Plextor Introduces M6S Plus SSDs with Toshiba’s 15nm MLC NANDby Anton Shilov on February 10, 2016 9:00 AM EST
Plextor this week introduced its M6S Plus SSDs that use Toshiba’s latest MLC NAND flash memory as well as a proven controller from Marvell. The new SSDs offer the same level of performance and the same feature set as their predecessors, the M6S drives introduced in 2014. The drives offer read and write speeds expected from SATA SSDs as well as Plextor’s technologies designed to guarantee high reliability and high performance after long periods of usage.
The new Plextor M6S Plus SSDs are based on the Marvell 88SS9188 controller as well as Toshiba’s newest 15 nm MLC NAND. Earlier Plextor offered M6S solid-state drives based on the same controller chip, but featured Toshiba’s older 19 nm MLC NAND. Apparently, Plextor will have to switch to a new type of NAND as manufacturers tend to decrease production of memory using outdated process technologies. The new drives are offered in 2.5-inch/7 mm form-factor and use the SATA-6 Gb/s interface.
The Plextor M6S Plus SSDs feature 128 GB, 256 GB and 512 GB capacities. They are rated to offer maximum sequential read/write speed of up to 520/440MB/s. Maximum random 4K read/write speed declared by Plextor is 94K/80K IOPS (input/output operations per second). It is noteworthy that to guarantee high performance, the new SSDs from Plextor are equipped with up to 768 MB of DDR3 DRAM-based cache.
|Specifications of Plextor M6S Plus SSDs|
|Capacity||128 GB||256 GB||512 GB|
|NAND||Toshiba, MLC 15nm|
|Cache||256 MB DDR3||512 MB DDR3||768 MB DDR3|
|Sequential Read Speed||up to 520 MB/s||up to 520 MB/s||up to 520 MB/s|
|Sequential Write Speed||up to 300 MB/s||up to 420 MB/s||up to 440 MB/s|
|4K Random Read||up to 88K IOPS||up to 90K IOPS||up to 94K IOPS|
|4K Random Write||up to 75K IOPS||up to 80K IOPS||up to 80K IOPS|
Notably, the performance specifications for the new M6S Plus lineup is identical to that of the previous M6S series. As Plextor didn't switch controllers and didn't switch NAND vendors, this means that they most likely are just using 15nm NAND as a drop-in replacement for their existing designs, and there won’t be any generational performance changes such as using fewer, higher capacity NAND dies. This has been an issue in the past when vendors have switched to newer generation NAND and labeled the resulting product under an old brand, so kudos to Plextor for making the 15nm drives their own line anyhow.
However while the switch to 15nm NAND doesn't impact performance, it's unclear whether the endurance of the newer M6S Plus series has been impacted, as the company hasn't published any endurance figures for the new drives. Typically, MLC NAND produced using thinner fabrication process has lower P/E cycle rating compared to flash memory produced using thicker manufacturing technology. The 19nm M6S series was rated for 72TB, so it will be interesting to see where the M6S Plus ends up.
Moving on, contemporary Plextor’s SSDs also ship with the company’s PlexTurbo RAM caching technology, which is aimed to improve performance beyond limitations of SATA-6Gb/s interface. The caching works like RAM disk, hence, uses system memory. The M6S Plus also supports TrueSpeed (supposed to guarantee maximum performance over long usage periods) and TrueProtect (a multi-layer error correction capability) firmware-based technologies developed by the company.
Finally, exact MSRPs of the Plextor M6S Plus drives are unknown. However, Amazon plans to start selling the 128 GB version for $62, the 256 GB model for $91 and the 512 GB flavour for $166 shortly.
The Plextor M6S Plus SSDs appear to be pretty affordable, in fact, even more affordable than their predecessors. Apparently, the recent price drops of NAND flash helped Plextor to build inexpensive SSDs based on 15 nm MLC NAND flash from Toshiba, which is not a bad choice, considering the fact that the company managed to maintain performance numbers of the M6S. By contrast, many of Plextor’s rivals are using TLC NAND to make cheap SSDs.
Post Your CommentPlease log in or sign up to comment.
View All Comments
BurntMyBacon - Monday, February 15, 2016 - link@Kristian Vättö: "It wouldn't be 1.5GB/s in real life, unless the spec was significantly changed ... SATA IV, at 12Gbps raw transfer rate, would be capable of about 1GB/s ..."
I does seem likely that the 8b/10b encoding scheme will persist through the next iteration of SATA given that SAS-3 12Gbps is already available and uses it. This would put a theoretical cap of 1.25GB/s performance. I'm not sure if other overhead scales linearly or even at all, so it is harder for me to say exactly where I would expect the practical limits to be. SAS-4 is apparently moving to a 128b/150b scheme, so that would be the alternate best guess, but I wouldn't expect it. It would only bring the theoretical cap up to 1.28GB/s anyways.
That all said, you can only fit a limited number of M.2 sticks on a motherboard. I'm sure some people wouldn't mind sacrificing PCIe slots, but others will want them for GPUs or other add-on cards. I don't see either of these supplanting SATA for volume use, though I expect it for laptops and perhaps small form factor devices. I would consider U.2 (SFF-8639) as the logical successor to SATA if only it would start picking up a little traction in the market. That said, SATA express doesn't seem to be getting any traction either, so who knows if an SATA-4 will do any better. With so much of the market switching to mobile and small form factor devices that can't support many storage devices, it's possible that SATA gets replaced with SAS.
I would argue that, if they are going to release another SATA standard, they need to do it quickly to maintain relevance. Otherwise it will be entirely supplanted by U.2 or SAS. Mobile and small form factor devices are likely already lost to the smaller profiles storage devices like M.2. Of course, there are always spinning disks, but SATA-3 seems to be well more than capable of handling them and a new SATA standard would just push cost up unnecessarily.
Kristian Vättö - Tuesday, February 16, 2016 - linkRemember that in order to utilize more than three SATA IV (assuming 1GB/s) at full speed, the DMI interface between the CPU and PCH would also need to be upgraded. DMI 3.0 is essentially just four PCIe 3.0 lanes, so without PCIe 4.0 there would have to be eight 3.0 lanes dedicated to DMI, which would be away from the other PCIe lanes (or increased die complexity and size). In other words, you would still be sacrificing PCIe lanes with SATA IV, and on the downside you would need three drives to match the performance of one PCIe 3.0 x4 SSD.
As you said, SATA III is still more than fine for volume use. The applications that require or benefit from multiple GB/s volume storage are quite rare, and the users likely have a workstation anyway with plenty of PCIe lanes for PCIe SSDs.
boeush - Wednesday, February 10, 2016 - linkPCIe and NVMe are fine if you only want one or two drives of relatively lower capacity and thermally limited performance in your system.
If you wanted to build a PB-class RAID-10 blazing-fast monster, then something like SATA IV might be the only reasonable way...
dakishimesan - Thursday, February 11, 2016 - linkI was lucky enough to pick up a Samsung 950 Pro M.2 for xmas last year, and I can tell you from hitting it hard that thermal throttling is not an issue, and it's one of the fastest components I've ever owned (in terms of jump from my last drive, and relative to the other parts of a modern skylake system). But I hear you on the RAID -- I don't think the Intel RS controller even supports raid on PCIe/NVMe devices, even for those motherboards that have 2+ M.2 ports.
Kristian Vättö - Thursday, February 11, 2016 - linkSkylake and Z170 chipset brought RAID support for PCIe SSDs.
jabber - Saturday, February 13, 2016 - linkExactly, if you want to run more than say three drives then PCIe etc. isn't going to cut it. SATA IV will have to come.
BurntMyBacon - Monday, February 15, 2016 - link@jabber: "Exactly, if you want to run more than say three drives then PCIe etc. isn't going to cut it. SATA IV will have to come."
Rather than wait on a standard that may or may not be coming, you could use SAS (a bit expensive I know), or U.2 (hopefully more players come to the market soon), or you could ask yourself if your third or more drive and really needs performance beyond SATA-3. Magnetic storage certainly doesn't and a Samsung 850Pro (and many competitors) put up a pretty decent showing of performance. While none of these solutions are perfect, they are all available today. On top of that, they downside of U.2 is the same as SATA-4 (availability) and it is both faster (if SATA-4 holds to the past upgrade pattern) and has a head start on SATA-4.
BurntMyBacon - Monday, February 15, 2016 - link@boeush: "If you wanted to build a PB-class RAID-10 blazing-fast monster, then something like SATA IV might be the only reasonable way..."
If they keep with the same upgrade pattern, SATA-4 would clock in at 12Gbps (1.25GB/s with encoding scheme). PCIe 3.0 x4 runs 32Gbps (3.94GB/s with encoding scheme). With two M.2 (4 lane) slots, you can run RAID-1. I would argue that you are better off running one M.2 drive than running two SATA-4 devices in RAID-0 under the above assumptions as you are getting less performance and more points of failure. Therefore, an M.2 RAID-1 setup is not only another reasonable option, but even more of a "blazing-fast monster" than a theoretical SATA-4 RAID-10 setup (and more reliable).
BurntMyBacon - Monday, February 15, 2016 - linkSide Note: Spinning disk is currently the way to go for capacity, but you certainly won't be blazing fast regardless of your setup. Also, business and data centers will probably continue to use SAS over SATA, so business class Network Attached Storage or Storage Area Networks are out of the question.
boeush - Wednesday, February 10, 2016 - linkHeh, I spoke too soon. Checking Wikipedia, SATA 3.2 standard provides 16 Gbit/s (1.97 GB/s) and was released in 2013. Just a matter of people actually implementing the standard...