The Samsung 960 EVO (1TB) Review
by Billy Tallis on November 15, 2016 10:00 AM ESTPerformance Consistency
Our performance consistency test explores the extent to which a drive can reliably sustain performance during a long-duration random write test. Specifications for consumer drives typically list peak performance numbers only attainable in ideal conditions. The performance in a worst-case scenario can be drastically different as over the course of a long test drives can run out of spare area, have to start performing garbage collection, and sometimes even reach power or thermal limits.
In addition to an overall decline in performance, a long test can show patterns in how performance varies on shorter timescales. Some drives will exhibit very little variance in performance from second to second, while others will show massive drops in performance during each garbage collection cycle but otherwise maintain good performance, and others show constantly wide variance. If a drive periodically slows to hard drive levels of performance, it may feel slow to use even if its overall average performance is very high.
To maximally stress the drive's controller and force it to perform garbage collection and wear leveling, this test conducts 4kB random writes with a queue depth of 32. The drive is filled before the start of the test, and the test duration is one hour. Any spare area will be exhausted early in the test and by the end of the hour even the largest drives with the most overprovisioning will have reached a steady state. We use the last 400 seconds of the test to score the drive both on steady-state average writes per second and on its performance divided by the standard deviation.
The 960 EVO's steady state random write speed is not quite as fast as the 960 Pro, but it's in the same league and much faster than most consumer SSDs.
The 960 EVO sets a new record for combining high performance with consistency. It's a bit slower than the 960 Pro, but less variable.
 |
|||||||||
| Default | |||||||||
| 25% Over-Provisioning | |||||||||
Highly consistent performance is a good thing, but it makes for a boring graph. The transitions from peak to sustained performance modes look the same for both the 960 Pro and the 960 EVO.
 |
|||||||||
| Default | |||||||||
| 25% Over-Provisioning | |||||||||
The 960 EVO responds to extra overprovisioning with even more consistent (and high) performance.
Facebook
Twitter
RSS
87 Comments
View All Comments
Foralin - Tuesday, November 15, 2016 - link
I'd like to see this kind of analisys for the new Macbook Pro's SSDphilehidiot - Tuesday, November 15, 2016 - link
I think that often Apple use a couple of different suppliers for their SSDs (certainly was the case when I bought my Air ages ago) and they're unlikely to hand out samples for testing as if there's one thing Apple seems to hate, it's scrutiny. This means that you might have to buy quite a few Macbooks, ID the SSD and then you'd still never know if they were using one, two, three or even four different suppliers unless you got loads of people to run the appropriate software and then went on a shopping spree. Hoping of course that you could return those you've unpacked, set up, tested and carefully repackaged.... Whilst it'd be nice, Apple don't make it easy and unless you're loaded it's not going to be practical.repoman27 - Tuesday, November 15, 2016 - link
Apple sourced SSDs from Samsung, SanDisk and Toshiba back when they used SATA SSDs, but went 100% Samsung when they switched to PCIe. The 2015 MBPs were all SM951, for instance. From what I've seen thus far, the 2016 MBPs use a new, in-house designed PCIe 3.0 x4 NVMe controller paired with SanDisk NAND.repoman27 - Tuesday, November 15, 2016 - link
And I take that back that last bit because I just saw a post with a photo of the internals of the MBP w/ TouchBar and it looked to have a Samsung SSD on board.Threska - Tuesday, November 15, 2016 - link
One disadvantage I see of the M.2 form-factor is inadequate cooling on some motherboards, compared to their more traditional SSD brethren.willis936 - Tuesday, November 15, 2016 - link
There's a quick fix for that: an ugly PCIe adapter with a heatsink. Or actually slapping some RAM heatsinks on the drive itself. I've been looking for a 2x M.2 to PCIe x8 adapter. The only ones I've found are expensive server adapters. Considering one of these drives nearly saturates 4 PCIe 3.0 lanes it seems that a regular consumer who wants to do RAID 0 should run their GPU in x8 (or go all out on HEDT) and get two PCIe adapters with heatsinks.TheinsanegamerN - Tuesday, November 15, 2016 - link
The issue is that is only possible on desktops. Laptops are more SOL in this regard.willis936 - Tuesday, November 15, 2016 - link
More performance = more power. It would be neat if they made different power profiles that could be set by the user through the OS. I don't want 5W pulled from my laptop just for my SSD to read 2 GB/s but I also don't need it to run that quickly.MajGenRelativity - Tuesday, November 15, 2016 - link
That's a nifty idea! I would like that too :)Billy Tallis - Tuesday, November 15, 2016 - link
NVMe already has that feature. Drives can define multiple power states, both operational and non-operational idle. The definition of those power states can include information about the relative performance impact on read and write throughput and latency, and how long it takes to enter and leave the different idle power states. For example, the 960 Pro declares a full-power operational power state with maximum power draw of up to 6.9W, and restricted operational power states with limits of 5.5W and 5.1W. It also declares two non-operational idle power states with limits of 0.05W and 0.008W, which my measurements have haven't accurately captured.Making full use of this capability requires better support on the software side.