We're enthusiasts. Whether we need more speed or not, fast hardware is attractive to us. And we're happy to explain to friends and family that they don't need to replace all of their mechanical disks with solid-state storage, but they should be booting up and launching their favorite applications from an SSD.
Yes, hard drives will continue serving as high-capacity user data repositories for years to come. But the rate at which they evolve is slow. Meanwhile, solid-state technologies push forward rapidly, pushing the limits of interfaces as soon as they're made available. It's not fair to keep SSDs married to SATA and the AHCI protocol. We don't even need traditional form factors. The solid-state nature of these devices makes them flexible and rugged. In the future, storage is going to look unlike anything around today.

The advancements are going to hit incrementally, though. Take the M.2 form factor, for example, which we previously introduced in SanDisk A110 PCIe SSD: Armed With The New M.2 Edge Connector. Here we are, eight months later, and we're just getting our hands on a second M.2-based device, Plextor's new M6e PCI Express SSD (which also goes by the model number PX-AG256M6E in 256 GB trim). Like SanDisk's drive, this is a natively-PCIe device with a PCIe-based controller that continues to rely on AHCI.
One day, it's probable that AHCI will go the way of Iomega's Zip disk. But for now it remains a viable standard for ensuring the interoperability of drives and controllers. It doesn't require proprietary drivers, since Windows, Linux, BSD, and OS X all support the interface. That also means important features like the TRIM command are enabled as well.
And so the M6e is a stepping stone on the way to a more SSD-optimized storage ecosystem. This hybrid configuration keeps progress moving forward without rocking the boat, thanks to its off-the-shelf Marvell 88SS9183 native PCIe controller, along with Plextor's familiarity with developing firmware for Marvell's processors. The M6e is basically a 22 mm-wide, 80 mm-long M.2-compatible SSD on two lanes of PCI Express. Our sample shipped on a half-height PCIe x4 adapter that makes it easy to drop into any PCIe-equipped desktop, even if it isn't M.2-enabled. As the ecosystem supporting M.2 grows, the M6e will eventually become available without the extra card.
This drive's real attraction is two-fold. First, it's the first readily-available M.2 PCIe-capable device we've seen. And second, there's the compelling performance story. Without SATA 6Gb/s limiting throughput, we're ready to break some speed records.

What about that SanDisk A110 review we were just talking about? As an OEM-only solution, you're going to have a hard time finding one without a lock pick, crowbar, and threat of jail time. A few specialized vendors might post it for resale, but it doesn't look like the A110 is available today.
Then again, it's not like there's a lot of demand for the M.2 form factor...yet. I'm guessing you'll see a lot more of it once Z97-based motherboards start surfacing in the very near future. And so Plextor is dropping the flag on enthusiast-oriented storage for M.2 ahead of the rush. It'll have competition eventually. But getting out in front with what it hopes is a solid product could earn the company some serious power user credibility.
That adapter card means you don't have to wait to try the M6e, though. It's available today, and it should work on any motherboard capable of second-gen PCIe signaling.

There are three versions of the M6e: one with 128 GB of capacity, a 256 GB model, and a 512 GB flagship. The first two are already on Newegg for $180 and $300, respectively. They'll apparently be available with and without the adapter card, though right now both retail packages include it. Before you grab one and rip off the M.2 module to add to your cutting-edge motherboard, bear in mind that voids your warranty. Just pick the version you need deliberately and you'll be fine.
| Products |
Plextor M6e 128GB
|
Plextor M6e 256GB
|
Plextor M6e 512GB
|
| Pricing |
|
|
|
| Controller | Marvell 88SS9183-BNP2 Native PCIe 2.0 Controller | Marvell 88SS9183-BNP2 Native PCIe 2.0 Controller | Marvell 88SS9183-BNP2 Native PCIe 2.0 Controller |
| NAND | 19 nm Toshiba Type A Toggle-mode, 64 Gb dies | 19 nm Toshiba Type A Toggle-mode, 64 Gb dies | 19 nm Toshiba Type A Toggle-mode, 64 Gb dies |
| Seq Read and Write (in MB/s) | 770 / 335 | 770 / 580 | 770 / 625 |
| Rand Read Write 4 KB IOps | 96,000 / 83,000 | 105,000 / 100,000 | 105,000 / 100,000 |
| Die Count | 16 | 32 | 64 |
| Warranty | Five-year | Five-year | Five-year |
See that warranty? Five years across the entire family. Plextor and Intel were ahead of the pack offering such lengthy guarantees on drives with sub-3x nm flash. And then there's the pricing, which reflects what you'll pay right now if you buy a drive on Newegg (at least for the smaller two capacities). We can only hope that post-launch pricing drops closer to $1/GB once Newegg no longer sells the drive exclusively.
We know that Plextor is planning a 1 TB SATA-based SSD. However, it won't employ the same Toggle-mode flash. It'll instead utilize Toshiba's new A19 (A for advanced) NAND, probably in 128 Gb densities. The M6e uses more performance-oriented memory from Toshiba, the same 3000 P/E-cycle stuff that rode aboard the M5 Pro. As a result, achieving 1 TB could be tricky on the M6e. As a performance boot drive and application launchpad, however, 128 to 512 GB should be ample.

There isn't any point in taking the M6e apart. It's basically naked right out of the box. But there are a few components that bear closer inspection. Let's start with those.
Disconnected from the adapter card, Plextor's M6e is downright diminutive. Again, this is a 22 mm-wide, 80 mm-long device, which we expect to become a fairly common side for M.2-based SSDs moving forward.


In fact, unless you're reading this on a smartphone or tablet, you're probably seeing the front and back PCB shots larger than they actually are. M.2 storage can be 12, 16, 22, and 30 mm-wide, though most of what we've seen thus far conforms to the 22 mm standard, easily accommodating the width of NAND packages. An 80 mm length offers enough room for eight placements (four on each side). And if that's not enough for a specific application, PCBs can grow as long as 110 mm.
This is the second time we've seen Marvell's 88SS9183-BNP2 controller, too. Its first appearance was on SanDisk's aforementioned A110. The 88SS9183 rocks two native PCIe physical layers, which means it natively supports two second-gen PCIe lanes. That's fairly special functionality. Most of the PCIe-based SSDs we've tested were based on SATA processors alongside host bus or RAID hardware. After all, a modern RAID controller's job is connecting SATA or SAS storage to the PCI Express bus.
Marvell's implementation is essentially the same processor used in a great many SSDs with specific considerations for connecting through PCI Express. It's significant in that it's as low-power as we're used to on the desktop, and capable of exposing similar features.
The M.2 connector and Marvell's 9183
Next to the controller, we have Toshiba's 64 Gb Toggle-mode DDR manufactured on a 19 nm process. Again, that's the same flash found on Plextor's celebrated M5 Pro, rated for 3000 P/E cycles. We already know this stuff is fast.
It's also worth a reminder that Marvell's 9183 controller and Plextor's firmware are in AHCI mode, supported by most modern operating systems without specialized drivers. The hardware does most of the things other SSD processors do, just over PCI Express. Differences to exist though, mostly with power consumption. DevSlp, for example, is a SATA command. The M6e should drop into a deep PCIe sleep state if the endpoint (that is, the slot Plextor's SSD populates) cooperates. We'll take a magnifying glass to power in just a bit.
Given the eight total NAND packages, we know each must employ a quartet of 64 Gb dies, totaling 256 GB. Plextor reserves just ~7% of that space for spare area.
Toshiba's NAND
Our consumer storage test bench is based on Intel's Z77 Platform Controller Hub paired with an Intel Core i5-2400 CPU. Intel's 6- and 7-series chipsets are virtually identical from a storage perspective. We're standardizing on older RST 10.6.1002 drivers for the foreseeable future.

Updates to the RST driver package occasionally result in subtle performance changes. They can also lead to some truly profound variance in scores and results as well, depending on the revision. Some versions flush writes more or less frequently. Others work better in RAID situations. Builds 11.2 and newer support TRIM in RAID as well. Regardless, results obtained with one revision may or may not be comparable to results obtained with another, so sticking with one version across all testing is mandatory.
| Test Hardware | |
|---|---|
| Processor | Intel Core i5-2400 (Sandy Bridge), 32 nm, 3.1 GHz, LGA 1155, 6 MB Shared L3, Turbo Boost Enabled |
| Motherboard | Gigabyte G1.Sniper M3 |
| Memory | G.Skill Ripjaws 8 GB (2 x 4 GB) DDR3-1866 @ DDR3-1333, 1.5 V |
| System Drive | Intel S3500 480 GB SATA 6 Gb/s, Firmware: 0306 |
| Drive(s) Under Test | Plextor M6e 256 GB M.2 PCIe x2, Firmware: 1.00 |
| Comparison Drives | Plextor M6S 256 GB SATA 6 Gb/s, Firmware: 1.00 |
| Plextor M6M 256 GB mSATA 6 Gb/s, Firmware: 1.00 | |
| Adata SP920 1024 GB SATA 6 Gb/s, Firmware: MU01 | |
| Adata SP920 512GB SATA 6 Gb/s, Firmware: MU01 | |
| Adata SP920 256 GB SATA 6 Gb/s, Firmware: MU01 | |
| Adata SP920 128 GB SATA 6 Gb/s, Firmware: MU01 | |
| Crucial M550 1024 GB SATA 6 Gb/s, Firmware: MU01 | |
| Crucial M550 512 GB SATA 6 Gb/s, Firmware: MU01 | |
| Intel SSD 730 480 GB SATA 6 Gb/s, Firmware: L2010400 | |
| Samsung 840 EVO mSATA 120 GB, Firmware: EXT41B6Q | |
| Samsung 840 EVO mSATA 250 GB, Firmware: EXT41B6Q | |
| Samsung 840 EVO mSATA 500 GB, Firmware: EXT41B6Q | |
| Samsung 840 EVO mSATA 1000 GB, Firmware: EXT41B6Q | |
| SanDisk X210 256 GB, Firmware X210400 | |
| SanDisk X210 512 GB, Firmware X210400 | |
| Intel SSD 530 180 GB SATA 6Gb/s, Firmware: DC12 | |
| Intel SSD 520 180 GB SATA 6Gb/s, Firmware: 400i | |
| Intel SSD 525 180 GB mSATA, Firmware: LLKi | |
| SanDisk A110 256 GB M.2 PCIe x2, Firmware: A200100 | |
| Silicon Motion SM226EN 128 GB SATA 6Gb/s, Firmware: M0709A | |
| Crucial M500 120 GB SATA 6Gb/s, Firmware: MU02 | |
| Crucial M500 240 GB SATA 6Gb/s, Firmware: MU02 | |
| Crucial M500 480 GB SATA 6Gb/s, Firmware: MU02 | |
| Crucial M500 960 GB SATA 6Gb/s, Firmware: MU02 | |
| Samsung 840 EVO 120 GB SATA 6Gb/s, Firmware: EXT0AB0Q | |
| Samsung 840 EVO 240 GB SATA 6Gb/s, Firmware: EXT0AB0Q | |
| Samsung 840 EVO 480 GB SATA 6Gb/s, Firmware: EXT0AB0Q | |
| Samsung 840 EVO 1 TB SATA 6Gb/s, Firmware: EXT0AB0Q | |
| SanDisk Ultra Plus 64 GB SATA 6Gb/s, Firmware: X211200 | |
| SanDisk Ultra Plus 128 GB SATA 6Gb/s, Firmware X211200 | |
| SanDisk Ultra Plus 256 GB SATA 6Gb/s, Firmware X211200 | |
| Samsung 840 Pro 256 GB SATA 6Gb/s, Firmware DXM04B0Q | |
| Samsung 840 Pro 128 GB SATA 6Gb/s, Firmware DXM04B0Q | |
| SanDisk Extreme II 120 GB, Firmware: R1311 | |
| SanDisk Extreme II 240 GB, Firmware: R1311 | |
| SanDisk Extreme II 480 GB, Firmware: R1311 | |
| Seagate 600 SSD 240 GB SATA 6Gb/s, Firmware: B660 | |
| Intel SSD 525 30 GB mSATA 6Gb/s, Firmware LLKi | |
| Intel SSD 525 60 GB mSATA 6Gb/s, Firmware LLKi | |
| Intel SSD 525 120 GB mSATA 6Gb/s, Firmware LLKi | |
| Intel SSD 525 180 GB mSATA 6Gb/s, Firmware LLKi | |
| Intel SSD 525 240 GB mSATA 6Gb/s, Firmware LLKi | |
| Intel SSD 335 240 GB SATA 6Gb/s, Firmware: 335s | |
| Intel SSD 510 250 GB SATA 6Gb/s, Firmware: PWG2 | |
| OCZ Vertex 3.20 240 GB SATA 6Gb/s, Firmware: 2.25 | |
| OCZ Vector 256 GB SATA 6Gb/s, Firmware: 2.0 | |
| Samsung 830 512 GB SATA 6Gb/s, Firmware: CXMO3B1Q | |
| Crucial m4 256 GB SATA 6Gb/s Firmware: 000F | |
| Plextor M5 Pro 256 GB SATA 6Gb/s Firmware: 1.02 | |
| Corsair Neutron GTX 240 GB SATA 6Gb/s, Firmware: M206 | |
| Graphics | MSI Cyclone GTX 460 1 GB |
| Power Supply | Seasonic X-650, 650 W 80 PLUS Gold |
| Chassis | Lian Li Pitstop T60 |
| RAID | LSI 9266-8i PCIe x8, FastPath and CacheCade AFK |
| System Software and Drivers | |
| Operating System | Windows 7 x64 Ultimate |
| DirectX | DirectX 11 |
| Drivers | Graphics: Nvidia 314.07 RST: 10.6.1002 IMEI: 7.1.21.1124 Generic AHCI: MSAHCI.SYS |
| Benchmarks | |
|---|---|
| ULINK DriveMaster 2012 | DM2012 v970, JEDEC 218A-based TRIM Test |
| Tom's Hardware Storage Bench v1.0 | Intel iPeak Storage Toolkit 5.2.1, Tom's Storage Bench 1.0 Trace Recording |
| Iometer 1.1.0 | # Workers = 1, 4 KB Random: LBA=16 GB, varying QDs, 128 KB Sequential, 16 GB LBA Precondition, Exponential QD Scaling |
| PCMark 8 | PCMark 8 2.0.228, Storage Consistency Test |
| PCMark 7 | Secondary Storage Suite |
Fantastic sequential read and write performance is a trademark of modern SSDs. To measure it, we use incompressible data over a 16 GB LBA space, and then test at queue depths from one to 16. We're reporting these numbers in binary (where 1 KB equals 1024) instead of decimal numbers (where 1 KB is 1000 bytes). When necessary, we also limit the scale of the chart to enhance readability.
128 KB Sequential Read

Using the popular Plextor M5 Pro as a comparison point gives us the opportunity to examine two otherwise-similar drives attached a couple of different ways. In the chart above, the biggest difference isn't controller or firmware, but rather the circumvention of SATA 6Gb/s' speed barrier. Both drives come armed with the same number of dies, and employ the same 19 nm Type A Toggle-mode DDR flash.
Anywhere above a queue depth of one, the PCIe-based M6e is significantly quicker than the M5 Pro, ending up more than 200 MB/s faster than the SATA-bound SSD. It's not that we haven't seen performance like this before. But it always makes me happy to see it again.
128 KB Sequential Write

The PCIe-attached M6e strikes again, besting the M5 Pro to the tune of 100+ MB/s. And the 256 GB model isn't even the fastest M6e. Plextor's 512 GB model should add even more sequential throughput.
Here's a breakdown of the maximum observed 128 KB sequential read and write performance with Iometer:
Of course, the only other truly comparable drive is SanDisk's A110, which uses the same M.2 form factor. In fact, both drives pack the same Marvell 88SS9183 heater and 19 nm Toggle-mode flash, backed by custom firmware. But unlike the A110, Plextor's drive is something you can buy right now, while the A110 remains OEM-only.
We turn to Iometer as our synthetic metric of choice for testing 4 KB random performance. Technically, "random" translates to a consecutive access that occurs more than one sector away. On a mechanical hard disk, this can lead to significant latencies that hammer performance. Spinning media simply handles sequential accesses much better than random ones, since the heads don't have to be physically repositioned. With SSDs, the random/sequential access distinction is much less relevant. Data are put wherever the controller wants it, so the idea that the operating system sees one piece of information next to another is mostly just an illusion.
4 KB Random Reads
Testing the performance of SSDs often emphasizes 4 KB random reads, and for good reason. Most system accesses are both small and random. Moreover, read performance is arguably more important than writes when you're talking about typical client workloads.
It's pretty rare to see more than 100,000 IOPS in a 4 KB random read test. And the M5 Pro running Plextor's Xtreme-branded firmware falls just short. The 256 GB M6e sneaks past that barrier at a queue depth of 32, though.
Separately, it's notable that Crucial's SATA 6Gb/s M550 at 1 TB also posts 100,000 IOPS. A conversion to MB/s would show this test to be device-limited, and not held back by its interface, which is why both Plextor drives come so close to each other.
4 KB Random Writes
Random write performance is also important. Early SSDs didn't do well in this discipline, seizing up even in light workloads. Newer SSDs wield more than 100x the performance of drives from 2007, though we also recognize that there's a point of diminishing returns in desktop environments.

We're still dealing with small, random transfers, so this metric cannot illustrate the benefit of PCI Express' faster pipe. But there's a performance difference to explain, and I'd suggest the difference comes down to drivers. Why software? Well, the M5 Pro is SATA-attached, meaning Intel's Rapid Storage Technology driver controls it. The M6e isn't a SATA device, though it still employs AHCI and leans on Microsoft's built-in AHCI drivers. In Windows 7, that's MSACHI.sys. We've seen those two drivers throw off our storage scores in the System Builder Marathon many, many times.
Here's a break-down of the maximum observed 4 KB sequential read and write performance with Iometer. The order the drives appear in our chart is determined by maximum combined read and write performance.
Plextor's M6e earns top honors. No other drive can match its combined maximum read and write transactional 4 KB performance. The margin of victory isn't large, though. PCI Express nets this SSD a big win in large-block sequential transfers. Lots of overhead limits the benefit of more headroom in random reads and writes, though.
The M.2-based M6e and A110 are both held back in our random performance benchmark, and we believe the culprit is AHCI. When the SSD-only NVMe standard takes off, then watch out. We're going to see big reductions in latency and improved random transfers.
Why does SanDisk's A110 fall so far relative to the competition? It might have shown better, except that its emulated SLC-like caching algorithm is swamped by random writes. We test with 16 GB LBA spaces (meaning random data gets sprinkled over 16 GB worth of addresses), hampering the caching mechanism. Tuned more for real-world storage workloads, the A110 exhibits lower results in synthetic random write metrics to larger address spaces. Had I tested just 8 GB, we'd see significantly better numbers.
Random Performance Over Time
My saturation test consists of writing to each drive for 12 hours using 4 KB blocks with 32 outstanding commands. But first I secure erase each drive. Then, I apply the write load, illustrating average IOPS for each minute (except for the last 20 minutes, where I zoom in and show you one-second average increments).
What we're doing here is taking a hard look at latency, quality of service, and consistency. Plextor continues to improve its products with an eye to the enterprise space. The M6e is decidedly enthusiast-oriented, but that doesn't mean some of the company's efforts don't trickle down into its behavior.
This chart comes from The SSD 730 Series Review: Intel Is Back With Its Own Controller. The 100% write (in pink), 50% write (in green), and 30% write (in blue) workloads are tightly grouped. There aren't any disturbing variations.
Now look at Adata's Premier Pro SP920 subjected to the same test:

The difference is significant. Each workload "band" is barely distinguishable, and there's a ton more variance from one second to the next, meaning a constant real-world application accesses I/O from the SSD inconsistently. If one operation depends on the previous one, there could be a comparatively long wait between them.
But the SP920 is representative of how most desktop SSDs behave. They typically aren't tasked with steady, demanding tasks. Conversely, in the enterprise space, predictably latency is key to building a reliable storage subsystem.
And that's why Plextor's result is so interesting. Have a look at this:

In the 12-hour scale, the company's M6e starts at 72,000 IOPS or so, which is typical after a few minutes. Then as the drive is filled, the PCIe-attached SSD starts putting up a fight, periodically reclaiming dirty, invalid blocks. Eventually, it gives up and ends up in a true steady state.
Break out a second-by-second graph of the three workloads shown above, and we see that the M6e looks a lot like Intel's SSD 730. With just 7% spare capacity to utilize, Plextor's M6e can't quite hang with the 730's significant over-provisioning, which means is doesn't achieve the same rarefied performance. But the variation is minimal, limited to a few percent.
If Intel is already celebrated for delivering I/O consistently, then Plextor deserves praise as well. By limiting the M6e's performance ceiling, it keeps its floor in check, too. That's not such an apparent advantage on the desktop. However, it's a good sign that an SSD is designed hold its ground under the most grueling storage workloads.
Storage Bench v1.0 (Background Info)
Our Storage Bench incorporates all of the I/O from a trace recorded over two weeks. The process of replaying this sequence to capture performance gives us a bunch of numbers that aren't really intuitive at first glance. Most idle time gets expunged, leaving only the time that each benchmarked drive is actually busy working on host commands. So, by taking the ratio of that busy time and the the amount of data exchanged during the trace, we arrive at an average data rate (in MB/s) metric we can use to compare drives.
It's not quite a perfect system. The original trace captures the TRIM command in transit, but since the trace is played on a drive without a file system, TRIM wouldn't work even if it were sent during the trace replay (which, sadly, it isn't). Still, trace testing is a great way to capture periods of actual storage activity, a great companion to synthetic testing like Iometer.
Incompressible Data and Storage Bench v1.0
Also worth noting is the fact that our trace testing pushes incompressible data through the system's buffers to the drive getting benchmarked. So, when the trace replay plays back write activity, it's writing largely incompressible data. If we run our storage bench on a SandForce-based SSD, we can monitor the SMART attributes for a bit more insight.
| Mushkin Chronos Deluxe 120 GB SMART Attributes | RAW Value Increase |
|---|---|
| #242 Host Reads (in GB) | 84 GB |
| #241 Host Writes (in GB) | 142 GB |
| #233 Compressed NAND Writes (in GB) | 149 GB |
Host reads are greatly outstripped by host writes to be sure. That's all baked into the trace. But with SandForce's inline deduplication/compression, you'd expect that the amount of information written to flash would be less than the host writes (unless the data is mostly incompressible, of course). For every 1 GB the host asked to be written, Mushkin's drive is forced to write 1.05 GB.
If our trace replay was just writing easy-to-compress zeros out of the buffer, we'd see writes to NAND as a fraction of host writes. This puts the tested drives on a more equal footing, regardless of the controller's ability to compress data on the fly.
Average Data Rate
The Storage Bench trace generates more than 140 GB worth of writes during testing. Obviously, this tends to penalize drives smaller than 180 GB and reward those with more than 256 GB of capacity.

Had I not securely erased the M6e myself and proceeded to repeat this test over and over, I wouldn't have believed the outcome. Despite multiple iterations on combinations of operating systems and platforms, 180 MB/s was the best I could originally manage. Frequently, the drive scored closer to half of that. The only other Marvell 9183-based drive in our test, SanDisk's A110 lands at 275 MB/s. Clearly, something was fishy. After troubleshooting several days away, I figured the issue out. What was the problem?
Secure erasing a PCIe SSD in Windows without a special tool from the drive vendor is a pain, in short. The device is frozen when the UEFI loads at boot, so normally you'd want to power cycle the drive. With a SATA-attached SSD, you can disconnect its power and plug it back in. A PCIe card requires more creativity. My workaround appeared to work, and the secure erase was reported as successful. But in reality, nothing was getting wiped.
Once I was able to confirm the secure erase, the results improve to 245 MB/s. In theory, that should be indicative of excellent service times on the next page.
In other news, the M5 Pro and newer M6S/M6M models fare quite well, dispatching a significant share of the field and coming within striking distance of the PCI Express-based M6e.
Service Times
Beyond the average data rate reported on the previous page, there's even more information we can collect from Tom's Hardware's Storage Bench. For instance, mean (average) service times show what responsiveness is like on an average I/O during the trace.
It would be difficult to graph the 10+ million I/Os that make up our test, so looking at the average time to service an I/O makes more sense. For a more nuanced idea of what's transpiring during the trace, we plot mean service times for reads against writes. That way, drives with better latency show up closer to the origin; lower numbers are better.
Write latency is simply the total time it takes an input or output operation to be issued by the host operating system, travel to the storage subsystem, commit to the storage device, and have the drive acknowledge the operation. Read latency is similar. The operating system asks the storage device for data stored in a certain location, the SSD reads that information, and then it's sent to the host. Modern computers are fast and SSDs are zippy, but there's still a significant amount of latency involved in a storage transaction.

Like my waist, this chart could probably stand to lose a couple of inches. But Plextor's results necessitate expanding the scale (at least in Windows 7). I'm giving you results from both Windows 7 and 8.1 because the M6e ends up in the rough under Windows 8.1, further out to the right than any other SSD. That's the opposite of where it'd want to be.
Under Windows 7, in green text, everything changes. Instead of demonstrating the worst read time, it yields the best. And write service time shows up just behind SanDisk's A110. We're working with Plextor to narrow down a more concrete explanation. Interestingly enough, average data rate doesn't change, so the figure on the previous page is correct for both operating systems.
After crunching the numbers, the real difference boils down to service time quality. The standard deviation with read/write service times is far lower (and more consistent) under Windows 7. Since we’re displaying mean service time (Tx), outliers add up, and there are more painfully-long operations due to Windows 8.

Think of latency like a golf score; lower is better. On this course, 296 µs is not good. Previously, the 120 GB Crucial M500 turned in the worst averages. But Plextor's M6e in Windows 8 relieves it. Switch to Windows 7, though, and everything is beautiful.
It turns out that this is a problem, and not that Plextor is responsible for. You'll see a lot more AHCI-conforming PCIe-based storage in the future, and it's looking like the issue is Microsoft's AHCI driver under Windows 8. All of our SATA-based drives are benchmarked under Intel's Rapid Storage Technology driver. But we have to use the built-in AHCI software for this type of native storage sitting on the PCI Express bus. MSAHCI.SYS was around forever, and is indeed what you use in Windows 7. That driver appears to behave itself. But it was recently superseded by STORAHCI.SYS in the latest version of Windows, and its performance isn't as predictable, manifested by requests serviced less consistently.
And before you start thinking this issue is limited to AHCI-based native PCIe storage, SATA-attached drives like Crucial's M550 get the same harsh treatment under STORAHCI. Its just that you're stuck with the operating system's generic AHCI driver when you use a device like the M6e.

The A110 is deadly with its 427 µs result, likely due to its emulated-SLC cache. Our trace seems to be ideally suited to SanDisk's nCache technology, providing small idle times the cache uses to flush information to the conventional MLC-configured storage. Plextor’s M6e slides in just behind, impressing with its performance as well. That's a spectacular result if you're in Windows 7. Making the move to Windows 8.1 pushes the M6e back another 200 µs worse on average.
Active Idle Power Consumption
Idle consumption is the most important power metric for consumer and client SSDs. After all, solid-state drives complete host commands quickly and then drop back down to idle. Aside from the occasional background garbage collection, a modern SSD spends most of its life doing very little. Enterprise-oriented drives are more frequently used at full tilt, making their idle power numbers less relevant. But this just isn't the case on the desktop, where the demands of client and consumer computing leave most SSDs sitting on their hands for long stretches of time.
Active idle power numbers are critical, especially when it comes to their impact on mobile platforms. Idle means different things on different systems, though. Pretty much every drive we're testing is capable of one or more low-power states, up to and including DevSlp. That last feature is a part of the SATA 3.2 host specification. And while it requires a capable SSD and a compatible platform, enabling DevSlp takes power consumption down to a very small number.
Measuring the power use of a SATA-based drive is fairly straightforward; you tap into the 5 or 12 V rail, do some multimeter work, and a bit of math. A PCI Express SSD is more difficult to get readings from. You can either use a riser card, like Igor does in our graphics card reviews, or employ a special purpose-built device.
I do the latter. Active idle is reported as a rule, so I disable the bus' sleep states to yield an equivalent to my SATA storage testing. Yes, the two PCIe-based drives can drop to lower idle power when the slot falls into a low-power condition. But we're concerning ourselves with active results in order to make the right comparison.
There is a trio of states to watch out for:
- L0 is active power/active use
- L0s is active power/idle state
- L1 is low power/slumber state
Plextor claims a .58 W idle specification, and that's pretty darned close to what we observe in PCMark 7.
PCMark 7 Average Power Consumption
If we log power consumption through a workload, even a relatively heavy one, we see that average use is still pretty close to the idle numbers. Maximum power may spike fiercely, but the draw during a PCMark 7 run is light. You can see the drives fall back down to the idle "floor" between peaks of varying intensity.

The M6e uses less power than SanDisk's A110 through PCMark 7. It also checks in just under Plextor's M6S and M6M.

After quite a bit of power testing, logging, and fiddling with Excel, we end up with the above chart.
SanDisk doesn't specify its idle consumption in the A110's documentation, though Plextor does. But it looks like both SSDs sit idle around the half-watt mark. The A110 does spent a lot more time above idle than Plextor's M6e, and after some discussion with Plextor, it appears there are ways to bring that figure down even more in certain scenarios. But PCIe power management is an entirely different beast, and I continue refining the testing.
We've been utilizing ULINK's DriveMaster 2012 software and hardware suite to introduce a new test for client drives. Using JEDEC's standardized 218A Master Trace, DriveMaster can turn a sequence of I/O (similar to our Tom's Hardware Storage Bench) into a TRIM test. JEDEC's trace is months and months of drive activity, day-to-day activities, and background operating system tasks.
ULINK strips out the read commands for this benchmark, leaving us with the write, flush, and TRIM commands to work with. Execute the same workload with TRIM support and without, and you end up with a killer metric for further characterizing drive behavior.
DriveMaster is used by most SSD manufacturers to create and perform specific measurements. It's currently the only commercial product that can create the scenarios needed to validate TCG Opal 2.0 security, though it's almost unlimited in potential applications. Much of the benefit tied to a solution like DriveMaster is its ability to diagnose bugs, ensure compatibility, and issue low-level commands. In short, it's very handy for the companies actually building SSDs. And if off-the-shelf scripts don't do it for you, make your own. There's a steep learning curve, but the C-like environment and command documentation gives you a fighting chance.
This product also gives us some new ways to explore performance. Testing the TRIM command is just the first example of how we'll be using ULINK's contribution to the Tom's Hardware benchmark suite.

On a 256 GB drive, each iteration writes close to 800 GB of data, so running the JEDEC TRIM test suite once on a 256 GB SSD generates almost 3.2 TB of mostly random writes (it's 75% random and 25% sequential). By the end of each run, over 37 million write commands are issued.
The first two tests employ DMA to access the storage, while the last two use Native Command Queuing. Since most folks don't use DMA with SSDs (aside from some legacy or industrial applications) we don't concern ourselves with those. It can take up to 96 hours to run one drive through all four runs, though faster drives can roughly cut the time in half. Because so much information is being written to an already-full SSD (the drive is filled before each test, and then close to 800 GB are written per iteration), SSDs that perform better under heavy load fare best. Without TRIM, on-the-fly garbage collection becomes a big contributor to high IOPS. With TRIM, 13% of space gets TRIM'ed, leaving more room for the controller to use for maintenance operations.
TRIM Testing
Average
To avoid adding too much data, I have the average performance for each benchmarked SSD with and without TRIM support enabled. Displayed in IOPS, this helps us make comparisons more quickly.

The M5 Pro is toward the bottom of the chart, while SanDisk's PCIe-attached A110 is closer to the top. Plextor's M6e is faster though, particularly with the TRIM command active.
Instantaneous
But I also want results for the instantaneous average of my TRIM test. How does the drive fare servicing writes with and without TRIM during each 100,000-command window? The purple line represents IOPS across the entire trace, without TRIM. The teal line is with TRIM. Each data point represents write IOPS per 100,000-command test reporting period.

The M6e's performance is improved quite a bit by TRIM. The troughs of the teal line don't dip quite as low, while the highs peak well over the purple indicator's levels. It's safe to conclude that the M6e is best-optimized for environments with TRIM enabled.
Were the reverse true, you'd see the purple line dominate, which is the case for Intel's SSD 730/S3500/S3700 and SanDisk's X210.

Just for fun, we pit the M6e against Plextor's celebrated M5 Pro. As we already know from the average IOPS chart, the outcome shouldn't even be close. And indeed, it isn't. There just isn't a lot to report from the M5 Pro. The lowest lows and highest highs aren't that far apart. Not so for the M6e, though. Its worst drops still tend to exceed the SATA drive's peaks.
Throughput
We collect and report the total throughput of each drive in the NCQ with TRIM test. It's one number that helps capture overall performance in the test.

Futuremark's PCMark 8 takes everything that was good about PCMark 7's storage testing and makes it better. The standard PCMark 8 storage components include real trace activity for 10 different activities. These recorded I/O segments include productivity, gaming, and photo/video manipulation, yielding individual sub-tests. At the end, each category is scored in seconds, a master throughput score is conferred (similar to the average data rate in our Tom's Hardware Storage Bench), and the benchmark generates a total score.
The Plextor M6e displays normal PCMark 8 Storage benchmark stats like so:
| PCM8 Storage | PCMarks | Bandwidth | WoW | Battlefield 3 | Photoshop (Light) | Photoshop (Heavy) | Adobe Illustrator |
|---|---|---|---|---|---|---|---|
| Plextor M6e 256 GB | 4971 | 280.57 MB/s | 58.4s | 133.4s | 113.7s | 359.9s | 70.8s |
| PCM8 Storage | Adobe InDesign | Adobe After Effects | Adobe Illustrator | Microsoft Word | Microsoft Excel | Microsoft PowerPoint |
|---|---|---|---|---|---|---|
| Plextor M6e 256 GB | 57.5s | 70.8s | 72.0s | 28.3s | 9.2 | 9.2 |
This is what the standard PCMark 8 storage test looks like.
Although that's illuminating, the newest version of PCMark 8 Professional includes extended storage tests, which allow us to dig deeper into SSD performance.
In truth, on a new or lightly-used drive, it's hard to distinguish one model from another. It takes tough workloads running over long periods of time to expose the issues that might affect our recommendations. PCMark 8's storage consistency component is designed to hit SSDs with those tasks, getting consumer-oriented products dirty enough to sink the bad ones.
Whether you run PCMark 8's normal or extended tests, those 10 traces form the backbone of this benchmark. As they're played against the drive under test, information is measured and recorded. The extended storage tests simply repeat the traces multiple times using varying conditioning procedures before each session.
First, the raw block device (there is no partition) is preconditioned twice by filling the entire accessible LBA space with 128 KB sequential writes. Once that is completed, the first Degradation Phase randomly writes blocks between 4 KB and 1 MB in size to random LBA spaces on the drive. Since the writes aren't 4 KB-aligned much of the time, the SSD's performance drops quickly. After all, non-4 KB-aligned accesses create overhead and generally increase write amplification significantly.
The first Degradation Phase begins with 10 minutes of those punishing random offset writes, after which each PCMark 8 activity trace is played against the SSD being tested. The successive degradation rounds are similar, except an additional five minutes are tacked onto each iteration. After eight repetitions, that write period expands to 45 minutes.
Next comes the Steady Phase. Each of five Steady Phases writes 45 minutes worth of random offset data prior to trace playback, pushing the drive even harder and making it more difficult to perform housekeeping duties. With fewer blocks available for writing, latency increases substantially.
Lastly, PCMark 8 moves into a Recovery Phase, which consists of five idle minutes before trace playback. Repeat that five times, and the test concludes.
This particular PCMark 8 test is brilliant in that it stresses SSDs methodically with preconditioning, followed by trace playback. You end up with a ton of data covering bandwidth, latency, and duration for all 10 traces and each of the 18 phases. But tracking that information lets us paint a picture of drive performance through each step of the benchmark. Every drive takes a beating during the Degradation and Steady Phases, but the Recovery Phase should push the most resilient drives to the forefront. SSDs that rely on garbage collection during write activity (and not background garbage collection) may not benefit much from this recovery time.
On to the testing.
Storage Consistency With PCMark 8's Adobe Photoshop (Heavy) Trace
Because there are 18 individual rounds packed with 10 traces each, we need to focus. We'll choose one trace, Adobe Photoshop (Heavy), and keep tabs on it through the entire extended run.
Bandwidth

First we have total bandwidth during each iteration of the Photoshop trace. The M6e is hammered right off the bat, averaging just over 15 MB/s of read and write bandwidth for the first eight rounds. As the test transitions into its Steady phases, the M6e continues to drag. But give it a few minutes to recover and watch out; the last two runs achieve more than 60 MB/s.

That might have been more impressive if the 256 GB SP920, 256 GB M5 Pro, and 250 GB 840 EVO SSDs didn't blow Plextor's PCIe-attached drive out of the water. Samsung's mainstream offering, in particular, is potent.
But given what we saw in our own trace test, latency should be more illuminating than bandwidth.
Latency
In this test, we're taking that same Adobe Photoshop (Heavy) trace and using average read and write latency to illustrate responsiveness. We'll sprinkle in some competing drives for comparison, too.

Plextor's M6e doesn't demonstrate the worst average read access time. And it gets better during the Recovery Phases, too.
Our Storage Bench trace has some idle time left in, though not much. This means the M6e could be going in and out of slumber more often in that test than it does in PCMark 8. For the sake of brevity, Futuremark cuts out all idle time.

Lower is better in this chart, and we see that Plextor's M6e starts bad, gets worse, and is at least able to beat the M6M and M6S by the end of the Recovery Phases. In fact, the M6e even approaches Intel's SSD 730 after some rest time.
But as it's getting hammered in PCMark's Degradation Phases, write access latency continues to pose an issue.

The M6e is fast. There's no doubt about that. It's able to hang with the other high-end SSDs we've ever tested. In fact, SanDisk's M.2-based, PCIe-attached A110 is similar, though also a bit faster. But you can't buy the OEM-only A110, whereas Plextor's offering is for sale on Newegg right now.
There's just one catch: two models are up on the site, and both include a four-lane adapter card for the two-lane M6e. The M.2-only version is still on its way. If you pop the little card off of its adapter, your warranty is voided. So, you don't want to buy the M6e for a notebook or desktop motherboard with an M.2 interface just yet. Rather, the 128 and 256 GB drives available today should live life inside a desktop machine with a spare PCI Express x4 link. That lets us narrow down our focus for making recommendations.
Our synthetic benchmarks show sequential read and write performance hundreds of megabytes per second faster than the quickest SATA-based SSDs. Random performance is also excellent, though it's not interface-limited, so PCIe doesn't really convey a quantifiable benefit. I can't say you're going to notice those blazing-fast sequential numbers in day-to-day use, and with random performance no better than the fastest SATA 6Gb/s drives, performance is ultimately a wash for the M6e.
We do see praiseworthy behavior in our write saturation testing, though. With one-second granularity, Plextor's M6e looks a lot like Intel's consistent SSD DC-series drives, including the enthusiast-oriented SSD 730. Those offering surpass Plextor's I/O throughput thanks to significant over-provisioning, but the M6e is at least able to serve up information in well-controlled bands.

This consistency is something Plextor is cooking up for enterprise-class storage, and we're glad to see it applied in the consumer space as well. Most enthusiasts won't see or feel the difference in consumer workloads, but that isn't the point. Devices like Intel's SSD 730 and Plextor's M6e are two of the only drives we've seen behave this way, displaying grace under the pressure of more taxing applications.
TRIM testing turns up another bright spot. Plextor's 9183-based drive behaves exceptionally for an SSD with just 7% over-provisioning, outshining the M5 Pro. It cannot lay a finger on SanDisk's X210 though, a device that shows itself to be outstanding in a number of metrics. You don't get the same screaming sequential numbers (it's hamstrung by the SATA interface, after all). In its degraded state, however, the X210 maintains the lowest write access latency imaginable.
All of that is to say the M6e's PCI Express controller and M.2 form factor, on their own, don't confer an advantage in the most meaningful benchmarks.

There are still questions left to answer about PCI Express storage and its interaction with the AHCI standard. In our Tom's Hardware Storage Bench, we recorded notably different service time profiles under Windows 7 and 8.1. In Windows 7, using Microsoft's built-in MSAHCI.SYS, the numbers look great. Windows 8.1, on the other hand, employs a newer AHCI driver that's far less kind to the M6e, other PCIe-based devices, and even familiar SATA SSDs. Intel may address this in the future with its own software for integrating AHCI- and NVMe-based PCIe storage into its platform architecture.
Finally, our newest PCMark 8-based storage test doesn't give the M6e much of an advantage either. Samsung's 840 EVO and Plextor's M5 Pro soar in the Storage Consistency test. Drives from Intel, SanDisk, and Adata dominate as well. Meanwhile, the M6e's performance is more ordinary than I was expecting given its interface and pedigree.
All of that would be fine if Plextor was hitting the right price points. But you can pick up three Crucial M550s for what the same $300 being asked for a 256 GB M6e. Sling them together in RAID 0 and you have enough throughput to saturate Intel's DMI interface (I'd be willing to guess that's almost 1500 MB/s in sequential reads and 1000 MB/s in writes).
Plextor's price tags could make more sense if we were looking at M.2 models (without the adapter) for notebooks, where multiple SATA drives aren't as easy to pull off. In a desktop, it's slightly more difficult to make a case for $1+/GB solid-state storage, even if it's attached through the PCI Express bus.
Of course, there's another way of looking at this. The M6e's simplicity and elegance (that is, a native PCIe controller) put it far ahead of the PCI Express-based SSDs we've reviewed in the past. Most of those needed multiple SATA controllers attached to host bus logic. And they were way more expensive than $300 for 256 GB.

There are pros and there are cons. No matter what, though, enthusiasts are going to find Plextor's M6e desirable. It's a great solution for loading Windows and launching performance-sensitive applications, leaving native SATA ports open for a big RAID array. As a storage fiend, it's easy to see how the M6e matched up to big mechanical disks would be a fun combination. And even if the M6e only really moves the needle in sequential transfers, the M.2-based version, without a bundled adapter, promises to satisfy mobile enthusiasts.
If Plextor can bring its price down over time, its unique brand of performance and this cutting-edge form factor should attract lots of attention. Again, though, we're most excited about the M.2-specific version. On the desktop, for what Plextor is charging today, there are more compelling solutions available.




