From novelties modest and often disappointing, such as Intel 540s, we move on to the more serious class SSD, the benefit of those appear in this year, albeit in a small range. With SSDs with PCI-Express interface in general is somehow tight. Samsung last year unveiled its 950 Pro and since then calm resting on its laurels, especially without straining – is no alternative to the last time this decision was not.
Micron swung to its Crucial Ballistics TX3, canceled but its release at the last minute. Plextor, once produced the first mass consumer M.2 PCIe SSD Plextor M6e, also canceled its M7e, only recently announced Plextor M8e. Kingston – public promises HyperX Predator NVMe release with all the trappings such as PCI-E 3.0. But this – later. And now – recently rolled out for all to see old HyperX Predator, we tested 16 months ago, just in a new volume of 960 GB. Phison – here and all wonderful: the controller Phison E7 ready, it works, but no mass expansion is not in sight – this creation is only used in one consumer SSD is not the most popular, to put it mildly, the brand Zotac.
And only one company swung to seriously compete with the Samsung 950 Pro. Name it – Toshiba. Its creation came to light only recently, and even the origins of his somewhat vague: and she Toshiba actively and boldly experimented with different controllers JMicron, SandForce, Marvell, and the OCZ, which was some time ago acquired by Toshiba, is also preparing its own controller, codenamed ” JetStream ». But the historical roots even understand those who are interested, but the average consumer interested in a purely practical side: Will Toshiba RD400 will impress performance and compete with widely advertised Samsung 950 Pro? Review and test processor AMD Athlon X4 845: Table excavator
Quietly and without much pomp AMD at the end of February led to a new segment of the desktop processor microarchitecture codenamed «Excavator». We will understand more with this processor.
In fact, the test sample was in my early as the end of June, but the conduct of the tests, it was decided to postpone until the moment has not yet come ordered Samsung 950 Pro 512GB. It is considered the king of the market, and without comparing them with an review Toshiba OCZ RD400 loses most of its practical meaning. But life imposes its own limitations, and at the moment when Samsung 950 Pro has arrived and it became possible to carry out the tests myself because of personal problems had to completely stop their work with articles.
OCZ. With this name, users connected, and a lot different. At the same time – not all is positive: the level of marriage in the SSD-drives produced by the OCZ, became the talk of the town and so far halloo to one another company, even though it already and independence lost, becoming a part of Toshiba, and an assortment of repeatedly updated completely (and recently himself Indilinx Barefoot 3 controller is sent to the rest).
But that does not take away from the OCZ, so that what the company is first to market with serial retail SSD with PCI-Express interface. Not just for show, as, for example, the ASUS with its ASUS RAIDR Express, released in a modification of 240 GB, namely a full retail product.
If my memory serves me (the case more than seven years ago, when the likes of SSD I have no interest, limited to 2.5 “and SATA), chronologically the first was a series of OCZ Z-Drive: a massive expansion card with PCI-Express x8 interface in 2009. The design was very peculiar: PCB was have a RAID-controller with associated strapping and metal casing, which recorded two more printed circuit boards, on which were placed the controllers of NAND-flash-memory (there were versions of OCZ Z-Drive with SLC and MLC- memory). And even then the proposed volume of up to one terabyte.
Soon followed a series of OCZ IBIS. Also very clever solution: the drive is physically composed of two parts – a separate board PCI-Express x4 slot with a special HDSL, to which, by means of a special cable, connected the drive itself, made in the form factor of 3.5 “and is installed in the battery bay. There were about ten modifications available in quantities of 100 to 960 Gbytes.
And, finally, the first OCZ RevoDrive. Now it has been more than a simple device that is installed in the expansion slot PCI-Express. But this series was also no less conspicuous in its design: on one printed circuit board is physically housed two separate SSD based on the SandForce SF-1222 combined into a RAID-array with an additional controller. Please note: as recently advocated Silicon Image 3124, which was equipped with a PCI-X interface, not PCI-E. And to ensure that the drive can work with the PCI-E bus, PCB installed also additional chip converter Pericom PI7C9X130DNDE. Impressive design.
But engineers OCZ and it was not enough. At the RevoDrive line subsequently appeared constructive under the symbol “the X2”, is a “sandwich” of two such printed circuit boards assembled in a single structure (the Laboratory overclockers.ru, for example, visited the OCZ RevoDrive 3 X2).
This was followed by further modernization resulting from shifts controllers and flash memory. Last the RevoDrive, released under the brand OCZ, submitted in April 2014, OCZ the RevoDrive 350 “non-standard the principle of” comply strictly: the hardware is RAID-array of several LSI independent of NAND-controller SandForce SF-2282 and the array of flash memory, with a total RAID Controller OCZ ICT-0262 (under this label, presumably hiding Marvell 88SE9548).
But this principle is not too advisable for full-scale competition in the market in economic terms, there is need to create a simple and versatile solution. And there is a new standard that allows the creation of very compact high-speed SSD – M.2 (nee – NGFF), began to receive wide distribution, being installed on many motherboards, not just in mobile devices. Add something like OCZ RevoDrive on the printed circuit board size of 22 to 80 mm can not be physically.
OCZ Company began developing a new controller, code name «JetStream», samples of which started in the near future on display at various exhibitions, such as at CES2015. Development was delayed for a long time, for a while gone even rumors of this controller.
In June 2015 the company Toshiba, which has already been the OCZ owner, showed samples of the Toshiba drives range XG3 made in the form factor M.2 2280 and even then, it was recognized that it was XG3 will be the basis for future OCZ RevoDrive 400. The future begins fairly long: Toshiba OCZ RD400 was introduced almost a year later – at the end of May 2016. Another one and a half months of the new product made it to our retailers.
Toshiba OCZ RD400 comes in two versions: only the drive, and complete with a card adapter for installation in a PCI-Express slot. In the latter case, the drive letter is added the title «A» (Toshiba OCZ RD400A). There are four variants of volumes: 128, 256, 512 GB and 1 TB.
It is interesting that the fastest version of the modification is 512 GB. Version terabyte – a little slower. Period original warranty is five years, but the Toshiba standard established additional restrictions: no more than 74, 148, 296 and 592 terabytes of stored data, depending on the drive capacity.
Page on the website of the manufacturer: Toshiba OCZ RD400 512GB (RVD400-M22280-512G).
Price (at time of publication):
In the box the purchaser of this option detects the shape of transparent plastic with information booklets.
Also included is embedded special shortened strap that allows the drive to be installed in a low-profile adapter HHL housing system blocks.
In the presence of the adapter-adapter M.2 >> PCI-E x4 (HHL) is the fundamental difference between the solutions from Toshiba Samsung 950 Pro: the user will not be forced to search for and purchase the adapter separately if on available to it do not have the motherboard seat M.2. And it’s not just the convenience of simultaneous purchase: There are several variants of execution of such adapters, key among which are two. Their main difference – in the food chain in laden with drive adapter. One type – Simplified, when the drive gets all it needs a voltage directly to PCI-E connector. The second option is more complex, in which the power is filtered and further stabilized, the drive is protected from interference and interference caused by the other expansion cards installed in adjacent slots (eg, video cards). Bundled with Toshiba OCZ RD400A offered it a second.
For such a “pleasure» Toshiba offers to pay exactly $ 20, which is quite adequate realities: if scour the most popular trading platform eBay and Aliexpress, you can find that kind of money, and it is similar to the adapter, if purchase it separately. Cheaper options with a range of prices starting from about $ 4 are of the type with a simplified power.
Does it make sense to try to save money? The answer to this question is not so simple: in fact, it all depends on the quality of performance of the motherboard, which is supposed to operate the drive. From my own experience and the results of practical experiments in the power of my opportunities and access to different sets, I can say that in most cases there is no problem. Specific names I will not mention that even two motherboards the same model from the same batch may not work properly, and the third – the drive can be lost from the system. Therefore, there is always a risk, and this should be ready.
Note that Toshiba did not object to the parse RD400A: here is not applicable any seals, the violation of which deprives the firm guarantee of the drive.
Thus, the drive can be removed if necessary and the adapter installed in the motherboard.
When parsing device can detect that Toshiba OCZ engineers attended an additional heat sink, between the printed circuit board and storage device adapter circuit board is made of soft material lining. Moreover, its position coincides exactly with the position of NAND controller.
In other words, it is understood that the user can wait for problems with temperature control. But that’s the element base, which was built Toshiba RD400, is placed only on one side of the PCB and covered with a label, the other – completely empty.
Under the label hides all four major chip: NAND-controller labeled Toshiba TC58NCP070GSB, two-chip flash memory Toshiba TH58TFT1JFLAEG chip buffer memory and DRAM Low Power DDR3-1600 Samsung K4E4E324EE-EGCE 512MB.
Each chip flash memory with Toshiba TH58TFT1JFLAEG marking carries sixteen (!) MLC NAND Toshiba crystals of 128 Mbps with a planar arrangement and executed on 15-nm process technology. The total volume of the array – 512 GB, but part of it is allocated to hidden reserves, and the amount indicated in the decimal system (used to indicate the volume of 1 GB is equal to one billion rather than 1,073,741,824 bytes). Therefore, in reality, the user is available only 476.94 GB, the remaining amount of the firmware of the controller operates for official purposes: for wear leveling, as reserve pool for the replacement of defective memory cells and others.
Thus, we have an array of 32 NAND crystals that according to available information, a structural limit for Toshiba same Toshiba OCZ RD400 1 terabyte is made in a form factor flash memory chips on both sides of the PCB. This may explain the fact that there is no more capacious versions of this drive.
Toshiba adheres quite original drives manufacturing policy: the company is not simply gets ready controllers from third-party developers, and modify them independently, making their own changes both at the program level in the firmware and the hardware, for example, the packaging of the controller chip and the DRAM buffer memory in a single housing and, often, changing the frequency of its work.
The resulting chips get their own markings Toshiba, and the company does not disclose technical details about the origin of the controller and only by circumstantial evidence have to wonder about what we’re dealing with. Identification is complicated by the fact that Toshiba has in recent years been able to include in its asset contracts with almost all the major developers in the industry: JMicron, SandForce, Phison and Marvell. And with the purchase of the Toshiba OCZ he had access to the documentation on the Indilinx Barefoot controller family and above OCZ JetStream.
Before us is PCI-Express storage running on a logical protocol NVMe 1.1, which cuts Indilinx Barefoot. With SandForce situation at the moment, with all resales group of developers, rather vague, ready-made solutions based on them in retail is still there, so they disappear. OCZ JetStream controller Fate is also unclear: in February 2015, spoke of him as the product is not ready for mass production in the coming months. Thus, taking into account the release date Toshiba XG3 four months later, he disappears and. The JMicron, although more than once showed working prototypes, retail and did not go.
Cooperation with Phison allowed Toshiba to release OCZ Trion 100 and OCZ Trion 150 (now Toshiba OCZ TR150) and it has E7 Taiwanese developer. But the scheme of reasoning here is about the same: is almost the end of 2016 and PS5007-E7 can be seen as a part of only one single drive – Patriot Hellfire, the volume of production an distribution which hardly allow us to call this controller serially manufactured product. Moreover, at the time of this writing, he began to disappear from shops and online Patriot he is already listed, and all unavailable for purchase.
Thus we are left with only one developer – Marvell. That he is the main (if not only) for Toshiba partner for several years in the production of SSD-drives. And there Marvell 88SS1093 codenamed «Eldora». It is well suited for terms debut (August 2014), and physically – visually seen chip samples coincide with Toshiba TC58NCP070GSB.
Marvell Eldora, until recently, was also not present in the retail (only recently begun Plextor M8Pe delivery). But that’s about the industrial segment do not say, is enough to recall, for example, Lite-ON EP2.
And with the simplest test Toshiba OCZ RD400 instantly detectable characteristics inherent Marvell 88SS1093: achieve maximum reading and writing speeds even on sequential access operations only possible if there is the request queue depth with some value clearly different from unity. If we talk about a simple single-threaded load, there RD400 figures slightly differ from the SSD interface PCI-E 2.0 x4. Visual illustration – to launch Crystal Disk Mark versions 3.h.h and 5.h.h.
To be more precise, its ceiling performance on the linear reading Toshiba OCZ RD400 512GB reaches only when equal to eight or sixteen requests queue depth, or four threads simultaneously.
Quite a specific feature, which is unrealistic in the “home” conditions and which, incidentally, is also available in Samsung UBX, underlying the 950 Pro and SM951, but he depends on it to a lesser extent:
Do not forget one more feature. We have before us a solution PCI-E NVMe, but because here in full growth there is a problem of backward compatibility with older systems. And the word “old” is meant not something like LGA775 and Socket AM2. We are talking about quite yet urgent Intel Z77 and X79: not all motherboards based on them may recognize such drives, and even more so – to boot from them. And do not confuse these concepts: drive and can be identified, but boot from it may not be possible. Speaking of motherboards based on the AMD system logic, then there is still sadder: I know only one motherboard that is able to work effectively with these SSD – ASUS 970 Pro Gaming / Aura. With it, I did not have any problems.
After loading the operating system (Linux or Windows – does not matter) NVMe driver installed on the motherboard with the above constraints drive will be available, and if the system supports PCI-Express 3.0, and then the speed will generally be appropriate. This is the case with Toshiba OCZ RD400 and my operating system based on the ASUS ROG Rampage IV Extreme.
Thus, in this case the drive can be used as a work.
By the way, the driver NVMe its own characteristics, but we’ll talk about a little later.
Software support for Toshiba OCZ RD400 worthy: drive is accompanied by a proprietary software package OCZ SSD Utility. The application is fully and accurately translated into Russian and is not experiencing any difficulty in identifying and working with the store (which, for example, can not be said about the universal application Crystal Disk Info).
You can get detailed information on the state of the drive, the interface, the SMART parameters, wear and temperature. There is even an identifier hardware configuration (although it is not useful to the ordinary user). You can create a bootable USB-drive, the image of which is loaded with Toshiba OCZ server, but two disadvantages: the image from the server downloads are on a very low speed and download files from the test stand I could not – it hangs during initialization of the equipment. Is the state of the control system of the drive and the possibility of mail when problems occur. You can allocate an additional amount of flash memory in addition to the factory reserve and update the firmware (such as a file, and with Toshiba OCZ server). There is a built-in test performance, although his testimony is not too stable, and therefore can not be considered absolutely reliable and more suitable for the general correctness of an assessment system assembly and setup.
The application is accompanied by a help system, also translated into Russian, but she soon realized in the form of tips than in the form of a full application.
The SMART drive is present in all 15 parameters.
Where you can find a counter showing the time in hours drive, the volume of written and read by the data interface, and others.
It would be a blessing in disguise. While the drive was in anticipation of the review of training, some tests were carried out in fits and starts, and in particular, during this time came out a few versions of the Crystal Disk Info, which had consistently run to update the illustrations to the article. And find out one interesting feature: version 7.0.1 and 7.0.2, and even “know» Toshiba OCZ RD400, but show it in the system through time.
With protocol NVMe, more precisely, the corresponding driver of the features. By default, Windows operating systems 10, 8 and 8.1 add-nominally staffed initially available for Windows 7 operating system specific update. But the driver for Microsoft authored no different performance, but because Toshiba OCZ (like Samsung) has released their version of it, you need to manually download and install Toshiba OCZ site.
For some users will be interested in the question of the work of several SSD in a single system, because the already released today a considerable number of manufacturers of motherboards with two or even three seats M.2. Clarifying the point: everything works fine, no conflict is detected either using the Microsoft full-time driver or drivers OCZ and Samsung.
Unfortunately, this motherboard with the possibility of building a RAID-file is not available, so we had to manage the program by means of the Windows operating system by installing a SSD into the slot from the processor, and the other – in the slot by chipset. Just at least a little to look at the potential of such a configuration:
The full set of tests not carried out due to the fact that it is, in terms of the accuracy of the indicators are not completely reliable. Incidentally, the Samsung driver can be manually set for Toshiba OCZ RD400 and despite the operating system warnings, such a bundle is quite efficient.
With CrystalDiskMark (64bit) 3.0.1 in random mode, the data is metered performance four times:
The initial state of the new drive ( “zero” state);
After the entire test cycle is filled with data from a disk user article previous partition (Word files, photos, audio and video), so that the total amount of data has been recorded at least three times the total volume of storage;
Thirty minute easy, for which is not made any transactions with the SSD – for background functions Drive Cleaning “garbage”;
Thus, we can see how well the drive firmware is up to the task of maintaining performance levels on low volume simultaneously written and read data – for the operating conditions in the domestic environment, this is enough.
Next, the complete cleaning of storage by applying Secure Erase command, and then run HDTune test in «Write» mode (block size is set to 1 MB) – This test produces a linear record of all volume of the support, simultaneously displays information about the recording process in an easy schedule. This test allows us to see how the whole drive is stable, if overheating occurs and what might algorithms “fast recording” implemented in the firmware.
Finally (and after the command Secure Erase) performed testing using Iometer.
Simulated work load drive in conditions close to the server (continuous random 4K recording units throughout the volume with depth inquiries line 32) in the absence of TRIM. That is how, for example, the database work: create one or unspecified number of large files, which are carried out within the read / write operation, the TRIM command generation is not happening. The test is carried out continuously for two hours with every second performance indicators are removed. The results of this test allow us to see the possibility of an experimental in its “pure” and the “use” conditions (achievement of “performance established” state).
Upon completion of this test, which has done one more goal will be finding out how well the algorithms “garbage collection» (Garbage Collection). At the final graph present speed performance drive in four situations: a “clean” cell array, after the continuous load for two hours in the absence of TRIM command, after being idle for 30 minutes, which should be sufficient drive for development of internal algorithms “garbage collection”, after TRIM command execution on the entire volume of the drive. The test is quite specific, and its results are important for those who aim to operate in the conditions of work without TRIM.
Yes, Crystal Disk Mark old version 3.0.3, rather than the new 5. The reason for this is the same, which we discussed above: multi-threaded load line operations – rare for a home PC, and therefore the situation will be synthetic, whereas we are interested in the practical side. We see that the Toshiba OCZ RD400 copes with demanding workloads and any claims to drive does not occur.
Drive at a linear recording does not use any “speed up the entry” algorithms, otherwise known as “the SLC-caching.” Recording speed is kept at the same level in the 1450-1500 MB / s. About the same speed we have seen in the actual copying of large files, Windows forces.
Guard sharp dips and fluctuations at the end of AIDA64 chart after recording approximately 415 GB. The reason is simple: overheating. This is despite the fact that the test was carried out under intense directional blowing.
If the airflow to remove, drive capabilities becomes more modest: thermal throttling starts to function after recording approximately 265 GB of data.
But why “only”, “only” and “just”? Just think about these numbers: talking about gigabytes of data records as measured by three-digit numbers. Continuously, one “unit”. At home, the generation load of this nature is akin to science fiction. If we talk about more than the usual load block, then there is practically impossible to achieve overheating.
What is the nature of the load is most typical for normal home PC? A small building block operation request queue depth (typically – less than 8.4). We generate this kind of a burden, but for reassurance, toughen scenario: building block will be recorded with random access 4KB blocks with the request queue depth of 16.
Testing is performed by the Iometer, the size of the test file – 32GB, test duration – 10 minutes. Control of the temperature sensor readings by using specially created at my request, one of the permanent members of the Conference overclockers.ru console application. However, its drawback is that at the moment it only works with standard Microsoft driver, the included Windows operating system, so they had to switch on the driver to carry out measurements. Readout occurs at an interval of 1 second.
In this 10-minute continuous test mode does not raise any problems with the speed of the fall, and not write about 303 GB of data, and only the drive warms up to +65 ° C and stabilized at that value.
Several complicate our task: remove the drive from the standard adapter which, as we know, is equipped with an additional heat-conducting material, and set to the normal adapter which was purchased on eBay (the most simple type). What will be close to the installation of the drive in the seat M.2 on the motherboard.
On used for testing the motherboard has a seat M.2, but due to some peculiarities of the implementation of it, could not be considered to set the temperature in his drive. Therefore it was necessary to use an adapter.
Tighter operating conditions leads to the fact that the amount of data that can be written without signs of throttling is reduced to 71.5 GB, which is roughly comparable to Samsung 950 Pro 512GB under similar conditions. Score a lot more modest, but it is still so great that this drive user explicitly does not have to worry about the presence of any problems.
The screenshots above were visible temperature + 50 ° C. It’s not a bug, but is true: our sample on an open testbed without blowing and load steadily demonstrated that temperature, and only under the direction of blowing going to fall to 32-35 ° C. But no problem it does not deliver.
In the building block random write performance Toshiba OCZ RD400 512GB we were able to get up to 170,000 IOPS. What is interesting, it is somewhat lower than the stated 190 000. In this case, the drive can not boast of a perfect regularity and speed characteristics if we consider the average of 30 seconds (black line in the graph), we get a more modest 120,000 IOPS.
After filling the entire array of free storage memory cells becomes “settled state” and its performance gradually stabilized at the level of 22,000 IOPS by the fact that the drive has to be cleaned before burning cells. This is typical behavior for a SSD-drive, except maybe the fact that no consumer-grade models even in its “pure” state able to provide the 22,000 IOPS.
The lack of TRIM command – a situation for PCI-E NVMe-drive is very unlikely due to the fact that even the driver himself NVMe is only available for those operating systems that “know» TRIM command natively. That is why it is logical to expect that Toshiba OCZ RD400 and not adapted to work in such conditions. More specifically, after the drive is able to take downtime approximately 1.1 Gbytes, but not complete, but at high speed. Full recovery speed and full maintenance performance level is carried out only with TRIM.
There was a move not only to the new configuration of the test stand, and a new operating system. And with this move we have a problem in this set of tests: the previously used TeraCopy program in Windows 10 environment showed inadequate results. It was therefore decided to abandon it.
Henceforth tests copying groups of files will be performed forces the operating system itself. For this batch file has been written that automatically copies files and records the time spent on the operation.
Pretty important attribute is the speed of data access time. It is understood that modern SSD drives in this regard have already reached such values that this issue will be of more academic interest. Average access time during reading and writing operations was obtained as a result of AS SSD Benchmark test version 1.7.4739.38088.
Almost a year on the market of retail NVMe SSD reigned duopoly of the two solutions: Intel 750 and Samsung 950 Pro. Moreover, direct competition, as such, was not among them. The first drive is inferior in speed and produced only in the form factor expansion card with PCI-E 3.0 x4 interface and 2.5 “SATA-Express interface, but offered volume up to 1.2 TB, and the second – only M.2 2280 and volumes of 256 and 512 GB, but with a higher level of performance.
Performance. With the goal to push the Samsung 950 Pro on the second place, Toshiba failed. Most drives are balanced on one bar, speaking with alternate success, with some margin in favor of Samsung models in custom application software. Creation Toshiba is more suitable for server workloads, rather than domestic.
Price. According to this criterion Toshiba OCZ RD400 looks interesting Samsung 950 the Pro, but worse Plextor M8Pe, who allegedly does not concede to the hero of the review to the fact that it uses the same hardware platform. Although, of course, the differences may well be due to, for example, firmware.
Operational and consumer characteristics. There Toshiba OCZ RD400 looks preferable by complete adapter, further modified for better heat dissipation (if we talk about RD400A). Do not forget and about the safeguards that are accompanied by solutions under the brand OCZ. For the Russian consumer, this means that you can never be in possession of any documents on the acquisition drive, just contact tech support and make the exchange of problematic devices. No Plextor, Samsung no such service is not offered.
However, no matter how it sounded unexpectedly, computer enthusiasts seeking uncompromising in terms of performance, it’s hard to recommend the acquisition of some of these SSD. The fact that Samsung has already established serial deliveries of new Samsung SM961. And in the near future the market will start to arrive Samsung 960 Series drives in versions with MLC NAND (960 Pro) and TLC NAND (960 Evo). All three offer a new line of Samsung Polaris controller and read speeds of up to 3500 MB / s. One of the first owners of the SM961, by the way, was marked last night in our conference, to share the results of tests.