by Ilya Gavrichenkov
06/20/2013 | 02:16 AM
We have tested a number of relatively fast yet affordable solid state drives recently, such as the OCZ Vertex 3, Kingston SSDNow V300 and Samsung 840. The mass arrival of such products might have been expected because fast synchronous flash memory has been getting cheaper with the introduction of new manufacturing technologies. As a result, SSDs now have a perfect opportunity to expand from their traditional habitat. The price of entry-level SSDs doesn’t frighten potential customers anymore, but they are not much slower than expensive flagship offerings. That’s why there is some agitation in the market of affordable SSDs. New players have arrived that want to attract the customer not only with low prices but also with high performance.
So today we have a chance to check out yet another product that sports an attractive price/performance ratio. Manufactured by SanDisk, the Ultra Plus SSD is far more exciting than the other inexpensive SSDs we’ve tested so far. While the majority of manufacturers create their entry-level products by combining time-tested controllers with cheap flash memory, SanDisk takes a different approach. The innovative SanDisk Ultra Plus features an inexpensive Marvell controller we have never met before, a special variety of flash memory, and customized firmware developed by SanDisk.
SanDisk has its own flash memory manufacture (jointly with Toshiba) but its consumer-class SSDs used to be rather dull. The company’s flagship model, Extreme, is based on the very common LSI SF-2281 controller, uses reference firmware and thus lacks any originality. But now SanDisk seems ready to correct its priorities and take the SSD market as seriously as its flash cards and embedded solutions. The Ultra Plus is the first fruit of the new strategy, which makes it the more exciting.
The SanDisk Ultra Plus embodies some original solutions rather than copies the conventional architecture of the majority of SATA 3 drives. The controller uses fewer than eight channels to communicate with flash memory but the available channels are utilized more efficiently. This has helped make the SSD design simpler and cheaper and also reduce the power consumption of the resulting product. Of course, with fewer controller channels, the SSD is likely to perform slower, but SanDisk, being a flash memory maker, has some tricky technologies to make up for that.
To be specific, the SanDisk Ultra Plus is based on the Marvell SS889175 controller. We have not met any SSDs with this chip so far, just because it is not designed for consumer SSDs. It is a simplified version of the popular Marvell SS889187 controller, which is employed in such products as the Plextor M5 Pro and the Crucial M500, but with only half the memory access channels. The SS889175 is meant for mobile applications that do not require high performance, but SanDisk has found a new purpose for it.
The controller is coupled with SanDisk’s exclusive eX2 ABL MLC NAND flash memory manufactured on 19nm tech process. Similar to conventional synchronous MLC memory with Toggle Mode interface, it features the proprietary nCache technology: some of the MLC memory cells work in faster SLC mode and act as a nonvolatile cache to increase writing performance and improve service life since caching involves consolidation of fragmented requests.
Now that we have a general picture of the Ultra Plus’s architecture, let’s take a look inside the 256GB model we’ve received to test. Its PCB is surprisingly small, populated by very few chips, yet we can find all conventional components here.
First of all, the SanDisk Ultra Plus 256GB only has four memory chips. Each of them is 64 GB and contains eight 8-gigabit NAND devices, so the controller uses 8-way interleaving in this SSD. By the way, that’s why the SanDisk Ultra Plus series cannot include any models with capacities above 256 GB.
Besides the flash memory and the Marvell SS889175 controller, the PCB carries a dedicated DRAM buffer typical of any non-SandForce-based SSD. It is a 128MB chip of DDR2-800 SDRAM. Thus, considering the nCache technology, the SanDisk Ultra Plus does double caching. And it is this double caching that is expected to ensure high performance with the 4-channel controller. Theoretically, it should work because SLC memory is about twice as fast as MLC memory.
There’s only one questionable thing concerning the size of the nCache. The flash memory installed into the SanDisk Ultra Plus has standard organization, i.e. 8-gigabit semiconductor devices. The SLC cache is allotted from the drive’s total capacity and cannot have a large size. The user-inaccessible part of SanDisk Ultra Plus drives is a typical 7% or about 17.5 GB. Some of this reserved pool must be used for block replacement, garbage collection and wear leveling instead of nCache. Moreover, SLC memory needs twice the memory cells for storing data in comparison with MLC memory. All in all, it seems that the nCache is only about 4 GB large in the 256GB model. That’s not much. On the other hand, this should be enough for regrouping and accelerated execution of random-address requests, which is the main point of the technology according to SanDisk.
The manufacturer suggests that this SLC cache is enough to make the Ultra Plus as fast as modern SSDs with 8-channel controllers. This is indicated by the official specs of the SanDisk Ultra Plus 256 GB:
Of course, the relatively high speed is ensured not only by the dual caching but also by firmware optimizations. The latest version is X2306RL and we recommend using it since it has certain improvements and corrects earlier bugs.
Thus, the SanDisk Ultra Plus looks like a typical inexpensive SSD. Its hardware components don’t seem advanced, but it may be competitive in real-life applications. Its manufacturer positions it lower than the SandForce-based Extreme series, though, which is reflected in the way the Ultra Plus series is offered to the customer.
The packaging is rather dull. It is just a small gray box.
The SSD is shipped together with a user manual and a frame that can increase its thickness from 7 to 9.5 mm. You can also download the SSD Toolkit utility from the SanDisk website. The utility isn’t very functional but offers some basic features like viewing general info about the SSD, checking out SMART parameters and updating firmware.
The SSD case is made of robust plastic. The model can be identified by two stickers: a large gray label with logos on one side and an info sticker with part number, serial number and barcodes on the other.
Talking about the new drive’s positioning, SanDisk claims that the Ultra Plus is especially good for multimedia data and recommends it for image and audio editing applications. We’ll check this out in our tests.
Although the nCache technology is expected to improve the service life of flash memory, the SanDisk Ultra Plus comes with a standard 3-year warranty.
For our today’s SSD test session we use a unified test system built on an Intel H77 based mainboard, which features two SATA 6 Gbit/s ports. We will use these ports to connect the tested SSDs.
As for the today’s testing participants, it turned out not as simple as we had expected. On the one hand, SanDisk Ultra Plus is an inexpensive SSD with obvious limitations on the hardware design level. However, during our practical tests it proved to be substantially better than we had originally anticipated. Therefore, we ended up comparing its performance against all current SSDs. It means that the performance diagrams below will show the results for many popular products on LSI, Marvell, LAMD, Samsung and Indilinx controllers.
The LSI SF-2281 is represented by the today’s fastest product based on it - Intel SSD 520, as well as a typical solid state drive - Corsair Force GS. LAMD LM87800 controller will be tested in Corsair Neutron GTX and Corsair Neutron SSDs. OCZ Vertex 4 and OCZ Vector defended the honor of the Indilinx Everest 2 and Indilinx Barefoot 3 controllers. Marvel platform was brought in by Plextor M5S on Marvell 9174 controller, and Plextor M5 Pro based on a more up-to-date Marvell 9187 controller. And Samsung products were represented by both contemporary drives – Samsung 840 Pro and Samsung 840. All above mentioned SSDs use exclusively synchronous MLC flash memory. Corsair Neutron, Intel SSD 520, OCZ Vertex 4, OCZ Vector and Plextor M5S are built with 25 nm memory from IMFT consortium with ONFI-interface. Corsair Force GS, Corsair Neutron GTX, Plextor M5 Pro and Samsung 840 Pro use Toggle Mode MLC NAND manufactured by Toshiba using 2x nm or 19 nm production process. Samsung 840 SSD stands out due to the fact that it is based on 21 nm TLC NAND with Toggle Mode 2.0 interface. We did our best to ensure that all testing participants came in the closest storage capacity, to ensure fairness of the comparison – 240/250/256 GB.
Overall our testbed was configured as follows:
We use Anvil's Storage Utilities 1.0.51 to measure random and sequential ref and write speeds. The synthetic benchmark integrated into this software suite provides a great overview of the products by experimentally checking out a wide variety of speed characteristics of the tested SSD.
The results you see here refer to the FOB (fresh out-of-box) non-degraded SSD performance. Moreover, we use incompressible data, which is formally the least favorable scenario for the LSI SF-2281 controller that employs on-the-fly data compression. Our tests show, however, that in today’s world when the data may only be partially compressed and the utilized flash memory has high-speed synchronous interface, the compression algorithms do not have a big effect on the real-life performance of SSDs with SandForce controllers. Therefore, we gave up the idea of testing SandForce-based SSDs with compressible data: These results would be exclusively artificial in nature and wouldn’t have any practical value for us today.
It’s interesting to watch the SanDisk Ultra Plus perform in the synthetic benchmarks. Despite its 4-channel design, it is as fast as the other modern SSDs. The reduced number of controller channels doesn’t affect the read speed of the new drive as its performance is limited either by the SATA 3 interface (in the case of sequential reading) or by the controller (in the case of random reading). As a result, the SanDisk Ultra Plus is comparable to the Plextor M5 Pro, which uses an 8-channel Marvell controller, when reading sequentially or randomly with a short request queue. The downsides of the 4-channel access only show up at a long request queue or when reading large random-address data blocks but such loads are not typical of regular PCs.
When it comes to writing, the SanDisk Ultra Plus makes use of its nCache technology, which is expected to make up for the fewer number of controller channels. And it does. At least, the SanDisk Ultra Plus doesn’t betray its different internal design under typical real-life loads with a short request queue. It is only at a long request queue that it slows down, but such loads can hardly occur in ordinary PCs.
So we can find no problems about the SanDisk Ultra Plus if it is used for everyday rather than server applications.
Unfortunately, SSDs are not always as fast as in their “fresh” state. In most cases their performance goes down after some time and in real life we deal with completely different write speeds than what we see on the diagrams in the previous chapter of our review. The reason for this phenomenon is the following: as the SSD runs out of free pages in the flash memory, its controller has to clear memory page blocks before saving data into them, which causes substantial delays. Therefore, contemporary SSDs usually try to free the memory in advance, and not when writes are underway. This usually happens in idle mode. At this time SSD controller can alleviate the performance drop almost completely by erasing unused flash memory pages ahead of time. The corresponding procedures are usually performed in idle mode, when the controller can fully restore the SSD performance by clearing out the unused flash memory pages. They use two techniques for that: idle-time garbage collection and TRIM.
An SSD controller doesn’t know which memory pages contain user data and which are considered empty by the OS. It happens this way because in file systems removing a file doesn’t involve its actual physical removal. Instead, the corresponding memory is marked in the file system as available for rewriting into. So, without involving the OS, an SSD controller can only pre-erase pages in the reserve pool (if it exists), which is not accessible by the OS. For a better solution to this problem, modern OSes have the TRIM command which improves garbage collection the efficiency. TRIM provides the SSD controller with information on which data could potentially be removed without any harm, as it is considered unused by the OS. As a result, the SSD controller can increase the cleared pages pool by physically removing unneeded data so that the user didn’t feel a performance hit during subsequent write requests.
This is how it should be in an ideal world. In reality, however, SSDs differ in their garbage collection and TRIM implementation. That’s why we check out the performance hit an SSD suffers when transitioning from its out-of-box (the flash memory is clean) to steady-state. This test follows the SNIA SSSI TWG PTS guidelines, which means that we measure the write speed in four cases one by one. First we measure the “fresh” SSD speed. Then we measure the speed after the SSD has been fully filled with data twice. The third test occurs after a 30-minute break during which the controller can partially restore performance by running the idle-time garbage collection. And finally, we measure the speed after issuing a TRIM command.
We ran the tests in synthetic IOMeter 1.1.0 RC1 benchmark, where we measured random write speed when working with 4 KB data blocks aligned to flash memory pages at 32 requests queue depth. The test data were pseudo-random. The following diagram shows the history of the relative speed changes, where 100% refers to the SSD performance in “fresh-out-of-box” state.
The SanDisk Ultra Plus is based on a Marvell controller, so we have no reason to suspect any problems with its garbage collection algorithms. And that’s indeed so. The TRIM command restores this drive’s performance to its original out-of-box level. Its background garbage collection is not so efficient, though. Plextor SSDs with Marvell controllers are much better in this respect. There’s a simple explanation: the SanDisk Ultra Plus has a small reserve pool, most of which is utilized for the nCache technology. So if you’re going to use this SSD in a non-TRIM environment, you may want to leave some of its storage space unpartitioned to improve its background garbage collection.
The popular PCMark 7 contains an individual disk subsystem benchmark. It is not a synthetic test, but is based on real-life applications. This benchmark reproduces typical disk usage scenarios and measures how fast they are completed in popular tasks. Starting with version 1.4.0, the PCMark 7 disk subsystem test generates raw performance results which do not take into account any pauses in the requests queue. New results are thus incompatible with old ones, but the differences between the performance numbers of different SSDs have now become more obvious. That’s why we decided to switch to the new version of the test from now on.
We ran PCMark 7 on “steady” SSDs, which is what they are going to be in actual computer systems most of the time. Their performance in this case is affected not only by their controller or flash memory speed but also by the efficiency of their internal algorithms that fight performance degradation.
The SanDisk Ultra Plus is very good here. In fact, it is only inferior to the flagship products like Samsung 840 Pro, OCZ Vector and Plextor M5 Pro (we don’t count in the high results of the SandForce-based SSDs, which reflect the specifics of PCMark 7 rather than their real-life performance). So although SanDisk positions the Ultra Plus as an inexpensive solution, it is actually faster than the majority of products of its class, at least according to PCMark 7, which tries to simulate real-life disk usage scenarios.
Let’s figure out what happened. The total PCMark 7 score is a more generalized performance metric. We will check out the individual tests to get a more detailed picture of what our SSDs are capable of under various types of operational load:
We’ve mentioned above that SanDisk recommends its new SSD for multimedia applications. We don’t quite agree with that recommendation. The Ultra Plus doesn’t show anything exceptional with multimedia files. It is only good in the sense that the alternative SanDisk Extreme is based on the SandForce platform and slows down on incompressible data. As opposed to it, the Ultra Plus delivers consistent performance across all of the tests. Its performance is only lower than we might expect in the Starting Applications and Gaming tests which are important for home users. It is slower than the inexpensive Plextor M5S in both cases.
Intel NASPT is another disk sub-system test that employs real-life usage scenarios. Like PCMark 7, Intel NASPT reproduces predefined disk activity traces and measures how fast they are executed. However, the default traces are designed for network attached storage devices rather than for SSDs. Therefore during our test session we replace them with the specially developed SSD Benchmarking Suite which offers more relevant usage scenarios such as compressing and decompressing files, compiling large projects, copying files and folders, loading 3D game levels, installing software, batch-processing photos, searching a digital library for data, mass-launching applications, and transcoding video.
Like PCMark 7, this benchmark gives us a true-to-life illustration of disk subsystem performance. Here the SSDs are again tested in their “steady” state.
Intel’s NASPT seems to provide a more realistic picture of SSD performance, and the SanDisk Ultra Plus has a rather high overall score. Like in PCMark 7, it is only inferior to the recognized leaders such as the Samsung 840 Pro, OCZ Vector and Plextor M5 Pro. The other SSDs from the same price category as the SanDisk Ultra Plus have lower scores.
Besides the average benchmark score, we would also like to offer you the results of individual usage scenarios, which will show where new SanDisk SSD can really shine. Note that the data-transfer rate is higher than the SATA 3 interface bandwidth in some subtests. That’s because INASPT is a high-level test that uses standard Windows functions to access the disk subsystem. As a result, the OS caching mechanisms also affect the results.
The SanDisk Ultra Plus is especially fast in two cases: when copying a large file from the SSD and when installing a large software bundle on the SSD. Apart from these, the Ultra Plus is no better than the Intel 520, Samsung 840 and Plextor M5S. So it is an interesting mainstream SSD but there’s nothing exceptional about it despite the averaged scores.
We use AS SSD version 1.7.4739.38088 test to benchmark the speed of copying different types of files within a single partition the size of the whole SSD. The SSDs are tested in their steady state.
Copying involves reading and writing files concurrently, and the SanDisk Ultra Plus shows average performance in this test. It is comparable to the OCZ Vertex 4 and the Plextor M5S, which is quite a good result considering its architecture.
We’ve been meeting some interesting products lately. Following Samsung’s new SSDs, the SanDisk Ultra Plus has proved a very exciting object for us to test. That’s why this review has a generally favorable attitude towards it. Any innovation that stands out from the crowd deserves our initial sympathy irrespective of the end result.
The Ultra Plus SSD is unusual for having a quad-channel Marvell SS889175 controller but making up for the lack of controller channels by using some of its MLC memory in SLC mode. And, according to our tests, this solution can be praised not only for being innovative. It is no flagship product and is not positioned as such but it looks competitive as an entry-level SSD. It is comparable to the Plextor M5S, Corsair Neutron and OCZ Vertex 4 in average performance. It copes well with typical PC loads, slowing down just a little if used as a system disk. The SanDisk Ultra Plus comes with a standard 3-year warranty but the nCache technology can be expected to make its service life longer compared to its competitors.
SanDisk should also be commended for things other than the engineering work. The Ultra Plus is currently one of the least expensive drives on the market, together with the Samsung 840. It is a very attractive product in terms of price/performance ratio. Moreover, it has high-quality 19nm MLC NAND flash with Toggle Mode interface, so we can recommend it to people who are still wary of TLC memory with 3-bit cells.
To help you with your choice, we offer the following summary table with test results of various SSDs. It contains basic hardware information about SSDs we’ve tested so far and allows to quickly determine the general position of a particular model among its competitors in terms of relative performance.