The Samsung 960 Pro (2TB) SSD Review
by Billy Tallis on October 18, 2016 10:00 AM ESTA Note About Drivers
The Intel SSD 750, the Samsung 950 Pro and the OCZ RD400 were all reviewed with the NVMe drivers supplied by the SSD vendors. In the past, vendors have sometimes cited performance as an advantage to using their NVMe driver over the one built in to Windows, but the primary reason has been that Microsoft's driver implements a limited feature set. The driver that was made available as an update to add NVMe support to Windows 7 SP1 and Windows Server 2008 SP1 did not include the necessary interfaces for updating SSD firmware, and even on Windows 8.1 and later the vendor-specific management tools require their own driver for performing tasks like a secure erase.
Samsung's NVMe driver for the 960 Pro was not ready in time for this review. They are planning to release it in mid-November in conjunction with their Magician 5.0 utility. The Samsung NVMe driver will be required to support Magician 5.0's new "Magic Vault" secure archive/backup feature and the new secure file erase feature.
In the meantime, rather than try to hack Samsung's NVMe driver for the 950 Pro to work with the 960 Pro, this review is relying on Microsoft's NVMe driver built-in to Windows 8.1. While most SSD vendors (especially the smaller ones) now say that Microsoft's NVMe driver offers adequate performance and that there is no need for a custom driver to get full performance, there are some pitfalls.
Windows provides two settings for drive write caching policy. By default, write caching is enabled on internal drives and there is an unselected option to turn off write cache buffer flushing. Both options have warnings attached about the possibility of data loss in the event of a power failure. It is normal for SSDs to cache and combine writes rather than immediately send all written data straight to the flash, and this is necessary to overcome the fact that NAND flash write operations are inherently much slower than read operations. Without write caching on the SSD, we would never see good random write performance, let alone random write performance that exceeds random read performance.
The default write caching policy settings work fine for SATA SSDs. This is not the case for NVMe SSDs when using Microsoft's driver. Microsoft's NVMe driver in the default configuration is extremely conservative about write caching, leading to extremely poor performance on some tests. Checking the second box gives performance that is as expected while leaving it unchecked for a high-end NVMe drive can lead to worse performance than a low-end SATA drive. Normally I would not review a drive with an obscure setting like this changed, especially since it can increase the risk of data loss, but Microsoft's default is clearly broken and not in line with the industry standard practices. The 960 Pro was benchmarked with the settings as shown above, and a more thorough comparison of how NVMe drivers and operating system versions affect performance will be coming in the future.
Facebook
Twitter
RSS
72 Comments
View All Comments
Billy Tallis - Tuesday, October 18, 2016 - link
The Intel 600p is very high on my to-do list. Since it's also using the Microsoft NVMe driver I want to run it through some more tests before publishing the review, but it should be done before I get the 960 EVO.WatcherCK - Tuesday, October 18, 2016 - link
Curious about the operating temperatures of the drive, Im guessing that Destroyer gave it a good thermal workout :) Puget Systems have done some testing on the effect of adding cooling to M2 SSDs:https://www.pugetsystems.com/labs/articles/Samsung...
Aquacomputer also make a PCIe Riser card equipped waterblock for M2 drives for those who want to water cool their SSD, the kryoM.2
Does anyone know if moving an M2 drive to a riser card improves cooling for the drive? I would think there is a little bit of improvement bringing the drive off the surface of the motherboard... How if at all is everyone keeping their motherboard mounted SSDs cool?
BurntMyBacon - Wednesday, October 19, 2016 - link
@WatcherCKYes, moving an M.2 drive to a riser card improves cooling (at least in my system). I checked temperatures mounted on my motherboard with a spot cooler directly on the 950Pro. I then compared it against mounting on a Silverstone ECM20 and an Asus Hyper M.2 X4 Mini (no spot cooler). If you are able to direct good airflow to the riser card, as I was, the Asus riser card is significantly cooler even though it had no spot cooler. If you use a thermal pad on the back of the M.2 drive on the Silverstone EMC20, it may in fact do even better than the Asus, but it my system, the larger standoffs allowed more air to flow behind the M.2 drive on the Asus solution. In any case, I can't seem to induce thermal throttling on my 950Pro in my setup regardless of how hard I try. Again, it will depend on how much airflow you can deliver it.
Gradius2 - Tuesday, October 18, 2016 - link
Those SSDs suffering from HEAT. If you don't put a forced cooling on them, it will DEGRADE the speed no matter what! This is why you won't see much than 3GB/s !iwod - Wednesday, October 19, 2016 - link
Where are these performance heading. Where Do we need to go. The destroyers and heavy definitely dont represent 95% of the consumer users usage. Do we need more higher Random Q1 Read Write? Or do we need higher Seq Read Write? If so then we dont need our firmware and powerful CPU core for QD32. Will that save cost? Why are we still within the 5W power usage? When can we get those dropped down to 2W or less.zodiacfml - Wednesday, October 19, 2016 - link
blazing fast. i will not be able to make useod this speed except for W10 upgradesprofdre - Monday, October 24, 2016 - link
@ Billy Talis: Would it be possible to test on another mainboard? There seems to be a clear bandwidth issue for sequential transfers, as other tech sites as https://www.computerbase.de/2016-10/samsung-ssd-96... achieved almost 3400 MB/s in CrystalDiskMark for the 512 GB model. At first they struggled to reach more than 3100 MB/s. (Samsung values were achieved with CrystalDiskMark according to computerbase).calbear88 - Tuesday, November 15, 2016 - link
Great table in the article summarizing the different Samsung NAND technologies. Here's a summary of the different types of NAND and which products they were in.27nm MLC 830 PM830
21nm MLC 840 pro
21nm TLC 840
19nm MLC XP941
19nm TLC 840 Evo
16nm MLC SM951
16nm TLC 750 Evo PM951
32 layer 86bit v-NAND MLC 850 Pro
32 layer 128bit v-NAND TLC 850 Evo
32 layer 128bit v-NAND MLC 950 Pro
48 layer 128bit v-NAND MLC 960 Pro SM961
48 layer 128bit v-NAND TLC 960 Evo PM961
Meteor2 - Wednesday, November 23, 2016 - link
How is new dies and an entirely new controller 'just a generational refresh of the 950 Pro', from any angle?anaconda1 - Monday, February 13, 2017 - link
The Samsung 2.1 driver has DPC latency issues on my end. I am usign a Samsung 960 500GB Evo nvme and using it with the Native Microsoft Driver (Windows 10 with latest updates installed) or the Samsung 2.0 driver, all is ok, DPC latency reports are fine - Latencymon and latency checker both report trouble free operation. However, installing the 2.1 driver, both Latencymon and dpc latency checker report problems with storport.sys and problems with DPC. Rolling back the driver from the Samsung 2.1 to the Microsoft or the Samsung 2.0, all goes back to normal