Assistance - Calling all Samsung EVO owners
- Thread starter Aurhinius
- Start date
Deleted member 138126
D
Deleted member 138126
Effectively the drive is defragging itself.
right but i think everyone would pick a drive that doesnt have to do that, unless it was very very cheap!
and there must be a short time right before the "periodic refresh" where the drive is not as fast as advertised
seems like a crazy positive spin to put on a fix that isnt really a true fix, and doesnt help consumers right now looking for refunds lol
and there must be a short time right before the "periodic refresh" where the drive is not as fast as advertised
seems like a crazy positive spin to put on a fix that isnt really a true fix, and doesnt help consumers right now looking for refunds lol
Deleted member 138126
D
Deleted member 138126
But what is the alternative? To recall potentially millions of drives? Imagine how much that would cost them, and how disruptive it would be to the owners (having to transfer files from one drive to another). Remember, we are the enthusiasts; the vast majority of users won't care about the nitty-gritty, as long as the performance is restored. I think it is the right call. Clearly they have a design issue that can't be fixed in software, and a slow background refresh is the best fix available to them.
Deleted member 138126
D
Deleted member 138126
The slow-down starts occurring after X number of days. So as long as a given cell is refreshed before X days have elapsed, you have optimum performance, and the user never notices a thing.
Deleted member 138126
D
Deleted member 138126
I reckon the reason it has taken 6 months is because developing this background refresh process is obviously very tricky. They are restricted to the processing power available in the existing controller, and it has to be seamless, and rock-solid. So it requires a lot of debugging and a LOT of testing.
Deleted member 138126
D
Deleted member 138126
Note that I am not defending Samsung. I own a 500GB 840 Evo, so I am affected like everyone else here. However, I also accept that a problem that develops over time (months), is very difficult to discover during development, and Samsung has limited options.
Would be interesting to see if they established a programme to replace any 840 Evo no questions asked if people wanted to go down that route.
Would be interesting to see if they established a programme to replace any 840 Evo no questions asked if people wanted to go down that route.
They should do an optional recall imo. Let the consumer choose if they wish to have fully working as intended drive or stick with the "fix".
Deleted member 138126
D
Deleted member 138126
100% agree.
Anandtech have installed the fix and even though they didn't give the drive any idle time after the firmware update, performance was mostly restored. That suggests the fix isn't purely to periodicly refresh the data, there must have been other changes in the firmware that instantly restores the majority of the performace. Of course this is presuming the data isn't rerwiten during the firmware upgrade, we will know if that's the case when we get our hands on it and install it, if it takes seconds it's not refreshing the data along with flashing the firmware, if it takes several minuites it likely is.
http://anandtech.com/show/9158/new-samsung-ssd-840-evo-read-performance-fix-coming-later-this-month
At the moment it seems to me the firmware tweaks have mostly fixed the issue and it will only need to resort to rewriting some of the data some of the time, rather than all of it at regular intervals. When get the fix time will know for sure.
I predict some drives will need more auto rewriting than others as there is natural quality variation in the nand dies, just like some CPUs overclock more than others. But on the whole the impact on write cycles will be negligible.
http://anandtech.com/show/9158/new-samsung-ssd-840-evo-read-performance-fix-coming-later-this-month
At the moment it seems to me the firmware tweaks have mostly fixed the issue and it will only need to resort to rewriting some of the data some of the time, rather than all of it at regular intervals. When get the fix time will know for sure.
I predict some drives will need more auto rewriting than others as there is natural quality variation in the nand dies, just like some CPUs overclock more than others. But on the whole the impact on write cycles will be negligible.
I don't think it's anything like a "periodic automated secure erase".
I think it's more likely to be a selective data refresh that happens "in the background".
There's two ways of looking at this.
1) Samsung "messed up". Have take quite a long time to (hopefully) fix the problem, and the fix might have a very small effect on performance (when it's carrying out the refresh) and might reduce the life of the SSD. Therefore we should all demand a refund/replacement.
or
2) The issue hasn't caused any data loss (which would be a big problem), and in a lot of cases went unnoticed, due to the fact that the reduction in sequential read speed looked at lot worse on paper than what it did in most real usage. If the new fix does maintain performance, and has little if any effect when the SSD is in use, then what's the problem ? The 840 EVO will perform as stated. Any reduction in lifespan is likely to be minimal, and "wearing out" will probably be due way after most users have replaced the drive anyway.
Personally, I'll be quite happy if the fix does do what we hope it will do. If, after a few months, we are still seeing performance degradation, then it'll be time to ask for replacements, as Samsung will have had a fair chance to solve the problem.
On the "increased wear/reduce lifespan" issue, I would have thought that if the new refresh algorithm only refreshes data that's been on the drive for a while (and therefore slowing down), then it won't be a problem. Heavily used drives will be getting a lot of data written by the user, therefore this data will not require much refreshing, as it will probably be re-written by the user fairly regularly anyway. Drives that have an easier life will not be having much data written on a regular basis, so the user is not giving the drive much wear. Yes, the refresh will have to do more work to keep the stagnant data up to speed, but I don't think that will mean that these users will be having to replace their SSDs any time soon. I think the wear issue is not really an issue, but it's one that still seems to get mentioned by some when deciding to buy an SSD.
TBH, my EVO (a "bad" one) has been OK.
Yes, I know that it slows down, but until the issue was announced, I wasn't really aware of any problem. I`ve carried out a few secure erases to restore the performance, but that was more to do with trying to see what was going on, than any real problem. My PC does slow down, but it isn't a drastic slow down.
I want the issue fixed, but I can't say that I'm too troubled by it. As I`ve said, I'd rather have slowing data than vanishing data. However, I expect it to be properly sorted this time, or I will want Samsung to replace my drive.
And I have switched to Crucial in the meantime, not just as a matter of principle, but also a matter of the MX100 being generally cheaper than the 840/850 EVOs.
Yes, I know that it slows down, but until the issue was announced, I wasn't really aware of any problem. I`ve carried out a few secure erases to restore the performance, but that was more to do with trying to see what was going on, than any real problem. My PC does slow down, but it isn't a drastic slow down.
I want the issue fixed, but I can't say that I'm too troubled by it. As I`ve said, I'd rather have slowing data than vanishing data. However, I expect it to be properly sorted this time, or I will want Samsung to replace my drive.
And I have switched to Crucial in the meantime, not just as a matter of principle, but also a matter of the MX100 being generally cheaper than the 840/850 EVOs.
Last edited:
When I asked OCUK about this, they referred me to Samsung.
Samsung told me that an update will be available by the end of March. Now I'm willing to give them a bit of leeway, but if the update isn't available by the end of April, I'm going to request more positive action from Samsung. If Samsung aren't willing to help, then I expect OCUK to sort something.
It really isn't an important decision.
The only problem is that it takes a while for Diskfrsh to do it's job. I used it a few weeks ago, it's no big deal.