Two Samsung 840 Pro's & 1 840 have failed, 4 at [H]ardOCP (Early review sample SSD's)

[fov001]

Thank God I came to this thread. I was just about to buy the 840 Pro, but will read around to see if there is a better option or a fix.

Any recommendations for a RAID 0 setup would be appreciated though

 

[zod96]

Just spoken to a friend of mine had no idea he bought one as well. He's had since release day and again not a single problem either. Anything I should look for when its failing or messing up?

 

[jwilliams4200]

If it dies like the others, you will know. It just bricks. Although one of the four failures from hardforum started giving 30sec pauses before it failed.

The one thing in common with all the failures I've seen reported (2 x 840Pro anandtech, 1 x 840 anandtech, 4 x 840 Pro hardforum) is that they were all being written to with a heavy workload. But heavy writing is not sufficient to kill one, since I've been really hammering my 840 Pro and it is still alive.

 

[Elixer]

The one thing in common with all the failures I've seen reported (2 x 840Pro anandtech, 1 x 840 anandtech, 4 x 840 Pro hardforum) is that they were all being written to with a heavy workload. But heavy writing is not sufficient to kill one, since I've been really hammering my 840 Pro and it is still alive.

Do you mean many I/O requests with small files / block sizes, or huge files / block sizes by the 'heavy workload' remark ?

 

[jwilliams4200]

Do you mean many I/O requests with small files / block sizes, or huge files / block sizes by the 'heavy workload' remark ?

Yes.

 

[Hellhammer]

I didn't think you had managed to write enough to kill the drive through writes? Does that mean that is just died?

Yes, it just died. NAND should have been good for another +100TiB of writes.

Don't those statements conflict with each other ?
I assume Sammy told you not to tell people something, which is why you wrote "...when we can talk about it" ?

I wasn't sure if I could talk about what Samsung had told us about the fix. Quite often manufacturers provide us preliminary info on fixes but that's under NDA because they don't want it to be published until they are certain it works. Anand mentioned about it in Twitter so I think I have a green light here:

Samsung is claiming that the issue is a firmware bug and they already have a fix for it (supposedly no retail drive should have shipped with the broken firmware). We'll be getting samples with the new firmware soon.

You still have lots of readers that have no idea there are potential issues with the 840 Pro, and they don't frequent every thread in the forums, so it should be made clear on the front page (not tacked on as a 'update' on the first review) that all your samples(?) have died.

True, could have been a bad batch, but this really should be made clear on the main site, just like when the OCZ reliability issues turned up.

The mention of a dead 840 Pro sample has been in the review since day one. Sure, it was now updated that the second sample died as well but without any further info, I don't think there is a reason to post a separate article. People who read the review when it was published are aware that the 840 Pro may have potential reliability issues and people who read it now will know that for sure.

 

[jwilliams4200]

Samsung is claiming that the issue is a firmware bug and they already have a fix for it (supposedly no retail drive should have shipped with the broken firmware). We'll be getting samples with the new firmware soon.

Hmmm. I hope it is something that can be fixed through firmware. But the claim from Samsung -- that it can be fixed through firmware, but that no retail drive had broken firmware -- is strange. Are they really claiming that all the retail SSDs shipped already have the fixed firmware? If so, why did they send a second 840 Pro to anandtech with the broken firmware? That was received within a week of the retail 840 Pro SSDs shipping, right? Certainly whatever firmware the retail units shipped with should have been available for the second anandtech sample.

Did Anand record the firmware version of the second 840 Pro sample? My retail 840 Pro (from newegg) has firmware version DXM03B0Q.

The person on hardforum had 4 out of 4 fail -- he should have received retail units, since he was not a reviewer. I guess it is possible that his company received pre-release units, but I did not get that impression from his posts.

I've been checking this page every day looking for firmware updates, but nothing yet:

http://www.samsung.com/us/support/owners/product/MZ-7PD256BW

EDIT:

I just went through 10 online reviews of the 840 Pro, but only 4 of them specified the firmware version (anandtech was one of the no-tells). It seems that firmware version is something that should be included in every review.

Anyway, all four of them had firmware DXM02B0Q, which is apparently one version older than what shipped on my retail unit. However, all of those reviews were from Sept. 24. It seems like there would have been time for the second sample sent to anandtech, which must have been a month later, to have the updated firmware. Did it?

For the record, here are the four reviews that specified the firmware:

http://www.custompcreview.com/reviews/samsung-840-pro-series-256gb-ssd-review/15768/4/
http://thessdreview.com/our-reviews...ew-killer-performance-and-untouchable-iops/2/
http://www.tomshardware.com/reviews/840-pro-ssd-toggle-mode-2,3302-2.html
http://hexus.net/tech/reviews/storage/45653-samsung-ssd-840-pro-series-512gb/?page=2

 

[Hellhammer]

Hmmm. I hope it is something that can be fixed through firmware. But the claim from Samsung -- that it can be fixed through firmware, but that no retail drive had broken firmware -- is strange. Are they really claiming that all the retail SSDs shipped already have the fixed firmware? If so, why did they send a second 840 Pro to anandtech with the broken firmware? That was received within a week of the retail 840 Pro SSDs shipping, right? Certainly whatever firmware the retail units shipped with should have been available for the second anandtech sample.

Samsung sent us the second sample right after our first sample had died, so that was over a month before the 840 Pro entered retail. I don't think Samsung had a fix for the issue when our first sample died, it's possible that they didn't even know about the issue.

Did Anand record the firmware version of the second 840 Pro sample? My retail 840 Pro (from newegg) has firmware version DXM03B0Q.

I'm pretty sure he did. I'll ask him when he gets online and report back.

The person on hardforum had 4 out of 4 fail -- he should have received retail units, since he was not a reviewer. I guess it is possible that his company received pre-release units, but I did not get that impression from his posts.

That's what shocked me and Anand as well. His units should be retail without the firmware bug...

 

[jwilliams4200]

Samsung sent us the second sample right after our first sample had died, so that was over a month before the 840 Pro entered retail.

Oh, I did not realize anandtech was sitting on the second sample for so long.

Okay, I guess they both had the same (older) firmware version then.

 

[jwilliams4200]

That's what shocked me and Anand as well. His units should be retail without the firmware bug...

He just posted that he won't be able to check the firmware version until Tuesday, but he did say that his SSDs "came as advance release versions from a Distributor". So it is looking possible that he did have units with a firmware earlier than the shipping retail version.

http://hardforum.com/showpost.php?p=1039334774&postcount=82

I hope it does turn out that his units had firmware -2B0Q or earlier, since that would neatly explain all the 840 Pro failures that have been reported, and we can go back to trusting our retail 840 Pro SSDs...

 

[mrali95]

Samsung sent us the second sample right after our first sample had died, so that was over a month before the 840 Pro entered retail. I don't think Samsung had a fix for the issue when our first sample died, it's possible that they didn't even know about the issue.



I'm pretty sure he did. I'll ask him when he gets online and report back.



That's what shocked me and Anand as well. His units should be retail without the firmware bug...

That's what shocked me and Anand as well. His units should be retail without the firmware bug...

Well hopefully they weren't retail units, I'd like my 840 Pro to live a good life.

Btw, does anyone know if the CXM03B1Q is a destructive update? The magician software gives a warning prompt when clicking the update tab "Warning: There's a risk of data loss when updating firmware and backup is recommended"

The only place I've seen the update is: http://www.samsung.com/us/support/owners/product/MZ-7PC256N/AM

 

[jwilliams4200]

Btw, does anyone know if the CXM03B1Q is a destructive update? The magician software gives a warning prompt when clicking the update tab "Warning: There's a risk of data loss when updating firmware and backup is recommended"

The only place I've seen the update is: http://www.samsung.com/us/support/owners/product/MZ-7PC256N/AM

That would probably be quite destructive on an 840 Pro, since that is firmware for an 830. :'(

 

[mrali95]

That would probably be quite destructive on an 840 Pro, since that is firmware for an 830. :'(

Damn, how could I miss that. Truly sorry , maybe there ain't any updates out yet. Have just installed the magician software and thought I would check for updates, but when I saw the warning I thought I'd check.

 

[Coup27]

mrali95 - do you have an 840 Pro purchased from a regular retail store or website? If so, what firmware is on it?

 

[GPz1100]

@mrali95, In general, I would never update a firmware on something unless I had a backup of its contents. While it is unlikely for data to get erased, it is all too easy for it to happen. A ssd backup shouldn't take too long to generate, and will pay huge dividends if its contents does get wiped out (lessons well learned from flashing other devices - phones, routers, etc..)

 

[mrali95]

mrali95 - do you have an 840 Pro purchased from a regular retail store or website? If so, what firmware is on it?

I have the DXM03B0Q and I bought mine at a online retailer

 

[mrali95]

@mrali95, In general, I would never update a firmware on something unless I had a backup of its contents. While it is unlikely for data to get erased, it is all too easy for it to happen. A ssd backup shouldn't take too long to generate, and will pay huge dividends if its contents does get wiped out (lessons well learned from flashing other devices - phones, routers, etc..)

That might be true, just haven't cleaned ip my HDD yet so have no place to backup.

Also, I just checked at Samsung (840 Pro this time ) and there's no avaiable firmware updates so far at their page so nothing to worry about anyway: http://www.samsung.com/us/support/owners/product/MZ-7PD256BW

 

[Coup27]

Thanks. It is certainly indicating at the moment that retail units shipped with DXM03B0Q and advanced or review samples shipped with DXM02B0Q.

Do Samsung use the advanced channel and review samples as a final form of testing? Put them in the hands of experienced users who throw the kitchen sink at them and if they receive no errors it's the final green light to release?

I'm a massive Samsung fan and considering their track record and worldwide customers I would find it nearly impossible to think they would be slack enough to release firmware they know to be faulty into the retail channel. If it does turn out that faulty firmware only made it to reviewers then I am not too worried about that.

Kristian, hopefully you will have some insight on this?

 

[Hellhammer]

Do Samsung use the advanced channel and review samples as a final form of testing? Put them in the hands of experienced users who throw the kitchen sink at them and if they receive no errors it's the final green light to release?

I think that at some level, this is the case. In-house validation always has limitations because it's very hard to mimic every possible usage pattern out there. When you provide reviewers and some higher-class clients with the drives, you get the first picture of how the SSD behaves out in the wild. While this doesn't cover all possible workloads, it's a much more efficient way than trying to do everything in-house (think about the amount of different systems you would need, let alone all the tests). Especially enterprises often use customized systems and software, so recreating such setup can be very hard. It's also rather cheap because all you need to do is to send out some drives.

Sometimes we even get engineering samples from manufacturers and they are even looking for our input on the functionality of the drive (for example, we may think the drive would be better if the manufacturer did more active garbage collection, or the idle GC should be triggered faster). Enterprise clients get engineering samples all the time because they are often the most important customers and their input really matters (not so common with consumer SSDs, though).

 

[Coup27]

So could Samsung have avoided all of this uncertainty by telling AT that this was pre-release firmware and had you put that in your article too when your samples failed then I don't think people would have gotten concerned?

I have read forums outside of here and there has been a lot of chatter about failing 840 Pro's with references to the AT review with people now thinking there is a major problem.

 

[Hellhammer]

So really Samsung could have avoided all of this uncertaintyby telling AT that this was pre-release firmware and had you put that in your article too then I don't think people would have gotten concerned.

I have read forums outside of here and there has been a lot of chatter about failing 840 Pro's with references to the AT review with people now thinking there is a problem full stop.

I think we made it pretty clear that our drive was a pre-production sample:

Both drives will be available on October 15th, however in advance of the release Samsung provided us with a beta sample for review. We were only able to get a 256GB 840 Pro initially but we've already asked Samsung for additional capacities. The other bad news is after running through our client test suite and preparing the drive for a run through our enterprise suite, our pre-production sample died. This isn't the first time we've had an SSD die during our test process, pretty much every company has seen a failure during one of our reviews, but despite Samsung's excellent track record even it isn't immune from early issues. These drives are a few weeks away from retail and Samsung will be getting our sample back this week to figure out what went wrong.

http://www.anandtech.com/show/6328/samsung-ssd-840-pro-256gb-review

The way I see it, there is a problem until Samsung provides an official explanation and a fix. Right now we don't know if retail samples are affected as well, even though Samsung is claiming they aren't. I can't be honest and recommend the 840 Pro until we have had a sample with the new firmware for at least a few weeks without bricking it.

 

[Coup27]

Ah sorry, my bad. I did read the review but must have forgotten those lines. Serves me right for not re-checking first.

 

[wpcoe]

I noticed this comment from Anand toward the end of his Intel Explains 20nm NAND Endurance Concerns on the SSD 335 article dated 11/17/2012, so maybe we'll have more info soon?

I've been hearing more reports of dying Samsung SSD 840 Pros and I believe I know the cause (firmware related, should be fixed in the latest shipping revision) but I'm still waiting for confirmation on one last thing before explaining what's going on there.

(Bolding in the quote is mine)

 

[plion]

Can't wait to hear the news. Hopefully I'll be able to install mine next week

 

[Mfusick]

So can I get cliff notes ? Samsung 840 any good to get or not?