No. I cannot artificial environments and data to test my ssd. This is what was shown to me and i did my best to run the tests. I don't know if this is the my problem or the ssd. For me, as a normal user, this ssd works exactly as it was worked for me when i bought it. My windows and linux boots in exactly the same time and my applications are opening exactly the same time.
That result looks to be due to the drive having a number of small files (small files take longer to read) and something interfering with the benchmark.
Yes, you are right, first Adarzh's test was very weird with a point well over 900 MB/s... His second one seems more realistic.Also, if you look closely you can see that the read speed at two points exceeds 600MB/s which isn't possible for an 850 EVO.
Or that if these results were accurate that read speeds drop way faster (initially at least) for the 850 EVO than for any other drive.
Which would mean that there would be something seriously wrong with the construction of the 850 EVO itself (something that we have seen little indication of).
It would also mean that Samsung's 16nm 2D TLC NAND leaks less than the 40nm 3D TLC NAND despite that going by litography the latter should have a huge advantage.
8 months sample for 850 PRO 256GB - tsvBe interesting to see an 850 EVO vs 850 PRO read speed test on identically aged 6-12 month old data.
The Windows partition test certainly looks a little odd (may be due to running that test in Power Saving mode as he mentioned), but unless I'm missing something big, opening the Storage Partition tsv file in Excel, highlight the three columns, then custom sorting of "Sort by Age, then by size", shows the two slowest "0 day" files were 0.6MB (186.6MB/s) and 1.1MB (211.1MB.s), which were also the smallest. Every file over 4MB size was reading over 400MB/s and every file over 20MB size was reading over 500MB/s. Same for Day 1 - slowest speed (302.2MB/s) was also the smallest 1.1MB file. Everything 450-495MB/s was between 5-10MB in size. Everything larger than 10-15MB in side was reading over 495-550MB/s. Same is repeated day after day, with a long term trend of speeds start slowing down after a few months to the point where 250 days in speeds are only 1/2 to 3/5 vs day 0. By 370 days, even large +500MB files average only 180-300MB/s vs a solid +500-550MB/s Week 1. However, all the recently written data in the Storage Partition test seems fairly normal to me?BSim500, I think the fact that read speeds seem to be able to drop within 24 hours of the file being written indicates that the low read speeds are not due to files being read slower as they age but rather that there is something else affecting the results.
Thanks for posting that coercitiv. 850 PRO looks solid throughout. :thumbsup:8 months sample for 850 PRO 256GB - tsv
That would be great. Thanks. :thumbsup:I'll re-run the test on my SSD (which will be ~6 months then) and post the results sometime over the weekend.
Thanks for posting the data Coup27. Speeds look very good! According to the tsv file, the only "slower" files were the smaller sub 5MB ones (perfectly normal for all drives). No idea why Adarzh's results were so different though.Just ran the test again. I don't run RAPID or any other form of RAM caching. I rebooted and shut down/powered on and disabled real-time AV. This installation is now 7 months old.
Full Results
All looks good to me.
Thanks for posting the data Coup27. Speeds look very good! According to the tsv file, the only "slower" files were the smaller sub 5MB ones (perfectly normal for all drives). No idea why Adarzh's results were so different though.
I'm not sure that it's only due to small files. There is a test made by Coup27 here: http://forums.anandtech.com/showpost.php?p=38124005&postcount=15 All small files with readings around 500 MB/s like in the second Adarzh's test.
Coup27's test is about only 12 weeks old files with no drops in speed and Adarzh's second test start showing slower speeds after 21-24 weeks old files (5 months).
Also dvsv message:http://forums.anandtech.com/showpost.php?p=38060549&postcount=60 seems to confirm it.
Maybe the 850 Evo has a issue similar to the 840 Evo but only after 5-6 months and with a slower degradation?
What about Glaring_Mistake's test? He says files are 1 year old but in the picture I see only 21-24 weeks old files. Why? I don't get it.
The Windows partition test certainly looks a little odd (may be due to running that test in Power Saving mode as he mentioned), but unless I'm missing something big, opening the Storage Partition tsv file in Excel, highlight the three columns, then custom sorting of "Sort by Age, then by size", shows the two slowest "0 day" files were 0.6MB (186.6MB/s) and 1.1MB (211.1MB.s), which were also the smallest. Every file over 4MB size was reading over 400MB/s and every file over 20MB size was reading over 500MB/s. Same for Day 1 - slowest speed (302.2MB/s) was also the smallest 1.1MB file. Everything 450-495MB/s was between 5-10MB in size. Everything larger than 10-15MB in side was reading over 495-550MB/s. Same is repeated day after day, with a long term trend of speeds start slowing down after a few months to the point where 250 days in speeds are only 1/2 to 3/5 vs day 0. By 370 days, even large +500MB files average only 180-300MB/s vs a solid +500-550MB/s Week 1. However, all the recently written data in the Storage Partition test seems fairly normal to me?
I've had another look at his results.
Adarzh, you are running an older version of the 850 EVO firmware. If you look at both our screenshots you are running 01 and I am running 02. I recommend you do the following:
1. In Windows, install or update Samsung SSD Magician to the latest version, found here. Use Magician to update your SSD firmware. I have been running this firmware for months without issue.
2. Manually trim your Windows and storage partitions. If you are using Windows 7 you can use the Magician "Performance Optimization" feature to achieve this. If you are using 8 or 10, click start and search "optimizer". You can use this to manually trim the partition(s).
3. After trimming, leave your system to idle for a few hours (you can still use it).
4. Reboot. Then shut down for 1 minute and power back up. Ensure real-time AV is off along with RAPID / RAM caching and then test again.
Also, could you post your make and model of laptop?
Did you trim the partition also?
There is an interesting article here: https://tinkertry.com/samsung-850-evo-benchmarking-and-firmware-upgrading#comment-2789509186
Test made after 6 months seems good. But in the comments below the author tested again after 1 year and there is a serious slowdown between 0,5k and 8k files with ATTO. Also HD Tune shows a very low minimum transfer rate (18 vs 280 of 6 months before).
Actually no. Look at the first 2 screenshoots, both made 6 months after buying the ssd. First is made just before firmware update and second is made after firmware update and after ssd busy-time (where firmware or windows kept busy the ssd doing some sort of refresh or whatever). Second screenshot shows better Atto results with big increase in 0,5k - 8k. So HD-Tune.The performance dropped shortly after updating the firmware but after the drive was given some time to recover speeds were back to normal.
What would be so odd about that?
OK, I understand but even so looks weird to me that in dvsv post (http://forums.anandtech.com/showpost.php?p=38060549&postcount=60) speed went back to normal after a diskfresh.About HD Tune, read speeds can sometimes drop even when they shouldn't, especially in the beginning even when you're using the settings that give you the most consistent results which he doesn't appear to do.
Instead it looks like he is using the standard settings for HD Tune which uses 64KB blocks instead of the 8MB that you should have it set to in order to get consistent results and he has likely not configured it for accuracy rather than speed either.
Actually no. Look at the first 2 screenshoots, both made 6 months after buying the ssd. First is made just before firmware update and second is made after firmware update and after ssd busy-time (where firmware or windows kept busy the ssd doing some sort of refresh or whatever). Second screenshot shows better Atto results with big increase in 0,5k - 8k. So HD-Tune.
After 1 year (see comments below the article) he tested again and Atto results dropped a lot between 0,5 and 8k (screenshot https://a.disquscdn.com/uploads/mediaembed/images/3960/4715/original.jpg). For example 4k read dropped from 247 MB (6 months before) to 97 and 4k write from 216 to 120.
The author also updated the article and he's thinking to RMA the unit (see comments) and even freeing space didn't seem to help.
OK, I understand but even so looks weird to me that in dvsv post (http://forums.anandtech.com/showpost.php?p=38060549&postcount=60) speed went back to normal after a diskfresh.
I'm more confused every day...
I think you are right. Who knows the culprit is... :hmm:Yes, speeds did drop.
Not sure why that is.
PCPer did note some performance issues with the 2TB 850 EVO, maybe it could be connected to that.
Anyway while performance has decreased it is not due to files being old since both reads and writes are affected and of course there's also the fact that ATTO doesn't read any old files.
Ok.About the post that dvsv made I'll say once again that I've seen a drive with files days old that according to HD Tune had degraded read speeds and after running DiskFresh they were restored.
Running tests on that same drive now and months later it still shows no sign of degraded read speeds.
So, I think it is likely that they're just false positives like I got with that drive.
If you have them, then yes please. Still a lack of real data on 16nm MLC vs 40nm TLC, so the more the better! Thanks.BSim500, still interested in seeing some results for the MX100?
I don't mind it posted here if it's easier. It's entirely up to you though.Do you want me to post it here or do you want to keep results from different drives than the 850 EVO to a minimum?
So here it is:I don't mind it posted here if it's easier. It's entirely up to you though.
Is that unpowered? If so that's pretty good. I have the 512GB version of the MX100 in a HTPC. It's powered on daily and there's no real controlled testing, but even after almost 9 months Average Read Speed is 551MB/s. And before that reinstall, read speed was still roughly the same after a year. For all the talk of how Crucials are up to 100MB/s behind Samsung's on synthetics, the 16nm MLC drives would not slow down no matter how much time passed.So here it is:
Seems to be holding up pretty well.