TR Forums

AnandTech wrote:

For those who are not aware of the issue, there is a bug in the 840 EVO that causes the read performance of old blocks of data to drop dramatically like the HD Tach graph below illustrates. The odd part is that the bug only seems to affect LBAs that have old data (>1 month) associated with them because freshly written data will read at full speed, which also explains why the issue was not discovered until now.

Musafir_86 wrote:

-Excuse me, but aren't ALL SSD (and NAND flash storage) have wear-leveling algorithm that supposed to move around (shuffle) the data internally? So those old data never get shuffled at all?

Musafir_86 wrote:

-What about data retention? What'll happen to the data in a drive that has been unplugged for months/years?

Ryu Connor wrote:

Benchmarks tend to be run on fresh drives, not on drives that have accumulated data over time and had that data age. Even TR's retention tests only go a week, which falls short of running into this bug.

Possible causes:

By this point you may be asking what the heck can be causing this? Here's my hypothesis:

1. Over time, flash cell voltages slightly drift.
2. Flash circuitry is designed to compensate for this by varying the read voltage thresholds and using varying levels of error detection and correction mechanisms.
3. Some unforeseen coordination issue between the flash and the controller, in the face of #2 above, is resulting in slower than normal read speeds for flash that has been storing data, undisturbed, for weeks or months.

Now I know there are many folks from many threads, that having read the above, are saying 'no way, it has to be worse than that!'. No, it doesn't. Why? Because out of all of the dozens or hundreds of folks running these tools and tests, not one has reported a single unreadable sector. Think about it, dozens of people are intentionally reading all data from these drives, which is way more than a user would be doing in typical use, and despite the slowdowns - no errors. If the flash cells or stored voltages were actually degrading so badly over time, the oldest stored data would likely be unreadable. Instead we see a steady speed taper down to a seemingly fixed (~50-100MB/sec, varying with drive capacity / die count) read speed. That indicates to me that the issue is one that can be corrected - maybe even in a manner that would snap speeds back to near full simply by updating the firmware. This is certainly possible, as we've seen it happen before.

Flying Fox wrote:

I mean many people have bought these drives more than 2 months ago. They probably have rebooted their systems or launch the same apps/games too. How come they did not report the problem? That's why I suspect a particular batch and/or a specific firmware version.

After the initial posting of this article, further details have come to light. BrainSplatter got back to me with some data on temperature effecting the slowing effect on old file read speeds. The same test was run first at 32C (90f) and then at 52C (126F), here were his results:

Newest: 510 -> 507 MB/s = - 0.8%
2 weeks old: 477 -> 427 MB/s = - 10.5%
3 weeks old: 278 -> 211 MB/s = - 24%
4 weeks old: 289 -> 177 MB/s = - 39%

The temperature difference he used was a bit on the extreme side, but it did certainly show a speed delta that amplified with file age. This supports the theory that the EVO's controller is busy crunching ECC data, and as a result is running up against its Dynamic Thermal Guard Protection, which in this case is proportionally clocking the MCX controller / CPU slower in hotter environments.

We can draw a few conclusions here:

* The flash itself is not degrading under higher temperatures, as the speed bounces back once the drive is cool.
* The controller was likely not meant to run at this sort of continuous ECC load, which further supports that a firmware fix might correct the issue (programmatically compensating for cell drift over time can mitigate the need for heavy ECC calculations and would theoretically restore speeds to near full for old files).

danny e. wrote:

210GB of used space.

Edit:
I do have my settings on "maxium reliability" whatever that does. So, that might impact it a tiny bit but not to the extent it's dropped.

danny e. wrote:

After fix:

morphine wrote:

Yeah, there's even some news about it in this obscure site.

http://techreport.com/review/27212/sams ... h-old-data

Samsung wrote:

Performance Restoration DOS Version 1.0

Performance Restoration DOS version 1.0 features
- Support Model : Samsung SSD 840 EVO, 840 EVO mSATA
- Firmware Update
- SSD Performance Restoration

Performance Restoration Version 1.1

Performance Restoration version 1.1 features
- Support Model : Samsung SSD 840 EVO, 840 EVO mSATA
- Firmware Update
- SSD Performance Restoration
※ Samsung Performance Restoration v.1.1 does not work, if users did performance restoration through v.1.0

Improvements
1. GUI (Progress bar) improved
2. Improved compatibility of detecting drives with some PC systems

The Egg wrote:

It's been pretty quiet. Now that some time has passed, I'd be interested to see results on how well the firmware update worked.

morphine wrote:

...You're going to be limited by the SD card's speed and the USB interface itself.
...

emphy wrote:

High speed SD cards are really,really fast nowadays