Any in-depth testing to see if 850 Evo is really not prone to degradation (edited)

[LTC8K6]

So I did check the two computers:

Performance Optimization with 850EVO and Magician 4.6:

Win7 - Yes

Win8.1 - No, with a message that says 8.1 supports TRIM automatically, so PO is not needed.

http://www.samsung.com/global/busin...site/SSD/us/html/whitepaper/whitepaper04.html

 

[omega3]

So I did check the two computers:

Performance Optimization with 850EVO and Magician 4.6:

Win7 - Yes

Win8.1 - No, with a message that says 8.1 supports TRIM automatically, so PO is not needed.

http://www.samsung.com/global/busin...site/SSD/us/html/whitepaper/whitepaper04.html

Then why does Samsung_Magician_46_Installation_Guide on p.7 section 26 say following:

Advanced Performance Optimization of Magician 4.6 is only supported in Samsung SSD 840 EVO 2.5” model

Just read you mentioned "PO".. is perhaps the Advanced performance optimization (APO) part exclusive to the 840evo?

 

[LTC8K6]

Then why does Samsung_Magician_46_Installation_Guide on p.7 section 26 say following:


Just read you mentioned "PO".. is perhaps the Advanced performance optimization (APO) part exclusive to the 840evo?

Just don't worry about it.

Buy something else.

You clearly don't trust Samsung, so let it go.

I've had it with the topic.

 

[dimovnike]

hello, I had an 840 EVO mSATA with the performance bug and the firmware update did not help. So I bought an 850 EVO mSATA in a hope to get rid of this issue and sadly the issue is present here too. It is exactly the same but it seems that 850 degrades faster! The issue manifests like this: When the drive is new the speed is blazing over 500mb/s but in a couple of months the speed degrades badly (some regions read as slow as 50mb/s). Then I rewrite it with dd (just rewrite the pages with the same data) and the speed is back up. I will try to post some screnshots with before and after benchmarks.

 

[dimovnike]

this benchmark http://imgur.com/4PGcld3 is done after 105 days of work:

9 Power_On_Hours 0x0032 099 099 000 Old_age Always - 2521
12 Power_Cycle_Count 0x0032 099 099 000 Old_age Always - 30
177 Wear_Leveling_Count 0x0013 099 099 000 Pre-fail Always - 6
235 Unknown_Attribute 0x0012 099 099 000 Old_age Always - 6
241 Total_LBAs_Written 0x0032 099 099 000 Old_age Always - 6235671697
(i removed attributes with zero value)

note the last part of SSD which is OK on the benchmark, that is unpartitioned space.

After the full rewrite with dd, the benchmark becomes this: http://imgur.com/OE8dgVY

As I already stated - the issue was absolutely the same on 840.

Do you have any idea about what is going on?

 

[omega3]

hello, I had an 840 EVO mSATA with the performance bug and the firmware update did not help. So I bought an 850 EVO mSATA in a hope to get rid of this issue and sadly the issue is present here too. It is exactly the same but it seems that 850 degrades faster! The issue manifests like this: When the drive is new the speed is blazing over 500mb/s but in a couple of months the speed degrades badly (some regions read as slow as 50mb/s). Then I rewrite it with dd (just rewrite the pages with the same data) and the speed is back up. I will try to post some screnshots with before and after benchmarks.

Anybody can tell if this also happens with the regular 850 EVO or just the mSATA version?

 

[coercitiv]

Anybody can tell if this also happens with the regular 850 EVO or just the mSATA version?

We need more data.

After seeing dimovnike's post I tested my PM851 again: average data rate was 460-480MB/s with with no spikes bellow 400MB/s. The PM851 is the OEM only 840EVO equivalent, uses same TLC. Both my SSD and my laptop BIOS required updates in order to fix the performance issues, previously spikes could be as low as 30MB/s. The data on my SSD is 2 months old.

The degradation is supposed to happen on the 850 EVO as well, albeit at a considerably slower pace, and it should be completely hidden by internal cell refresh algorithms.

We could really use data from more 850 EVO owners.

 

[john3850]

I took a new 30 month old 840 that was never used and put it in the wifes 2500k which is only used 15 minutes a week.
Read speeds were 500 MB or normal after a dirty copy install.
Now in less then two months or only 2-3 hours of total use the As-ssd benchmark stated read speeds dropped to be 225 MB.
Now I ran DiskFresh which brought the read speeds back to over 500 MBs that was six months back.

 

[Glaring_Mistake]

this benchmark http://imgur.com/4PGcld3 is done after 105 days of work:

9 Power_On_Hours 0x0032 099 099 000 Old_age Always - 2521
12 Power_Cycle_Count 0x0032 099 099 000 Old_age Always - 30
177 Wear_Leveling_Count 0x0013 099 099 000 Pre-fail Always - 6
235 Unknown_Attribute 0x0012 099 099 000 Old_age Always - 6
241 Total_LBAs_Written 0x0032 099 099 000 Old_age Always - 6235671697
(i removed attributes with zero value)

note the last part of SSD which is OK on the benchmark, that is unpartitioned space.

After the full rewrite with dd, the benchmark becomes this: http://imgur.com/OE8dgVY

As I already stated - the issue was absolutely the same on 840.

Do you have any idea about what is going on?

I have an 850 EVO that is about a year old and which contains a lot of static files since it is my game drive and tested it to see if it would be affected.
Didn't think so since I have tested it previously without any issues.

And here you can see the read speed as measured by HD Tune:

Mine seems to be ok.
Not sure what has caused yours to slow down.

Actually after discovering that ADATA SP550 suffers from read speed slowdowns (at least when it is left without power) I've begun testing several SSDs to see if their read speeds will slow down too.
850 EVO is one of the SSDs included in the test, but even if read speeds do slow down it is going to take time for that to become visible.

 

[dvsv]

Same problem on a 850 Evo 250 GB 2.5" !

Day 1:

After 8 months:

A little more than 1 year later:

After diskfresh:

There's a lot of old stale data (game instalation) on this disk!

Edit: Test was done w/ default 64KB block size.

 

[john3850]

Maybe the data has to be read a few times before cell voltage is set to a normal pattern.
If your minimum read speeds come only from old stale unused data that should not have much effect on normal reads.

 

[Palorim12]

Make sure when testing, you guys use HD Tune Pro 5.6. You can use the trail version. The free version 2.55, doesn't bench SSDs correctly. Also, drivers affect the results of the speed on HD Tune Pro as well, So9 make sure you have the most up to date AHCI drivers in your system.

Make sure in the settings you use Full Test, 8mb blocks, and Most Accurate.

 

[Palorim12]

this benchmark http://imgur.com/4PGcld3 is done after 105 days of work:

9 Power_On_Hours 0x0032 099 099 000 Old_age Always - 2521
12 Power_Cycle_Count 0x0032 099 099 000 Old_age Always - 30
177 Wear_Leveling_Count 0x0013 099 099 000 Pre-fail Always - 6
235 Unknown_Attribute 0x0012 099 099 000 Old_age Always - 6
241 Total_LBAs_Written 0x0032 099 099 000 Old_age Always - 6235671697
(i removed attributes with zero value)

note the last part of SSD which is OK on the benchmark, that is unpartitioned space.

After the full rewrite with dd, the benchmark becomes this: http://imgur.com/OE8dgVY

As I already stated - the issue was absolutely the same on 840.

Do you have any idea about what is going on?

What program did you use there?

 

[aviator79]

Now in less then two months or only 2-3 hours of total use the As-ssd benchmark stated read speeds dropped to be 225 MB.
Now I ran DiskFresh which brought the read speeds back to over 500 MBs that was six months back.

AS SSD is NOT showing, if your SSD is degrading in read speeds of old files. It only benches with fresh data...you don't even know what the problem is, and post such a nonsense.

All others:
Please use the tool SSDReadSpeedTester2.04 from here:
http://www.overclock.net/t/1512915/...needed-to-confirm-affected-ssds#post_22842513
to bench your SSD and confirm the problem.
And also check this tool File Bench 007a here:
http://www.computerbase.de/forum/showthread.php?t=1394418&page=10&p=16314483#post16314483
Board is german but the tool is in english.

Please do NOT use AS SSD, Crystal Disk Mark or some other crap that is either benching with fresh data or some low level benchmarking tools like HD tune or HD tach, which are NOT suitable for SSD testing!

 

[john3850]

AS SSD is NOT showing, if your SSD is degrading in read speeds of old files. It only benches with fresh data...you don't even know what the problem is, and post such a nonsense.

That was my mistake I also used SSD Read Speed Tester first then As-ssd.
Read speeds dropped to be 225 MB with SSD Read Speed Tester first.

Now on my 850evo read speeds are good with SSD Read Speed Tester.

 

[Alfhw]

Hi,
I'm looking to buy a Samsung 850 EVO 500 GB but I'm worried about degradation. Messages on this theread are old and mixed: some people say there is the problem, someone else no... Are there any update news about possible degradation in the 850 EVO like in the 840 EVO?

 

[omega3]

Hi,
I'm looking to buy a Samsung 850 EVO 500 GB but I'm worried about degradation. Messages on this theread are old and mixed: some people say there is the problem, someone else no... Are there any update news about possible degradation in the 850 EVO like in the 840 EVO?

Would also love to hear a year later from 850 EVO users if they still don't suffer from any degradation problems. Where there any updates for the drive btw?

 

[Glaring_Mistake]

Hi,
I'm looking to buy a Samsung 850 EVO 500 GB but I'm worried about degradation. Messages on this theread are old and mixed: some people say there is the problem, someone else no... Are there any update news about possible degradation in the 850 EVO like in the 840 EVO?

Would also love to hear a year later from 850 EVO users if they still don't suffer from any degradation problems. Where there any updates for the drive btw?

There's a pretty recent thread where I present some information on an 850 EVO after it's been in use for a year and another that has been powered down for five months (except when I've run tests on it).
Said thread can be found here: http://forums.anandtech.com/showthread.php?t=2479834

Also Allyn Malventano over at PCPer has said that he has been running tests on the 850 EVO since the beginning and has yet to notice any read speed slowdowns.