Lol. I paid $65 for a 32gb x25. Wanted to get my toes wet with ssd and when it lost it's usefulness I found out it 32gb was enough to install gran turismo. Tracks load faster but the menu is still awfully slow. It's gonna get repurposed as an external drive and I'm gonna throw the Intel 320 160gb in it soon.
-
After reading many posts about this issue with 840 evo I want to know if I got it right?
1. Samsung in it's statement and even title of this thread says "some drives" are affected. Am I correct in thinking, it actually means all drives, since they're all pretty much the same? Just some people won't notice it since it affects least used data and therefore hard to spot without benchmark.
2. most talk about SSD drives, am I assuming correctly msata is the same as SSD and also affected?
3. I'm not clear if this issue affects only old data that hasn't been read at all for long time (for example old back up of hard drive), or does it affect any old written data, even if it was read many times since, but not rewritten (for example program files of games etc. which are read and loaded into memory often.
Thanks for any insight you may have. -
tilleroftheearth Wisdom listens quietly...
1. Samsung is only covering it's ***. ALL TLC drives it has manufactured prior to the 850 EVO are affected. For all we know, the 850 EVO is also affected but just needs time to show the effects. This includes the 840 EVO and the original 840 TLC drive which Samsung is still in denial about.
2. 2.5" SSD's and mSATA variants both affected.
See:
http://www.overclock.net/t/1507897/...-written-data-in-the-drive/2170#post_23672741
(and search the linked thread to see more examples).
3. Old data is defined as data written once and not moved or updated since. Does not matter how often it is read. It will continue to slow down to pre-4200RPM HDD levels. When it is moved (even to a new location on same drive and back), it is old data that has been refreshed and will not read slowly for about 3 weeks to 3 months, depending on the luck of the draw.
Here is what the issue is in a nutshell:
See:
http://forum.notebookreview.com/thr...-and-1tb-samsung-850-pro.773267/#post-9962224
Samsung rigs every benchmark it can. This issue cannot be caught with standard benchmarks. Most BM tools write a new file and read it and the owner gets a warm glow and says 'wow' great 'score'. The custom bm tool in the following thread cuts through that loophole and shows the issue for what it is.
See:
http://www.overclock.net/t/1512915/...es-benchmarks-needed-to-confirm-affected-ssds
There is no way I would believe Samsung did not know this issue existed in it's internal testing. It simply decided that the best course of action would be to deceive it's customers and release this time bomb. And it continues to deceive and stall on this issue today. Even after the first fix is shown as non-effective. And still denying the original TLC 840 is affected at all.
Best course of action? Boycott Samsung products. Even their so called 'pro' line.
All they have is marketing and the hoopla the online media can create for their products via advertising dollars.
See:
http://www.tomshardware.com/reviews/silicon-motion-sm2256-ssd-preview,4066-6.html
For years; Middling performance in real world (non-cheatable) benchmarks, not to mention the 'snap' that an SSD should have. Shortcuts taken that undercut quality and performance and basic fit for purpose. Continued denying of any problem that is easily shown is spread wide and deep (across product lines). The very real possibility that new issues will appear and in true form; Samsung will deny and effectively not address them either.
Look at the graphs in the link above. Now, tell me which is the real 'Pro' model? -
^the M550
What it performed the best after the 3rd recovery phase in both total and light workload bandwidth, and 99.9% of those who buy Crucial SSDs aren't going to use it for heavy work anyway.
*cough*
I get this eerie feeling Samsung is going the way of OCZ with regards to their SSDs. Also, if SanDisk named their next SSD "Extremely Pro" I'd buy it just for the name hahaTomJGX and alexhawker like this. -
Hmm ... wasn't in a hurry since the large/important files are rewritten daily anyway, but boot-time was affected too, so ...
Before:
After:
Fine if this is a one-off, yet hardly a happy camper if this needs a monthly re-run (or weekly
?). Anyone tried RMA-ing successfully because of this?
Kinda hoping for a 1TB MLC mSATA soon, but if that isn't going to happen then there's still some 512GB alternatives. Not looking forward to image-swaps on striped disks, but if that's what it takes ... -
Robbo99999 Notebook Prophet
Ah, I've just found this thread, was posting in another thread which wasn't quite as applicable as this one! Here's my post copied & pasted from the other thread:
Hi All, I have an update for you on the Performance Degredation of my 840 EVO 250GB, for those that have followed this thread then you'll know what I'm talking about.
It looks like my EVO is finally fixed, although only more time can truly tell. No longer am I experiencing Read Speed Degredation of Old Data. I fixed it by doing a Secure Erase of the drive and then restoring Windows from a previous image. To me it looks like the Secure Erase AFTER having updated to the latest firmware (installable from Samsungs Performance Restoration Tool) has been the deciding factor. It's been 11 weeks since the re-image and no slow downs at all, whereas prior to the secure erase I had slow downs after less than 8 weeks! Following is the summary of the entire history of the drive so you can know what's going on without reading all my previous posts:
1) Applied Samsungs Performance Restoration Tool (which also upgrades to latest firmware, and should have prevented any Read Speed Degredation of Old Data) as soon as I had bought the drive (after installing Windows).
2) After less than 8 weeks Read Speed of Old Data decreased from 530MB/s to the 300's.
3) Secure Erased Drive and installed Windows by re-imaging. Read Speed now back to approx 530MB/s. Also made some other changes to a few configurations: a) RAPID not used b) Write Cache Buffer Flushing Enabled (previously I had had this disabled for supposed better performance), although I still have Write Caching enabled.
4) 11 weeks after the secure erase & reinstallation Read Speeds still at approx 530MB/s (see attached, C:OS, F:Games) = RESULT!!!
In conclusion, I think it was the Secure Erase that was applied AFTER the firmware fix had been applied which made the difference (although it can't be ruled out it could be the other configuration changes that I previously listed that might have made the difference: buffer flushing turned on / RAPID turned off). Interestingly the amount of writes recorded by the drive (viewable in Samsung Magician or HWInfo for instance) has increased since this procedure - increased to about 10GB per day, previous to the secure erase this figure was lower; perhaps the firmware is now working as it should and it is gradually & continually re-writing data to maintain performance(?) - because I don't as a user do anywhere near 10GB of writes to the drive, I hardly ever save anything to that drive!t456 likes this. -
That's great news, thanks for sharing
!
Think the 10GB/day sounds fine and perhaps the increase can be attributed to the erase + image restore action? Other than that, the firmware flash might've caused the drive's system date to reset ... There's also some games that induce a vast amount of small-file writes. -
Everyone here knew why it was cheaper. Everyone was told not to buy it.
You got what you paid for, why complain? No, seriously, why?
It's called "drawback of such technology" -
I have to agree with James on this.. You were warned Robo... Well at least it works alright now after a lot of hassle..
-
Robbo99999 Notebook Prophet
The 10GB/day is not due to the extra writing that happened during the image restore, it's an ongoing 10GB/day that happens from now until...eternity I guess! (Over a period of a couple of weeks each day I've been writing down the TB of writes that the drive has done as seen in Samsung Magician or HWInfo, so it's about 10GB/day). Maybe it's the games, but I doubt it, I don't think they write 10GB/day - I think it's the firmware doing it.
I bought the drive before I read about the Read Speed Degradation issues, but yes I hope it's fixed now - only time will tell but it's looking rosy so far! -
let's see 10gb/ day = 3.65 tb /yr so at this rate you'll reach 40 TB in 11 years and
1. I don't think 840 evo is warranty restricted on TB
2. in testing 840 evo started having errors at 600TB and failed at around 900 TB
I wouldn't really worry about it.
Now I have been looking at my drive and it seems I can have about 5GB writes /hr, just surfing a web. For example if you have hibernation enabled, that's about 6 GB right there, every time you hibernate and Win8.1 could be hibernating instead of shutting down if you have fast boot enabled. Also you could be swapping RAM , even if there is free RAM still available etc.
BTW 10 GB a day is assumed average user, for average writes in most tests and specifications for SSD, so I would say you're as average as can be.Cloudfire likes this. -
I don't think that anyone who posted in this thread has expressed their concern about the longevity of the SSD in question.
About massive and well-documented slowdowns, yes.Robbo99999 likes this. -
OP suspected the fix for slowdown, also well documented, (rewrite older data a new) is using up his write cycles, so I stated the obvious, that the fix either build into firmware by Samsung or manually by diskfresh etc., shouldn't affect longevity of the drive that much, so maybe slowdown itself is not the end of the world, as some make it, since there is a fix more or less.
As a matter of fact, just yesterday I experienced massive slowdowns (down to 6MB/s ) on my barely 1 yr old Seagate HDD, so probably it will fail soon, so here I actually wish I had 840evo type of issue than what I have with this HDD. -
The poster that you were responding to was not the OP...that's why it helps when one quotes the actual post that they're responding to, at least in my opinion...
No one said that it was the end of the world, but it's definitely unacceptable, and the manner in which Samsung has handled the whole problem is nauseating. Why would anyone in their right mind accept an SSD that slows down to HDD speeds within months?
Also, judging by the thread on OC forum, the fix is definitely "less" and not "more".
http://www.overclock.net/t/1507897/samsung-840-evo-read-speed-drops-on-old-written-data-in-the-drive/2330
Post # 2335 is one of my personal favourites...tilleroftheearth likes this. -
tilleroftheearth Wisdom listens quietly...
If they only slowed down to current HDD levels. But of course, the plummet way below that. 26MB/s reads on EVO drives in a few short weeks/months is what makes Samsung the OCZ of 2014/15.
Burn them at the stake, sell their remains for pennies on the dollar, simply ignore them, but whatever you do... Don't buy Samsung junk until and if they address this issue in an appropriate way.
And yeah; post 2335 on that thread is a good analogy and the course Samsung should follow too at a minimum.
I would take a refund any day though than even a 'pro' replacement from them.TomJGX likes this. -
Robbo99999 Notebook Prophet
Yep, thanks for working out the writes per year, etc, based on my 10GB/day usage, but I wasn't worried about the extra writes & longevity of the drive because I knew it wasn't an issue. But it does look like writes have increased since my drive is working better since the Secure Erase after the firmware upgrade, and I'm not worried about that. My main point is that I might have found a way for 840 Evo users to fix their drives for good (hopefully, but more time will tell for certain), by doing the following:
1) Apply Samsung Performance Restoration Tool (which upgrades firmware to latest version too).
2) Then Secure Erase drive & reinstall Windows (either from an image or fresh).
3) Disable RAPID. Enable write caching, Enable Buffer Flushing, (options available in Device Manager or through Samsung Magician).
I think it's more likely it was the Secure Erase after applying the Samsung Performace Restoration Tool that made the difference, I don't think point #3 above was the cause (although I can't prove it).Cloudfire likes this. -
That makes sense; either they've made a workaround or this is what it should've done all along. It's not much of problem though; it's already at 40GB/day, so an extra 10 only reduces longevity by a quarter. Will reach the TBW in 3 years hence and will have 1TB's long before that. And, as mentioned, endurances tests indicate even more leeway, live one running here; SSD Endurance Test – Live.
Same here, though, being TLC, mine were a cheap/fast interim option anyway.
It is, but they don't advertise this quite so prominently:
Couldn't find these for the 840 Pro.
Hmm ... (thinking in lawyer-mode) ... since they call it “ Total”, the warranty actually expires after 75 bytes written. -
I'm quite sure you got the 840 Evo after these read speed issues came about even though a lot of people said not to get it... Anyways it doesn't matter at the end of the day.. it's working fine for you now and that what matters
-
Robbo99999 Notebook Prophet
Haha, naughty me! No, even though read speed issues obviously existed when I bought the drive, I had no knowledge of them. Not that anyone cares!TomJGX likes this. -
Robbo99999 Notebook Prophet
Well this is a bit of a revelations for the 840 EVO, looks like Samsung are releasing a new firmware at the end of the month that fixes the Read Speed Degradation of Old Data Issue, along with a new version of Magician (v4.6). Apparently the new firmware periodically refreshes old data on an ongoing basis, as well as optimising the readability of old cells that have not been refreshed yet. In addition, Magician 4.6 will also contain an 'Advanced Optimisation' button which users can manually click at any time to reorganise data on the SSD into a sequential pattern (to defrag files), which looks to increase read performance of the SSD beyond the initial capabilities of a previously perfectly functioning 840 EVO. The article is written by a review site that have been testing Samsungs Beta Firmware & Beta Magician 4.6:
http://www.pcper.com/reviews/Storag...40-EVO-EXT0DB6Q-Firmware-Review-Finally-Fixed
(EDIT: my 840 EVO is still running today at perfect read speeds since the last fix I posted earlier in this thread (Post #56), but I'll probably apply the new update anyway when it comes out at the end of the month.)Papusan likes this. -
From anandtech; All in all, I hope this fix will finally put an end to the performance degradation. The issue has been bugging many users for months and it's critical that the users get what they initially paid for. On one hand I'm confident enough to say that this fix is permanent given the way it works, but on the other hand I don't want to be too optimistic this time around because the first fix didn't turn out so great. Either way, I think this fix is the last chance for Samsung to provide a permanent solution because they already failed to do so once and it would no longer be fair to ask the customers to wait months for a fix that might or might not fix the issue. For now the only thing we can do is wait for user reports and hope for the best, but at least in theory the new firmware should be a permanent fix.
http://anandtech.com/show/9158/new-samsung-ssd-840-evo-read-performance-fix-coming-later-this-monthLast edited: Apr 14, 2015Robbo99999 likes this. -
TechReport also has a piece: (emphasis mine)
If you read between the Samsungs lines (it's 1am), I think it's pretty clear Samsung is trying to convey -- in nVidia PR doublespeak fashion -- that the 840 Evo is flawed at the hardware level and the only fix is to refresh the data in the NAND cells on a regular basis permanently.
In other words, the 840 Evo is Samsung's GTX 970, and they're not stepping up to the plate either aside from releasing a permanent band-aid fix. Gotta love big corporations these days.
And so glad I pawned off my 840 Evos before fecal matter hit the fan.
TomJGX, alexhawker and ajkula66 like this. -
Samsung Gtx 970 Evo Edition ... Are Nvidia and Samsung same company? YES...LoL
Last edited: Apr 15, 2015TomJGX likes this. -
GTX 970 Evo edition now with 1GB of G-Cache™
Gotta love that G-Cache man
-
Ssd firmware from Samsung and Overclock blocked vbios for the new gtx970/980m are perhaps made from the same developer = NVISAM ... A new cooperative by Nvidia and Samsung.
-
Robbo99999 Notebook Prophet
Latest firmware available now through Magician 4.6 for 840 Evos. Supposedly this has sped up many of the Evo drives when it comes to read speed of old data issue. I've upgraded my Evo sucessfully through Magician 4.6 yesterday, no issues so far, read speeds of old data fine. I'll monitor over the months to come to see if I get any slowdowns. Here are some places you can get the latest Samsung Magician 4.6:
http://www.samsung.com/global/busin...I53ztyBF44w6R5Y76YEphWZC01OmRfBxVfxoCpgLw_wcB
http://www.techspot.com/downloads/5345-samsung-ssd-magician.html
http://www.majorgeeks.com/files/details/samsung_ssd_magician.html
(Samsung website experiencing download limits on Magician 4.6 - I think they're doing this on purpose to trickle out the new firmware to people, as a way of testing their new firmware before 'widely' distributing it to everyone. You can download the same program though from the other links above - you can then upgrade the firmware to the latest version through Magician). -
Spartan@HIDevolution Company Representative
-
tilleroftheearth Wisdom listens quietly...
And yet, and yet...
See:
http://www.overclock.net/t/1507897/...-written-data-in-the-drive/2710#post_23835322
After applying the latest and greatest Samsung junk, the performance gets worse.
I can see them pulling this heap soon (glad I didn't try too hard to download it) if the post/experience by http://www.overclock.net/u/110246/tekwarfare is mimicked by even one more person with an EVO. -
Robbo99999 Notebook Prophet
The post you quoted there, his read speeds were fine before the update, so the update didn't have anything to improve upon. (And his read speeds did increase by 10MB/s anyway).
EDIT: apart from the HDTach results, which got worse, but I remember reading that HDTach is not very reliable. -
tilleroftheearth Wisdom listens quietly...
You are more forgiving than me.
Samsung is a 'score hoar' who believes that the biggest numbers sell the best; I can't really believe they would allow an 80MB/s drop in any synthetic benchmark just to fix a real problem...
-
Robbo99999 Notebook Prophet
His SSD is fine, the guy you quoted, he has no concerns. I just don't think HDTach is that reliable or relevant, the other benchmark scores were fine for him. Most reports seem to be showing an improvement with the latest firmware, my own SSD for instance too (although mine was performing top notch before the update, and now still so). Time will tell though for sure, to see if there's any degredation, and it will be interesting to see more graphs from other users with the new firmware too (at the link you provided, I've been keeping an eye on that thread as well). -
tilleroftheearth Wisdom listens quietly...
Time will tell, of course.
But I'm not betting on Samsung if my life depended on it.
This is an issue of crap hardware. Software/firmware won't fix this.Incontro likes this. -
Robbo99999 Notebook Prophet
Samsung Magician now downloadable from Samsung website, previously it had reached it's download limit:
http://www.samsung.com/global/busin...t/downloads.html?CID=AFL-hq-mul-0813-11000279
Yeah, it's not the best SSD on earth, but I do like the fast SLC cache, which makes small drives like the 250GB have fantastic write speeds for 'small - medium' duration, which is perfectly fine for light usage. Read speeds are awesome (if they've now permanently fixed the Read Speed Degredation of Old Data!), especially 4K QD1 at 38MB/s on my system, which makes it a snappy performer for the OS for a good bargain price. You wouldn't want one for a server though I guess! -
SanDisk Ultra II is also TLC NAND, and also has a fast SLC cache (achieving 460MB/s write for a tiny 128GB drive), but isn't plagued by unacceptable slowdowns.
TomJGX likes this. -
Robbo99999 Notebook Prophet
Good!
I don't know anything about that drive, but 460MB/s writes is good, 840 Evo does 490MB/s sequential writes on my system though - we're both comparing the SLC cache speeds I guess. Hopefully now the Evo's won't be having the read speed degredation of old data now anyway (fingers crossed!).
-
Is Firmware Samsung_SSD_840_EVO_EXT0DB6Q.iso compatible with mSATA 840 Evo?
-
No...
-
Spartan@HIDevolution Company Representative
No and there won't be an mSATA firmware update according to one guy at OC forums. Samsung will leave us mSATA users in the dust the same way they did with the 840 Vanilla users -
Yet another reason for you to skip Samsung...
No more Samsung...
TomJGX and Spartan@HIDevolution like this. -
Robbo99999 Notebook Prophet
That's weird that mSATA may not get the firmware. I'd always thought that they were essentially the same SSDs with the same hardware, but just in a different format - looks like there's probably some significant differences considering Samsung haven't released a fixed mSATA firmware yet. -
The firmware is specific to the very last bit of the hardware. The pinout of the mSATA is different, thus a small portion of the firmware should differ. The outrageous part is that its probably just a couple of lines code and Samsung doesn't make the effort to provide it.Spartan@HIDevolution likes this.
-
Robbo99999 Notebook Prophet
There must be something significantly different that's delaying/preventing a firmware release for mSATA, I'm pretty sure it's not just Samsung not making the effort to provide, because if it was so simple then of course they would have released it already - they've committed themselves to fixing the problem so if it was an easy fix it would have been done for mSATA already. I'm just surprised that there seems to be such a difference between the mSATA and 2.5" when initially I had thought there would be minimal differences. -
Spartan@HIDevolution Company Representative
Well to be honest and fair, my mSATA 1TB 840 EVO benchmarks faster than my previous 1TB 840 EVO 2.5" and never did suffer from any performance loss so I am really not concerned about them not issuing a fix. Guess I got lucky with some golden sample but it has been flawless with 24 TB written to it so much. I tortured this SSD badly. -
If the bad torture means lots of rewrites the Old data bug should not come up, since there might not be any old data. Or maybe the pieces of the smaller mSATA units are more carefully selected?Spartan@HIDevolution likes this.
-
Spartan@HIDevolution Company Representative
I cannot really tell since all I have on the mSATA is MP3s, Videos, Pics.....so I wouldn't notice to be honest but benchmark wise it always beat the 2.5" version heck its 4K reads/writes are even higher than my 850 PRO at times......
-
Seems perfectly healthy and indeed, very fast
Maybe you activated the secret SLC mode in the firmware
Spartan@HIDevolution likes this. -
Spartan@HIDevolution Company Representative
Didn't touch a thing bro....... my 4K scores have never hit over 37 even on the 850 PRO. I don't know what's the magic in these mSATA drives I bought a while back....
I have one that I am using with 24TB written to it and another one sealed still in the box that I've been trying to sell for months but noone has even called me for it LOL -
Maybe the rest of the mobo's infrastructure is better suited for 4k in case of mSATA, no clue...
Jeah, selling an SSD is hard, especially an 840...
Some 840 EVOs still vulnerable to read speed slowdowns
Discussion in 'Hardware Components and Aftermarket Upgrades' started by Spartan@HIDevolution, Feb 7, 2015.