Performance Consistency

We've been looking at performance consistency since the Intel SSD DC S3700 review in late 2012 and it has become one of the cornerstones of our SSD reviews. Back in the days many SSD vendors were only focusing on high peak performance, which unfortunately came at the cost of sustained performance. In other words, the drives would push high IOPS in certain synthetic scenarios to provide nice marketing numbers, but as soon as you pushed the drive for more than a few minutes you could easily run into hiccups caused by poor performance consistency. 

Once we started exploring IO consistency, nearly all SSD manufacturers made a move to improve consistency and for the 2015 suite, I haven't made any significant changes to the methodology we use to test IO consistency. The biggest change is the move from VDBench to Iometer 1.1.0 as the benchmarking software and I've also extended the test from 2000 seconds to a full hour to ensure that all drives hit steady-state during the test.

For better readability, I now provide bar graphs with the first one being an average IOPS of the last 400 seconds and the second graph displaying the standard deviation during the same period. Average IOPS provides a quick look into overall performance, but it can easily hide bad consistency, so looking at standard deviation is necessary for a complete look into consistency.

I'm still providing the same scatter graphs too, of course. However, I decided to dump the logarithmic graphs and go linear-only since logarithmic graphs aren't as accurate and can be hard to interpret for those who aren't familiar with them. I provide two graphs: one that includes the whole duration of the test and another that focuses on the last 400 seconds of the test to get a better scope into steady-state performance.

Steady-State 4KB Random Write Performance

For a mainstream drive, the 850 EVO mSATA/M.2 does relatively well in IO consistency except for the highest capacity 1TB model. Strangely enough the 2.5" 1TB 850 EVO does just fine, so this issue seems to be limited to the mSATA version. 

Steady-State 4KB Random Write Consistency

Looking at the standard deviation reveals why: the IO consistency of the 850 EVO mSATA 1TB, even with overprovisioning, is horrible compared to the rest of the 850 EVO lineup. 

Samsung 850 EVO mSATA 250GB
Default
25% Over-Provisioning

The issue with the 1TB mSATA is actually worse than I expected because it's literally stopping for seconds in a frequent manner. The pauses can even be over 50 seconds, so this isn't just some normal garbage collection that's happening in the background. I find this to be very alarming because it may have dramatic impact to user experience and it's simply something that no modern SSD should do anymore. I did let Samsung know about my findings before publishing this review, but I wasn't really able to get any comment from them regarding this issue and whether Samsung has noticed something similar in its internal tests. Adding over-provisioning seems to help as the pauses become much more infrequent, but for now I would still advise against buying the 1TB mSATA version until there's a fix for the IO consistency.

As for the other capacities, the 850 EVO has excellent consistency and steady-state performance for a mainstream drive. The capacity has some effect on performance, but even the 250GB model has roughly twice the performance of 240GB Ultra II thanks to faster V-NAND.

Samsung 850 EVO mSATA 250GB
Default
25% Over-Provisioning
Introduction, The Drives & The Test AnandTech Storage Bench - The Destroyer
Comments Locked

58 Comments

View All Comments

  • WackyDan - Saturday, April 4, 2015 - link

    Eh... Nope. These won't work in the ThinkPads.
  • blanarahul - Tuesday, March 31, 2015 - link

    Hopefully, this drive will put an end to the mSATA/M.2 versions of the 840 non-EVO that companies are shipping in their laptops.
  • Samus - Tuesday, March 31, 2015 - link

    DoH! I just paid $20 more for the M550 (basically the MX100) 512GB M2 drive, which is double-sided and needs to be crammed into my laptop :\
  • kgh00007 - Tuesday, March 31, 2015 - link

    Hey, is there any chance you could fill one of these m.2 drives with data, power it off for a month then check the read speeds when you power it back on?

    I need some sort of evidence that this TLC V-nand does not have the same issues as the TLC in the 840 EVO.
    Otherwise I'm not going to be able to buy one of these!

    Cheers!
  • goodyes - Wednesday, April 1, 2015 - link

    What the hell does not the charts compares with 840 ev msata ? Bad numbers Why not true sequential test ??
    And this site posting SUCKS, Why in the world I cannot use my facebook or so login, Did you know about disquss ? Get out the absurd captcha that fucking me 30 times or so trying to get a magical potion to know what the words saying,. THIS IS NOT ADVANCE IN TECH THIS IS TRASH
  • Kristian Vättö - Wednesday, April 1, 2015 - link

    This is the third review with the new 2015 SSD Suite and I've only had limited time to test drives, hence the lack of 840 EVO in the graphs.
  • goodyes - Wednesday, April 1, 2015 - link

    Ya, but results that I have a 1TB 840 msata and write sequential at more than 500MBps around 520max MBps, and now I see than new 850 msata tops at ?? 480MBps ?? cannot be possible what my eyes look, AND WHY THE HELL NO ONE REVIEWER COMPARE With olders 840 msata, so I must think that all of you guys are a paid reviewers and you get money from samsung, if not, YOU MUST compare to older model
  • cgorange - Wednesday, April 1, 2015 - link

    Other than providing samples, I can assure you that Samsung doesn't pay Anandtech to review its products
  • Ekitrak - Wednesday, April 1, 2015 - link

    The Final Words page has 2 entries of "Samsung 840 EVO mSATA" on the Amazon Price Comparison. I'm guessing this is an error and one of them is supposed to be the Sata III version?
  • Kristian Vättö - Wednesday, April 1, 2015 - link

    I'm not seeing this -- maybe you accidentally mixed up the 850 EVO mSATA and 840 EVO mSATA as they are both in the table (or maybe this was already fixed by another editor).

Log in

Don't have an account? Sign up now