold.

82

@nightcap wrote:

Understatement of the year. Corporate neglect on a grand scale. No more, no less. Totally unacceptable.

 

I’ve downloaded, burned ISO to CDR and updated two Extreme SSDs to R211 firmware. So far, so good. Will monitor performance.

 

In the meantime, various doubts arise:

  1. After 6 months usage with broken TRIM how is overall drive longevity (NAND cycles) affected? Were we beta tester guinea pigs?
  1. Is firmware R211 guaranteed to provide a permanent remedy that does not negatively affect other performance parameters of the SSD? (Another Sandisk Extreme user elsewhere reports read rates are down and write rates are up following his initial firmware update.)
  1. Will any Sandisk executive finally come forward and do the honourable thing by providing a full explanation of events that led to broken TRIM, and how Sandisk went about fixing it? If not, are end-users to learn facts third-hand from distant cyber-forums? What then, is the purpose of Sandisk hosting this forum?

  1. No, I’ve run an endurance test on one of these drive based on the on old firmware, and got close to 1PiB total writes before drive death, which is a bit over twice the specified life for the NAND, and fairly typical for a sandforce drive.

  2. There are no guarantees in life. In the case of the other user, the reduced read performance of about 3% is well within error  margins for these kinds of benchmarks.

  3. I wouldn’t hold my breath … every other company has been relatively hush hush on this whole TRIM buisness. What probably happened was that they implemented a new algorithm for TRIM/garbage collection for the 5 series firmware and it simply didn’t work well (there is evidence that TRIM did work, just plenty that it wasn’t working well).

* 5.0.3 reverted to the older algorithm (found in series 3 firmware)

* 5.0.4 was implemented with fixes that allowed WHQL certification to pass.

* Sandisk, being the huge slow moving company it is, applied updated to disable 6gbit SATA for macs, then took several weeks intergrating and testing it on its update system. Too long you say … maybe, but you were not the engineers tasked with making sure this random update didn’t brick half the drives it came in contact with. 

83

@canthearu wrote:

 

…Sandisk could have been better with communication, but their implementations for ssd updates are the best/simplest I’ve seen. …Given they are also suppling updates specificly to resolve Mac Issues (poor negotiation between nvidia SATA and sandforce controller), it is clear that sandisk are not leaving their users in the cold.

Agreed. Both seem fair comments.

canthearu continued to write:

  1. No, I’ve run an endurance test on one of these drive based on the on old firmware, and got close to 1PiB total writes before drive death, which is a bit over twice the specified life for the NAND, and fairly typical for a sandforce drive.

So you’ll guarantee that no damage was done? Or that there is little relative damage? Or the damage was worth rushing an immature product to market?

  1. There are no guarantees in life. In the case of the other user, the reduced read performance of about 3% is well within error margins for these kinds of benchmarks.

Incomplete and incorrect. Further data indicate an approx. 14% decrease in 512kb reads. Statistically significant?

  1. I wouldn’t hold my breath … every other company has been relatively hush hush on this whole TRIM buisness. What probably happened was that they implemented a new algorithm for TRIM/garbage collection for the 5 series firmware and it simply didn’t work well (there is evidence that TRIM did work, just plenty that it wasn’t working well).
    * 5.0.3 reverted to the older algorithm (found in series 3 firmware)
    * 5.0.4 was implemented with fixes that allowed WHQL certification to pass.
    * Sandisk, being the huge slow moving company it is, applied updated to disable 6gbit SATA for macs, then took several weeks intergrating and testing it on its update system. Too long you say … maybe, but you were not the engineers tasked with making sure this random update didn’t brick half the drives it came in contact with.

Non-brick is good. And many here have witnessed DOA or lame-duck SSDs at close quarters. Most failures involved Sandforce controllers, if I’m not mistaken. Do you notice any pattern emerging?

84

You all keep thinking TRIM was not working, but it was, but the controller was too slow to clean the trimmed cells before you do another benchmark. If you have let idle time, you would have gotten the same performance.

Stop telling the product was immature. You can say that for OCZ because they are first to rush a new product to market, but not Sandisk.

85

“You all keep thinking TRIM was not working, but it was, but the controller was too slow to clean the trimmed cells before you do another benchmark. If you have let idle time, you would have gotten the same performance.”

Do you have proof of that? You might want to read the white papers on SandForce’s site that talk about garbage collection and write amplification. 

86

I can’t proove it myself, I can only report what this guy has found:

http://forums.sandisk.com/t5/SanDisk-Extreme-SSD/TRIM-EFFICIENCY-OF-240GB-SANDISK-EXTREME-WHEN-USED-AS-A/td-p/283036

87

I’m NOT a routine benchmarker.  I did a set of benchmarks when I got the drive, and that was as far as I planned to go.  BUT, when after a few months the drive FELT NOTABLY SLOWER, I did another set of benchmarks to see how things were going.  The results were roughly half of what they were originally, so far as write speeds.  Read speeds were only slightly slower compared to the original.

Call it what you will, but I NOTICED THE DIFFERENCE and went searching for answers.  Only then did I find the TweakTown article about broken TRIM.

SanDisk’s UTTER SILENCE has convinced me NEVER to buy or recommend a SanDisk product.

88

Sandisk Company Q3 2012

Sandisk appears to be generating revenue, yet trivial income. Company can’t afford to communicate with customers?

According to recent data released by CEO Sanjay Mehrotra and CFO Judy Bruner, quarterly revenue of approx $1.5 billion may not provide sufficient means for one Sandisk moderator spending (let’s say) 3 or 4 hours briefing end-users on TRIM issues in this forum. Seems Sandisk’s enterprise clients warrant full service provision but “small fry” end-users don’t?