Although SandForce was formed in 2006, the company didn’t infiltrate PC enthusiast circles until last year, when its SF-1500 controller broke cover in OCZ’s Vertex 2 solid-state drive. SandForce entered a market populated by big names like Intel and Samsung, and the upstart managed to impress with a unique controller design laced with an intriguing black box of compression and encryption technologies known as DuraWrite. Before long, controllers from the SF-1000 family could be spotted in a wide range of SSDs from OCZ, Corsair, Mushkin, Patriot, and others.
Now, more than a year old and restricted by 3Gbps Serial ATA connectivity, the SF-1000 family appears to be on its way out. In its place, SandForce has developed an updated line of 2000-series controllers destined for next-generation SSDs. Fittingly, the family is making its desktop debut inside OCZ’s Vertex 3.
The Vertex 3 isn’t due to hit store shelves for another couple of weeks, but it’s arguably the most anticipated SSD of the year. Indeed, we’ve been so curious about the drive’s performance that we’ve been testing a pre-production sample with firmware that hasn’t yet been finalized. OCZ expects the firmware only to get faster from here on out, so we’ve decided to take a closer look to see where it’s at now.
Before dipping into our benchmark results, let’s unravel the drive to see what makes it tick. The Vertex 3 features an SF-2281 controller drawn from SandForce’s SF-2000 family of “client” SSDs. SandForce also makes enterprise-oriented controllers based on the same architecture, but they’re targeted at servers rather than notebooks and desktop PCs. This isn’t a case where power users need to be dipping into the enterprise pool, either; SandForce names enthusiasts as a specific target market for its client family.
Perhaps the most notable feature of the SF-2281 is 6Gbps Serial ATA connectivity. The faster interface nicely matches what’s being offered by the latest SSDs from Crucial and Intel, but it’s more than just a checkbox feature to keep up with the Joneses. According to SandForce, the SF-2281 offers sequential read and write speeds up to 500MB/s—more than enough to saturate a 3Gbps SATA pipe. OCZ is even more optimistic about the Vertex 3, which it says can push 550MB/s in sequential reads and 525MB/s in writes.
|Flash controller||SandForce SF-2281|
|Flash type||Micron 25-nm MLC NAND|
|Available capacities||120, 240GB|
|Sequential writes||500MB/s (120GB)
|Random 4KB reads||60,000 IOps|
|Random 4KB writes||60,000 IOps (burst)
20,000* IOps (sustained)
|Warranty length||Three years|
For a system drive, the ability to handle a lot of random read and write requests is arguably more important than blazing sequential throughput. According to the SF-2281’s spec sheet, the controller is capable of chewing through 60,000 IOps with random 4KB reads. The same IOps rating applies to 4KB random writes, but only in short bursts. With a sustained random-write load, the controller’s performance rating drops to 20,000 IOps, at least according to SandForce. OCZ instead prefers to say that the Vertex 3 can sustain 60k random-write IOps until the drive gets “dirty.” After that, garbage collection will kick in and random-write performance will drop—how much it drops depends on a variety of factors, including how the drive has been used up to that point.
Interestingly, OCZ tells us the 120 and 240GB flavors of the Vertex 3 should perform almost identically. The 120GB variant is limited to 500MB/s sequential writes, the company says, but all other performance specifications are the same. 120GB may be as small as the third-gen Vertex gets; there’s no indication 60GB or lower-capacity variants are in the works.
Like the SF-1000 series, SandForce’s new SSD controllers are designed to operate without the aid of dedicated DRAM cache memory. Internal buffers within the controllers provide a measure of built-in cache capacity, but I suspect they’re not nearly as large as the 128-256MB of cache available on competing SSDs. The cacheless SF-1000 family was pretty darned fast without, so I wouldn’t worry too much about the lack of DRAM on the Vertex 3.
Besides, the drive’s NAND flash memory chips are much more interesting. The SF-2000 family is compatible with a wide range of flash technologies, including ONFI 2.0 and Toggle DDR specifications. SandForce says per-die transfers top out at 166MT/s, a figure that becomes considerably more impressive when you factor in the controller’s eight memory channels.
On the Vertex 3, OCZ pairs the SF-2281 with MLC NAND courtesy of Micron. Most of the flash found inside contemporary SSDs is built using 34-nano fabrication technology. The Vertex 3’s chips are fabbed on an advanced 25-nm process that squeezes more dies into every square inch of wafer area. Provided there are no problems with yields, these chips should be cheaper to purchase than their 34-nano counterparts, leading to lower SSD prices. There is a catch, though. NAND chips fabbed on finer process technologies tend to withstand fewer write-erase cycles than those fabbed with older process tech. OCZ says that each of the Vertex 3’s flash cells is good for 3,000 write-erase cycles.
SandForce controllers are a shining example of why a given flash technology’s write-erase endurance shouldn’t be viewed in a vacuum. The number of flash cycles that each cell can survive is but one part of a larger equation that defines how much data can be written to a drive over its usable life. Another important variable is a controller’s write-amplification factor, which refers to how much data is actually written to the flash for each write request that comes from the host. SSDs typically have a write-amplification factor that’s greater than one but very close to it. Thanks to a suite of compression techniques, SandForce controllers can write less to the flash than is asked by the host, resulting in a write-amplification factor that’s less than one.
DuraWrite is the name SandForce gives this black box of compression mojo, and the company has been reluctant to give us a closer look inside. We suspect there’s an element of encryption involved, since SandForce controllers offer on-the-fly AES encryption that can’t be disabled. The latest SF-2000 series support 256-bit AES encryption, which is a nice upgrade over the 128 bits of protection offered by the SF-1000 family.
DuraWrite’s ability to lower a drive’s write-amplification factor is especially valuable as the market transitions to flash chips that tolerate fewer write-erase cycles. Another unique SandForce technology that’s well-suited to potentially flaky flash is RAISE, which stands for Redundant Array of Independent Silicon Elements. Similar to RAID 5, this redundancy scheme dedicates the capacity of one flash die to storing parity-like information. You lose a little capacity in the process, which is why RAISE is optional for the SF-2000 family. With RAISE disabled, the SandForce controllers rely on ECC error correction capable of recovering up to 55 bits per 512-byte sector.
RAISE is enabled on the Vertex 3, which uses a typical 7% overprovisioning percentage to offer 224GB of Windows storage capacity on a drive with 256GB worth of NAND. The Vertex 2 debuted with a much higher 28% overprovisioning percentage familiar from enterprise-class SSDs. Having more “spare area” at the controller’s disposal can improve performance somewhat, but it can also inflate drive’s cost per gigabyte—arguably a more important consideration for folks shopping for consumer-grade SSDs.
Our testing methods
Before dipping into our benchmark results, let’s take a quick look at the mix of rivals we’ve put together to face the Vertex 3, and the methods we use to test storage devices here at TR. We include these details to help you better understand and replicate our results, but if you’re already familiar with our approach to storage testing, feel free to skip ahead to the benchmarks. We won’t be offended.
Today, the Vertex 3 will face off against a collection of SSDs based on a handful of different controllers. Note that the Vertex 3, the RealSSD C300, and the 510 Series are the only ones with 6Gbps SATA interfaces. We’re using a Sandy Bridge motherboard with 6Gbps SATA connectivity, so those drives have a distinct advantage over the others. The Agility 2 also has somewhat of an edge thanks to a 28% overprovisioning percentage, four times what’s typical for consumer-grade SSDs. We’ve found SandForce-based SSDs tend to run slower when they set aside a more traditional 7-8% of their flash capacity as spare area.
|Flash controller||Interface speed||Cache size||Total capacity|
|Corsair Nova V128||Indilinx Barefoot ECO||3Gbps||64MB||128GB|
|Crucial RealSSD C300||Marvell 88SS9174-BJP2||6Gbps||256MB||256GB|
|Intel X25-M G2||Intel PC29AS21BA0||3Gbps||32MB||160GB|
|Intel 510 Series||Marvell 88SS9174-BKK2||6Gbps||128MB||250GB|
|OCZ Agility 2||SandForce SF-1200||3Gbps||NA||100GB|
|OCZ Vertex 3||SandForce SF-2281||6Gbps||NA||240GB|
We’ve updated all the drives to their latest and greatest firmware revisions with the exception of the Nova. This Indilinx-based drive debuted well into the controller’s life, so the initial release should have all of the kinks ironed out. Corsair tells us there are no firmware updates for the Nova.
We’re in the midst of overhauling our storage test systems here at TR, a plan that was stalled briefly by Intel’s Sandy Bridge chipset bug. The new suite of tests is coming soon, and it should be worth the wait. In the interim, we’ve whipped up an abbreviated version with a handful of new and old tests that cover the basics.
The block-rewrite penalty inherent to flash memory, the TRIM command designed to offset it, and the last workload an SSD tackled can all impact drive performance, so we’ll provide a little more detail on exactly how we test SSDs. Before testing, each drive is returned to a factory-fresh state with a secure erase. Next, we fire up HD Tune and run a series of read and write tests covering transfer rates and random access times. HD Tune is designed to run on unpartitioned drives, so TRIM won’t be a factor. The command requires a file system to be in place.
After HD Tune, we partition the drives and fire up a series of IOMeter workloads using the latest version of that app. When running on a partitioned drive, IOMeter first fills it with a single file, firmly putting SSDs into a used state in which all of their flash pages have been occupied. We delete that file before moving onto our used-state file copy tests, after which we tackle disk-intensive multitasking. Our multitasking benchmark requires an unpartitioned drive; like HD Tune, it shouldn’t be affected by TRIM.
With our multitasking tests completed, we secure-erase the drives once more and launch a final instance of our scripted file copy test. This procedure should ensure that each SSD is tested on an even playing field—and in best- and worst-case performance scenarios.
We run all our tests at least three times and report the median of the results. We’ve found that IOMeter performance can fall off after the first couple of runs, so we use five in total and throw out the first two. Each drive’s performance over the last three runs has been pretty consistent thus far. We’ve also seen remarkable consistency with our new FileBench copy test, which we’re currently running five times while we tune the scripting. We used the following system configuration for testing:
|Processor||Intel Core i7-2500K 3.3GHz|
|Motherboard||Asus P8P67 PRO|
|Platform hub||Intel P67 Express|
|Platform drivers||INF update 184.108.40.2065
|Memory size||8GB (2 DIMMs)|
|Memory type||Corsair Vengeance DDR3 SDRAM at 1333MHz|
|Audio||Realtek ALC892 with 2.58 drivers|
|Graphics||Gigabyte Radeon HD 4850 1GB with Catalyst 11.2 drivers|
Corsair Nova V128 128GB with 1.0 firmware
Intel X25-M G2 160GB with 02M3 firmware
Intel 510 Series 250GB with PWG2 firmware
OCZ Agility 2 100GB with 1.29 firmware
Crucial RealSSD C300 256GB with 0006 firmware
OCZ Vertex 3 with 1.11 firmware
|Power supply||OCZ Z-Series 550W|
|OS||Windows 7 Ultimate x64|
Thanks to Asus for providing the system’s motherboard, Gigabyte for the graphics card, Intel for the CPU, Corsair for the memory, OCZ for the PSU, and Western Digital for the Caviar Black 1TB system drive.
We used the following versions of our test applications:
The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at a 75Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.
Most of the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
HD Tune — Transfer rates
HD Tune lets us look at transfer rates in a couple of different ways. We use the benchmark’s “full test” setting, which tracks performance across the entire drive and lets us create the fancy line graphs you see below. This test was run with its default 64KB block size.
As you can see, we’ve painted our results in a rainbow of colors to make the graphs easier to interpret. In the bar graphs, drives are colored by manufacturer, with the Vertex 3 highlighted in golden yellow. The line graphs follow a similar color scheme with an additional shade of blue to differentiate between the Intel drives.
The Vertex 3 may not hit its rated peak in HD Tune’s sequential read speed test, but it’s the only drive to break 400MB/s. Intel’s latest falls 25MB/s short on average, and the C300 trails by a much bigger margin. With the Vertex 3 nearly 70% faster than the Agility 2, it’s clear that that SandForce has made substantial gains in sequential read throughput.
That’s one funky looking line graph. The Vertex 3’s read speed spikes modestly at pretty regular intervals, but its write speed climbs by over 150MB/s in a series of short spurts across the drive’s total capacity. We’ve seen this before from SandForce-based drives with 7% overprovisioning, so it’s not a terribly surprising result. This time around, the spikes are more widely spaced than with SandForce’s last-gen controller.
Most of the time, the Vertex 3 writes at a little more than 250MB/s. The peaks pull up the overall average, but not by enough to catch Intel’s new 510 Series, whose write speed is considerably more consistent overall. Barring the first few moments of the test, which seem to trouble all SandForce-powered SSDs, the Vertex 3 offers higher sequential write speeds than the rest of the pack.
HD Tune’s burst speed tests are meant to isolate a drive’s cache memory. How will they run on the cacheless Vertex 3?
Pretty quickly. The newest Vertex trumps the 510 Series with reads and ties it with writes. 6Gbps connectivity clearly helps those drives, but it’s not the only factor. Despite its 6Gbps SATA link, the RealSSD C300 is only a smidgen faster than the 3Gbps drives.
HD Tune — Random access times
In addition to letting us test transfer rates, HD Tune can measure random access times. We’ve tested with four transfer sizes and presented all the results in a couple of line graphs. We’ve also busted out the 4KB and 1MB transfers sizes into bar graphs that should be easier to read.
The line graph is a little hard to read when looking at the smaller transfer sizes, but the Vertex 3 is right in the thick of things. In fact, the Vertex 3 comes out on top in three of four transfer sizes, including one tie with the old X25-M at 4KB. With that particular transfer size, the Vertex is substantially quicker than the 510 Series. Intel’s latest SSD draws closer with larger 1MB transfers, though.
SandForce has to share the stage a little more when it comes to random writes. The Vertex 3 is every bit as fast as the Intel 510 Series in the two transfer sizes we’ve pulled out for closer inspection. The two SSDs are also closely matched with the 512-byte and 64KB transfer sizes.
TR FileBench — Real-world copy speeds
Our resident developer, Bruno “morphine” Ferreira, has been hard at work on a new file copy benchmark for our storage reviews. FileBench is the result of his efforts. This shining example of scripting awesomeness runs through a series of file copy operations using Windows 7’s xcopy command. Using xcopy produces nearly identical copy speeds to dragging and dropping files using the Windows GUI, so our results should be representative of real-world performance.
To reduce the number of external variables, FileBench runs entirely on the drive that’s being tested. Files are copied from source folders to temporary targets that aren’t deleted until all testing is complete. Copy speeds were tested with the SSDs fresh from a secure erase and in a tortured used state after more than half a day’s worth of IOMeter thrashing.
To gauge performance with different kinds of files, we tested with four sets. The movie set includes six video files of the sort one might download off BitTorrent. Total payload: 4.1GB. Our MP3 file set uses a chunk of my music archive, which is made up of high-bitrate MP3s and associated album art. This one has 549 files that add up to 3.47GB. The Mozilla file set includes the huge selection of files necessary to compile Firefox. All told, there are 22,696 files spread across only 923MB. Finally, we have the TR file set, which contains several years worth of the images, HTML files, and spreadsheets behind my reviews. This set has the largest number of files at 26,767, but it’s heftier than the Mozilla set with 1.7GB worth of data.
Like father, like son. SandForce controllers are known for slowly reclaiming trimmed flash pages, and the SF-2281 inside the Vertex 3 upholds that tradition. Interestingly, this relaxed attitude to clearing unused flash pages only affects copy speeds with the larger files contained in the movie and MP3 file sets. With those two sets, the Vertex’s used-state copy performance is 15-21% slower than it is fresh from a secure erase.
The Vertex 3’s used-state copy speeds with the smaller files in the Mozilla and TR file sets are lower, as well, but by a comparatively trivial margin. I suspect the additional overhead associated with the massive number of individual files in the Mozilla and TR file sets is responsible for leveling the playing field somewhat. All the drives copy those file sets much slower than they do the other ones.
Overall, the Vertex 3 fares pretty well. With the Mozilla and TR file sets, it’s the fastest SSD of the bunch regardless of the state of the drive. Switch to larger files, and the Vertex 3 does no worse than a second-place tie with the C300.
TR DriveBench — Disk-intensive multitasking
TR DriveBench allows us to record the individual IO requests associated with a Windows session and then play those results back on different drives. We’ve used this app to create a set of multitasking workloads that combine common desktop tasks with disk-intensive background operations like compiling code, copying files, downloading via BitTorrent, transcoding video, and scanning for viruses. You can read more about these workloads and desktop tasks on this page of our SSD value round-up.
A new version of DriveBench complete with updated traces is in the works. This old suite of workloads still has some life left in it, though.
Below, you’ll find an overall average followed by scores for each of our individual workloads. The overall score is an average of the mean performance score with each multitasking workload.
Well, that’s making a statement. The Vertex 3 tears through our multitasking workloads faster than any other SSD we’ve tested. Although the Vertex has a healthy lead over its closest competitors, the gap between it and the Agility is even more impressive. Last year’s SandForce controller is nearly 35% slower than the latest and greatest, at least when we look at our overall DriveBench average.
Let’s break down the overall average into individual test results to see if the Vertex 3 stands out anywhere in particular.
Nope, it’s consistently fast across all five workloads. Intel’s Marvell-based 510 Series SSD fares better with the file copy workload, which should come as no surprise given that drive’s strong performance with our real-world file copy tests. Otherwise, the Vertex 3 runs the table.
As a control, we also recorded a trace of our foreground tasks, while nothing was going on in the background.
Without multitasking upping the ante, the Vertex 3 settles for a first-place tie with the C300. Both have a distinct edge over the 510 Series.
DriveBench lets us start recording Windows sessions from the moment the storage driver loads during the boot process. We can use this capability to gauge boot performance, this time with TweetDeck, Pidgin, AVG, Word, Excel, Acrobat, and Photoshop loading from the Windows startup folder.
This workload is quite short, and it runs very quickly on all the SSDs involved. The Vertex 3 completes the startup trace one second faster than the others, resulting in a healthy lead.
Our IOMeter workloads are made up of randomized access patterns, making them perfect candidates to exploit the wicked-fast access times of solid-state storage. The app bombards drives with an escalating number of concurrent IO requests and should do a good job of simulating the demanding environments common in enterprise applications. We tested using the “pseudo random” data pattern, which is IOMeter’s old default and somewhat amenable to the compression mojo built into SandForce controllers. Additional testing with the “full random” data pattern revealed only a minor drop in the Agility 2’s performance, so we’re sticking with pseudo random for now.
Over the last few years, we’ve watched new storage controller drivers (including the Intel RST drivers used in this review) effectively cap IOMeter performance scaling beyond 32 outstanding I/O requests. The Serial ATA spec’s Native Command Queue is 32 slots deep, and more than one drive maker has told us that this queue is rarely full. As a result, we’re only testing up to 32 concurrent I/O requests.
Surprisingly, the Vertex 3 has lower transaction rates than the Agility 2 with three of four access patterns. Those particular workloads are made up of a mix of read and write requests, while the web server access pattern consists of reads exclusively. With that access pattern, the Vertex 3 easily trumps the Agility and hits the highest peak of any of the drives.
We’ve seen SandForce-based SSDs with higher overprovisioning percentages perform better than drives with less flash capacity dedicated to spare area, and that’s probably what’s going on here. Remember that this particular Agility 2 has an overprovisioning percentage more than four times higher than that of the Vertex 3.
Take the Agility 2 out of the picture, and the Vertex 3 looks pretty compelling. OCZ’s most recent SandForce-based SSD is locked in a back-and-forth battle with the C300 for overall dominance. The 510 Series is only close with the web server test pattern, and even then it’s no faster than OCZ’s new hotness.
Although the Vertex 3 we tested is running pre-production firmware, it’s hard not to get excited about the results. This is one of the most balanced SSD performances we’ve seen in a while, and that’s exactly the sort of thing we look for when shopping for a solid-state system drive.
SandForce controllers have always had excellent performance with random access patterns. The Vertex 3’s SF-2281 looks to be no different in that regard, and its sequential throughput is much improved over the last generation of SandForce controllers. Not only does the Vertex 3 boast quick access times and excellent performance with our disk-intensive multitasking and multi-user loads, but it’s also pretty quick when copying files and performing long reads and writes.
To be fair, Intel’s 510 Series will be a little faster with some sequential transfers. Crucial’s RealSSD C300 also offers higher transaction rates in certain scenarios. But the Vertex 3 is a better all-around performer than both of those drives.
Thanks to DuraWrite and SandForce’s propensity to reclaim trimmed flash pages slowly, the Vertex 3 might also last longer than its competitors. Projecting SSD endurance is difficult to do, especially when you’re dealing with the latest 25-nano flash chips and a write-amplification factor that changes depending on whether the data being written is easily compressible. However, the combination of DuraWrite and RAISE nicely positions the Vertex 3 to deal with any reduction in write-erase endurance or increase in error rates that might arise from its jump to the next generation of NAND fabrication technology.
OCZ tells us the Vertex 3 will be available for sale in the next few weeks. Pricing hasn’t been finalized, but early indications are that 120GB versions of the drive will sell for $250, while 240GB units like the one we looked at today will run $500. Those prices closely mirror the cost of the RealSSD C300, and they’re much lower than what you’ll pay for a like-sized 510 Series. Assuming nothing changes on the pricing front, I’d have a hard time justifying either of those other drives over the Vertex 3.
Things do change, though. Prices fall in response to new competition, firmware gets updated to improve performance, and we’re still not done with this season’s string of SSD releases. Don’t forget that we have a more comprehensive suite of storage tests in the works, as well. We’re holding back on making any official recommendations until the dust settles and we have a full collection of results from our new suite. In the meantime, OCZ’s Vertex 3 is the definite front-runner for our favorite next-gen SSD.