Plextor M6e Black Edition 256GB review

Plextor M6e Black Edition 256GB review

Plextor sent us her M6e Black Edition 256GB SSD, a new version of the marketed in 2014 PCI-Express SSD. We have tested it extensively.

Rarely has a name as effective as the M6e Black Edition, because it shows exactly what we have to do with a black version of Plextor M6e. Back in repetition: the M6e early 2014 was one of the first PCI Express-based SSDs and anyway about the only PCI Express-based SSD that you could buy loose through normal roads. In fact, it is a product with M.2 form factor, but certainly since the beginning of 2014 few people had a motherboard with M.2 Finally, the Plextor M.2 combined module with a PCI-Express card converter, allowing the SSD a normal PCI-Express x4, x8, or x16 slot can be placed. Those that want to read back what is special about SSDs with PCI Express interface and how it is now with M.2 and other form factors, please refer to our December article published: 2015: the year of PCI Express SSDs.

The new M6e Black Edition is basically the same as last year tested M6e even now we have to do with a PCI-Express SSD form factor M.2 into a converter module. The fact that the Plextor black PCB now has SSD and a metal cap annex has provided cooler, however, gives the product directly a completely different look. Where the original M6e was actually a green monster that a beautiful high-end PC would disfigure considerably, the M6e Black Edition is a product that just a high-end system that can give cachet.

The actual SSD is identical, though the M6e Black Edition is now output with firmware version 1.5, where we tested the original Plextor M6e with firmware version 1.0. The SSD is based on the Marvell 88SS9183 contrroller that has a PCI-Express 2.0 x2 interface. That means a theoretically maximum doorvooersnelheid 1000 MB / s; significantly more than the 600 MB / s limit of Serial ATA 600, but significantly less than with PCI Express 3.0 and / or 4-lanes is possible, as we have already described.

The Marvell controller is combined with 19nm MLC flash memory from Toshiba factory. The SSD has a 512MB DRAM buffer. Plextor indicates that the M6e Black Edition read speeds of up to 770 MB / s and write speeds of up to 580 MB / s should be able to pick up. For random workloads promises Plextor 105,000 4K read IOPS and 100,000 4K write IOPS. These specifications are not changed from the original M6e. Not changed is the support for 256-bit AES encryption, though the product is not suitable for Microsoft eDrive. Plextor offers a 5 year warranty on the M6e Black Edition, to achieve a minimum number of writing assignments, called TBW value, however, is not specified.

Plextor bundles the M6e with the latest version of its software with PlexTools PlexTurbo functionality. In addition, unused RAM used as a buffer for the SSD to improve the pretations. Further on we come back here.

The price of the product is spicy; average you pay € 257 for the M6e Black Edition 256GB, or a bit more than a dollar per GB. In comparison, the fastest SATA SSD 600 of this moment, the Samsung 850 Pro, costs on average € 168 or converted 66 cents per GB for the 256GB version. It therefore makes this Plextor PCI-Express SSD is about 50% more expensive. The big question is whether this product that extra cost is worth.

Test

Our SSD test consists of several, mostly on Iometer based synthetic tests, then two on PCMark based practice tests and two further continuous tests. Our test system consists of an Intel Core i3 on ASRock Z97 Extreme6 motherboard with M.2 slot that can accommodate SSDs with both two and four PCI-Express lanes.Tests running on Windows 7 x64. SATA SSDs we naturally connect to a Serial ATA 600 port with AHCI enabled. In Windows, we use the Intel RST drivers.

Iometer tests

Using Iometer, we first determine the random read and write with 4k data blocks. Precisely these tests with small data blocks as an indication of the performance of SSDs in practice very important: on Windows, but also in other operating systems is 4k normally the most common block size that allows hard drives and SSDs are controlled. The test we do with queue depth 1, 2, 4, 8, 16 and 32. This queue depth indicates how many concurrent instruction be fired on an SSD. In QD32 may SSDs themselves show their best side, but rather the performance at low queue depth are of great importance in consumer applications.

With Iometer we also measure the reading and writing of large data blocks of 1 MB. This test we do at QD32, but precisely such data blocks minimizes the amount of parallel requests little to nothing, because the SSD controller that can divide multiple SSD excellent channels.

All Iometer tests we perform with fully randomized data, making SSD controllers compression tricks that can not obtain benefit from. Also run all the different tests at least 30 seconds each, much longer than comparable tests that are part of various benchmark tools.

Although we actually can get all the relevant synthetic test data from IOmeter, we still also run the popular AS SSD benchmark, which is internally based on the performance in 4k data blocks (QD1 and QD64) and sequential read and write performance. Although so redundant relative to the Iometer tests, AS SSD has two advantages: first of all the scores are neatly turned into a clear total score, which gives good first impression of the performance of a SSD. Additionally, you can also run AS SSD simple home and so easy to compare scores.

Real-world benchmarks: PCMark

For consumer applications, real-world benchmarks and PCMark7 PCMark8 are much more important. PCMark7 simulates hard disk access of real programs and indicates the drive performance in different scenarios. The total score provides a measure of general use, the sub-scores to give an indication of the speed at different usage models. The so-called traces are based on software from the Windows 7 era. PCMark 7 plays the scenarios in real time; performance gains that you would not find in reality, you see in PCMark 7 is therefore not reflected in the scores.

We also show PCMark7 ‘Raw’ scores. It is idle time omitted, so the scores are no longer one-to-one correlation with the performance of the applications in practice, but you can nicely see the actual performance difference between SSDs.

Furthermore, we turn PCMark8, the latest version of the benchmark. Also this benchmark has another hard drive / SSD test, again based on traces, but now modern software. PCMark 8 contains traces of Adobe Photoshop, Adobe Illustrator, Adobe InDesign, Adobe After Effects, Microsoft Word, Microsoft Excel, Microsoft PowerPoint, World of Warcraft and Battlefield 3.

Continuous tests

Finally, we do have two continuous tests using a workload for 30 minutes on an SSD, and per minute using the average performance. First, we do this with the Iometer 4k random write test, in addition to the Iometer database simulation. Both of these tests are carried out continuously and in a file with QD32 that covers 75% of the capacity of the existing flash memory. More information about this continuous tests away. For consumer applications, these continuous tests of little importance for the professional use of SSDs – for example servers – the results of these tests are much more important than any other benchmark whatsoever.

Consistency tests

Consistency we understand how performance can deviate from second to second. To bring consistency in map, we turn again Iometer 4k random write workload QD32 but instead of every minute to determine the performance, we determine an average for each second. Let it be said immediately: the consistency of performance is mostly something for the enterprise market is important. Who put an SSD in a heavy duty database server, wants to be sure that the performance of the used storage are predictable: large peaks and valleys can cause unpredictable behavior of workloads. For consumer use is actually (almost) irrelevant except that the performance consistency undoubtedly has an effect on the performance of SSDs when you put them in RAID 0. After all, in a RAID 0 array are blocks of data to and written to multiple disks, and the performance is in fact true that the slowest disk (the “weakest link”) provides the ultimate level of performance. If one drive has a moderate consistency of performance, the overall performance of a RAID 0 array will be relatively low, given the likelihood that either a “dip” than is relatively large.

The power consumption of SSDs we measure using a BTO W540EU laptop based on Intel Core i3 3120M CPU and the Intel H77M chipset. We measure the power consumption of SSDs both idle and during a 4k random write and 1 MB sequential write workloads. PCI Express SSDs if Plextor we have unfortunately not yet possible to measure the power consumption.

Graphs

For the graphs somewhat reduce clutter, we only have a limited set of other popular SSD series, of course, all in 240-256 GB format.

AS SSD

AS SSD works with oncomprimeerbare data. As with our Iometer tests SSDs with SandForce controller so there is no advantage of the built-in compression tricks.

AS SSD does both read and write tests. First they tested with data blocks of 4KB, one instruction at a time, and then with 4 KB data blocks with 64 simultaneous instructions. Finally, is also carried out a sequential read and write test, which gives a measure of the work with very large files. Based on all tests AS SSD also determine an overall score. Although the tests are largely redundant with the Iometer tests we run SSDs because this benchmark also can easily rotate in order to make a comparison with your own SSD.

When we entered the M6e with firmware version 1.0 tested, we achieved a new record score 1231 points. Now we get to 1158 points, a fraction less. This is comparable to the fastest SATA600 SSDs.

Looking at the sub-scores on the next page, we see that the lower total score mainly because of a lower random write score high queue depth, which we already saw in our Iometer tests. With sequential read is the M6e in AS SSD 640 MB / s over, so faster than SATA600 SSDs. With sequential write, we measure approximately 540 MB / s.

PCMark7

PCMark7 uses traces based on relatively modern software from the Windows 7 era. Because this benchmark uses real and modern applications, this is along with PCMark8 actually the most important benchmark on which to base your decision to buy the SSD, if you buy an SSD for consumer applications. A difference with PCMark Vantage, a benchmark that we no longer run for SSDs because he is now too old, that PCMark7 playing the traces in real time. It left scores of SSDs closer together, but also get the results much more in line with what you would expect in actual practice.

In the synthetic benchmarks, we concluded cautiously that the new firmware is more tuned for consumer applications, since the writing with low queue depth has increased and the write speed with high queue depth decreased. That is reflected in PCMark7: with 5632 points, the M6e Black Edition in second place, just behind the PCI Express 2.0 x4 based Samsung XP941.

Plextor M6e Black Edition 256GB review

PCMark8

The storage benchmark PCMark8 is again based on traces, but now modern software. PCMark8 contains traces of Adobe Photoshop, Adobe Illustrator, Adobe InDesign, Adobe After Effects, Microsoft Word, Microsoft Excel, Microsoft PowerPoint, World of Warcraft and Battlefield 3. The total score is calculated by taking the geometric mean of the different parts. Again the traces in real time play, so including idle time. This means that, as with the scores of PCMark7 SSD are very close to each other. This therefore corresponds to the ground: in normal consumer applications, it is virtually impossible to distinguish from each other modern SSDs.

Where PCMark7 for all tests also provides a raw score, PCMark8 shows in the same way an average speed of play drive activity traces. This is no idle time and again shows the actual underlying differences in performance of SSDs.

The differences are very small, in practice, simply because you hardly notice a difference between the various popular SSDs right now – at least in these scenarios. Almost all SSDs score between 4900 and 5000 points. The M6e Black Edition 256GB with 5014 set a new record sip. The measured storage bandwidth is 334.7 MB / s, which is also faster than any other SSDs, including the Samsung XP941.

Conclusion

A year ago, Plextor the first native PCI Express SSD threw for consumers in newsstand sales and now, in early 2015, it still has a unique position with it. The new M6e Black Edition is functionally basically the same as last year, but the beautiful appearance instead of the ugly green circuit board of yesteryear, the product makes it a lot more high-end PC fähig.

Looking at the performance, we must unfortunately again conclude that this SSD with PCI Express interface is not really faster than the fastest SATA 600 SSDs. The maximum transfer rates than are in fact somewhat higher than with SATA600 possible, the difference is very limited. That is of course because the use Marvell controller uses only two PCI Express lanes, which also of the 2.0 type. Even if you have lots of sequential writing assignments to your purpose, you will therefore have little benefit.

Especially when you consider that the M6e Black Edition fetches some EUR per GB, we can only conclude that you are a high-end SATA SSD 600 as the Samsung 850 Pro 256GB actually does a better deal than this Plextor product. The PlexTurbo functionality certainly offers added value, although you do not expect miracles in practice. And it is also true: just Samsung offers such functionality as well.

Or if you want something special in your PC, you can consider this Black Edition. But if we look strictly at performance and price, we can not really recommend the product.

Plextor M6e Black Edition 256GB review updated: February 11, 2015 author: John Malkovich