Performance Consistency

Performance consistency tells us a lot about the architecture of these SSDs and how they handle internal fragmentation. The reason we do not have consistent IO latency with SSDs is because inevitably all controllers have to do some amount of defragmentation or garbage collection in order to continue operating at high speeds. When and how an SSD decides to run its defrag or cleanup routines directly impacts the user experience as inconsistent performance results in application slowdowns.

To test IO consistency, we fill a secure erased SSD with sequential data to ensure that all user accessible LBAs (Logical Block Addresses) have data associated with them. Next we kick off a 4KB random write workload across all LBAs at a queue depth of 32 using incompressible data. The test is run for just over half an hour and we record instantaneous IOPS every second.

We are also testing drives with added over-provisioning by limiting the LBA range. This gives us a look into the drive’s behavior with varying levels of empty space, which is frankly a more realistic approach for client workloads.

Each of the three graphs has its own purpose. The first one is of the whole duration of the test in log scale. The second and third one zoom into the beginning of steady-state operation (t=1400s) but on different scales: the second one uses log scale for easy comparison whereas the third one uses linear scale for better visualization of differences between drives. Click the dropdown selections below each graph to switch the source data.

For more detailed description of the test and why performance consistency matters, read our original Intel SSD DC S3700 article.

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

The 850 EVO presents a healthy increase in IO consistency. The 840 EVO wasn't exactly inconsistent in the first place, but the 850 EVO takes the steady-state IOPS from ~3,000-5,000 IOPS to 5,000-8,000 IOPS, which is actually nearly on par with the 850 Pro. The 850 EVO has without a doubt one of the highest performance consistencies out of the value/mainstream drives we have tested.

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

 

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


TRIM Validation

To test TRIM, I filled a 120GB 850 EVO with sequential 128KB data and proceeded with a 30-minute random 4KB write (QD32) workload to put the drive into steady-state. After that I TRIM'ed the drive by issuing a quick format in Windows and ran HD Tach to produce the graph below.

And TRIM works as expected.

Endurance: Close to Planar MLC NAND AnandTech Storage Bench 2013
Comments Locked

97 Comments

View All Comments

  • R3MF - Tuesday, December 9, 2014 - link

    my mistake, i had presumed that the Pro was not a consumer part.

    still, six months on with the arrival of tons of X99 and Z97 boards sporting m.2 slots, and the drives based on the marvell controller just months away, i'd have thought it would merit a mention.
  • hojnikb - Tuesday, December 9, 2014 - link

    There is a separate article adressing this....
  • cm2187 - Tuesday, December 9, 2014 - link

    But out of curiosity, what are you going to do with the extra performance? Who actually has any use specs higher than what the EVO already offers. It is certainly the case on some heavy load server but for end users, even enthusiasts like me, I am not sure I would get an even slightly better experience by beating the SATA 3 specs.
  • R3MF - Tuesday, December 9, 2014 - link

    I presume that SATA express and m.2 were invented for no reason then?

    Bandwidth is useful, as is lower latency.
  • cm2187 - Wednesday, December 10, 2014 - link

    Well, it's not because it is invented that it is useful. It's like having dozen of cores in a CPU. Some applications will have some use for these cores (certainly relevant on servers or for virtualization). But the vast majority of common applications are single threaded so people should rather focus on higher clock rates. I'm always happy to see higher specs but I just wonder which of my application will be faster with M2.
  • Supercell99 - Sunday, December 14, 2014 - link

    Virtualization. I run VMware with several OS's running at the same time on my desktop. Being able to startup and have these run off a low latency disk is nice. Power users always have a need for high bandwidth, low latency I/O.
  • hojnikb - Wednesday, December 10, 2014 - link

    These can be used for PCI-E or SATA protoco. In fact, most m.2 drives run sata instead of pci-e
  • dcaxax - Tuesday, December 9, 2014 - link

    I'm unconvinced by Samsung. My SSD 830 is doing ok, having suffered an acceptable 30% performance decline (which may be correctable via secure erase but I will not test this).

    But my 840 (non-EVO) which works in my HTPC and sees limited use outside of the hibernation file (60 of the drive is empty by the way) is now running more than 60-70% slower.

    This is just unacceptable in a system which supports trim (win 7 x64). Samsung have done nothing to rectify this, claiming these problems occur on their EVO line. Until they change their approach, I'm inclined to distrust their latest cost-saving "innovation" and give my money to crucial instead.
  • simonpschmitt - Tuesday, December 9, 2014 - link

    Dear Mr. Vättö,
    while I don't think it would necessary belong into this article R3MF has a point. What is the state of SATAExpress, NGFF, m.2, ... currently? My laptop is 18 Month old and has an unuses m.2-slot witch, to my knowledge, nobody ever put an SSD into. You seem to have an ear to the ground when it comes to the SSD-Industry. There are a few questions you might have a qualified opinion about:
    - Will we be seeing current-gen (meaning 850 EVO-gen) m.2/SATAExpress SSDs?
    - If yes, up to wich capacity in wich form factor (2242, 2280, ...)?
    - with regards to m.2: Will there be mainly PCIe (2 lanes/4lanes) drives or SATA?
    - When do you suppose these will be an economically viable alternative to 2.5" given both slots are available?

    In your personal opinion:
    - What is the point of SATAExpress when literally every SATAExpress-Device also has m.2?
    - Will there be a subjective improvement for the normal or enthusiast (non Datacenter) user with the switch to PCIe?

    A quick blurb, perhaps in the form of a short pipeline aticle, would be much appreciated.

    An other thing I always wondered about: While I am amazed with the percieved benefits of an SSD vs. an HDD game-load-times often seem not to change at all. It's more of an oddity than a real concern but my new system (i5 4200, 8GB, 840 EVO) often has the same load times than my old system (i3 330, 4GB, HDD). I always thought load times were mainly dependent on how fast the data can be read (HDD/SDD bound) and how fast it can be processed/extracted (CPU bound). Is there a factor I'm missing or do games just not take advantage of certain kinds of faster hardware.

    Perhaps you or some of the other readers can help me with my curiosity.

    Thanks, Simon
  • metayoshi - Tuesday, December 9, 2014 - link

    I can't answer all of your questions, but I can answer the gaming part.

    With regards to gaming, it really depends on the game. Many games these days are relatively optimized on loading, so running them on an HDD or SSD doesn't matter too much since they like to load parts of the game in the background. However, there are some games where having an SSD is completely noticeable. As an avid World of Warcraft player, I can tell which raid members have SSDs and which don't because those of us with SSDs simply appear in the raid much faster than those on HDDs when switching zones. I mean, it definitely doesn't hinder gameplay too much since the only thing that takes a while is actually getting into the zone. The rest of the zone is in RAM already, so getting to and fighting bosses are instantaneous. I used to have a 7200 RPM drive before too. For me, the difference is completely noticeable now that I have had an SSD for a couple of years now.

Log in

Don't have an account? Sign up now