Samsung makes quite a lot of different products. Their name is emblazoned on the plasma television that sits in my living room, my girlfriend’s cellphone, and my sister’s camcorder. The electronics giant also builds our favorite terabyte hard drive and manufactured the 17″ LCDs that used to be hooked up to my primary workstation. Heck, Samsung even makes a $2,000 washing machine with something called 3D Advanced Vibration Reduction Technology. Ooooh.
You may not have any of those products, but chances are a bit of silicon made by Samsung sits inside one of the many electronic devices that are a part of your everyday life. The Korean juggernaut has a rather formidable semiconductor wing that makes everything from ARM-based processors to flash memory. With the latter, Samsung most recently had a hand in developing Toggle DDR NAND, a new kind of flash memory that promises more than triple the throughput of traditional NAND.
Although Toggle DDR NAND may eventually end up buried inside the next iDevice, the chips are coming to market first in something better equipped to take advantage of their performance potential: a 470 Series solid-state drive of Samsung’s own design. Fancy flash is only one part of the equation with this new SSD, which features dual cache chips and a multi-core processor with apparent ARM underpinnings. The combination of those elements offers rather intriguing potential, so we’ve run Samsung’s latest through our exhaustive suite of performance tests against a deep field of solid-state rivals and mechanical hard drives to see how it fares. In a moment, all the gory details.
Before diving into a mountain of benchmark data, we should take a closer look at the mix of components that make this drive tick. We used to find that an SSD’s flash controller played the biggest role in determining drive performance, but I’m not sure that’s still the case. As we saw with Crucial’s RealSSD C300, faster flash chips can make a big difference. The RealSSD’s flash conforms to the second-gen ONFI spec, which is one of two competing standards in the high-performance NAND world. ONFI counts Micron, Intel, and Hynix among its collaborators, while the competing Toggle DDR NAND standard is backed primarily by Samsung and Toshiba. Supporters of both specs have worked with the JEDEC standards body to get their offerings an official seal of approval.
Even though only Toggle DDR makes it explicit, both solutions effectively offer a double data rate by allowing reads and writes to be executed on the rising and falling edges of a data strobe. ONFI does this via a source synchronous mode that relies on the combination of a clock signal and a data strobe. In a bid to save power, Toggle DDR dispenses with the synchronous clock and gets by with a data strobe that’s disabled unless transfers are actually taking place. This allows Toggle DDR to maintain an asynchronous interface and backward compatibility with controllers designed for less exotic NAND.
According to Samsung’s website, the current generation of Toggle DDR NAND can enable data rates up to 133Mbpsmore than a threefold increase over so-called SDR NAND, which is said to top out at around 40Mbps. I suspect those figures are for individual NAND devices rather than flash chips as a whole. Samsung’s website identifies the chips used in the 470 Series as having an x8 organization that, with a max data rate of 133Mbps per device, would yield an effective 133MB/s per chip. That speed nicely matches the second-gen ONFI spec, which supports interface speeds up to 133MB/s. Subsequent ONFI revisions have added 166 and 200MB/s interface speeds, while Samsung is promising 400Mbps for a next-gen version of Toggle DDR NAND due out next year.
The 470 Series uses Samsung K8HDGD8U5M flash chips manufactured by Samsung on a 32-nano fabrication process. Moving to finer process technologies allows chip makers to squeeze more NAND dies out of a single wafer, but it also reduces the lifespan of those chips. Flash fabbed on a 50-nano process typically carries a write-erase endurance of 10,000 cycles, while 34-nano parts are generally rated for 5,000 cycles. According to Samsung, the 470 Series’ flash chips have a write-erase endurance of only 3,000 cycles, which is on par with the longevity of 25-nano flash chips currently rolling off the line at Micron.
Each of the 470 Series’ memory chips weighs in at 16GB, so it takes 16 of them to reach the 256GB capacity of the drive we’re looking at today. Although the Toggle DDR NAND specification has a provision for low-voltage chips that require only 1.8V, these particular modules call for the same 3.3V as standard flash memory.
Along with its flash, the 470 Series sports a couple of 128MB DRAM chips. Labeled K4T1G164QE-HCE6, these puppies have a 667Mbps data rate, and Samsung puts one on each side of the drive’s circuit board. SSDs typically tap a single DRAM chip as their cache memory, making this arrangement rather unusual. Add in the fact that Samsung’s website makes no mention at all of the 470 Series’ cache, and the intrigue builds. The dual memory chips could be treated as a single 256MB cache, but it’s also possible that the drive controller uses each chip for different things. We asked Samsung for more information, but the company declined to explain exactly how the drive’s caching system works.
Despite our requests for additional details, the 470 Series’ S3C29MAX01 controller chip also remains a bit of a mystery. The controller is made by Samsung and is claimed to have a “multi CPU” design. However, we’re still in the dark on just how many cores there are and whether their origins have something to do with the ARM initials inscribed on the chip. A dual-core design seems likely, and I very much doubt that ARM’s appearance is accidential. The letters are underlined, after all.
Given the lower write-erase endurance of Samsung’s Toggle DDR NAND, we were particularly concerned about the 470 Series’ write amplification factor, which is calculated by dividing the amount of data written to the flash by the size of the actual write request. Unfortunately, Samsung declined to reveal specifics, leaving us to speculate once more. Drive makers want to minimize the amplification factor to preserve write-erase cycles, and without funky compression schemes like those employed by SandForce, they usually target a factor of 1. If Samsung achieved that goal, the 470 Series 256GB should be able to withstand around 750TB worth of host writes. The write amplification factor is directly related to total drive capacity, so you’re looking at close to 375TB for a 128GB model, and in the neighborhood of 187TB for the 64GB variant.
Wear leveling and garbage collection algorithms are usually responsible for amplifying the amount of data written to an SSD, and the 470 Series has both. The drive also supports the TRIM command, and I suspect that the mechanism used to clear trimmed flash pages has a hand in amplifying host writes, as well.
|Sequential reads||Sequential writes||Random reads||Random writes|
|64GB||250MB/s||170MB/s||31,000 IOps||11,000 IOps|
|128GB||250MB/s||220MB/s||31,000 IOps||20,000 IOps|
|256GB||250MB/s||220MB/s||31,000 IOps||21,000 IOps|
At least we don’t have to guess about the 470 Series’ Serial ATA interface, which tops out at 3Gbps. The last-gen standard has since been succeeded by a 6Gbps SATA spec with higher throughput, but this SSD isn’t fast enough to take advantage. According to Samsung’s own performance ratings, the 470 Series tops out at 250MB/s for reads and 220MB/s for writesspeeds easily attainable within the 3Gbps spec. Those performance ratings apply to 256GB and 128GB versions of the drive, while the 64GB variant is pegged at 170MB/s for writes.
Lower-capacity SSDs tend to use fewer memory chips and thus tap fewer controller-level memory channels than their higher-capacity siblings. This leads to a reduction in performance that affects more than just sequential writes. While all three capacity points share the same 31,000 IOps rating for random reads, the 128GB model is a little behind the 256GB flagship when it comes to random writes. The 64GB drive really suffers here, offering just over half the random-write throughput of the larger drives.
Solid-state drives typically fill the 2.5″ hard drive form factor to allow their memory chips to spread out on a larger circuit board. Samsung’s approach is different with the 470 Series, and it hints at the company’s focus on selling these drives to notebook makers. Inside the 470 Series’ 2.5″ casing lies an SSD whose dimensions more closely match the 1.8″ hard drive form factor. You can probably expect to find variations of this drive, if not the exact circuit board, inside the next generation of ultraportable notebooks. Samsung probably has one of its own in development.
With its outer shell in place, the 470 Series doesn’t look nearly as interesting. Our review sample came in a matte black casing that’s considerably duller than retail samples, which have brushed aluminum enclosures with bright orange accents. It’s nice to see Samsung making an effort on the aesthetics front, even if the drives will quickly be buried out of sight inside desktop or notebook systems. A little style never hurt anyone.
With Samsung’s logo always highlighted in blue, orange seems like an odd choice of color. Green might have been an more appropriate hue for the 470 Series, whose idle power consumption is rated at a measly 0.14W. The drive’s “active” power draw is claimed to be just 0.24W, which seems impossibly low unless Samsung’s definition of active involves little more than a pulse. Toggle DDR NAND does seem to have a focus on power efficiency, though; in a moment, we’ll see just how much power the 470 Series consumes in the real world.
Before getting into those and other performance results, I should point out that the 470 Series is covered by a three-year warranty. That’s a pretty standard warranty length for internal hard drives, including SSDs, although some premium mechanical models do come with five years of coverage.
Our testing methods
Before burying you under a deluge of benchmark graphs, let’s take a quick look at the mix of rivals we’ve put together to face the 470 Series, 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. I won’t be offended.
On the solid-state front, we have direct competition based on the latest and greatest controllers currently available on the market. For comparative reference, we’ve also included a stack of 2.5″ and 3.5″ mechanical drives. And, thanks to a little cross-pollination, we have results for Seagate’s Momentus XT flash/mechanical hybrid, as well. Below is a chart outlining several key characteristics that can affect the performance of the contenders we’ve lined up.
|Flash controller||Interface speed||Spindle speed||Cache size||Platter capacity||Total capacity|
|Corsair Force F100||SandForce SF-1200||3Gbps||NA||NA||NA||100GB|
|Corsair Force F120||SandForce SF-1200||3Gbps||NA||NA||NA||120GB|
|Corsair Nova V128||Indilinx Barefoot ECO||3Gbps||NA||64MB||NA||128GB|
|Crucial RealSSD C300||Marvell 88SS9174||6Gbps||NA||256MB||NA||256GB|
|Hitachi Deskstar 7K1000.C||NA||3Gbps||7,200 RPM||32MB||500GB||1TB|
|Hitachi Travelstar 7K500||NA||3Gbps||7,200 RPM||16MB||250GB||500GB|
|Intel X25-M G2||Intel PC29AS21BA0||3Gbps||NA||32MB||NA||160GB|
|Intel X25-V||Intel PC29AS21BA0||3Gbps||NA||32MB||NA||40GB|
|Kingston SSDNow V+||Toshiba T6UG1XBG||3Gbps||NA||128MB||NA||128GB|
|OCZ Agility 2||SandForce SF-1200||3Gbps||NA||NA||NA||100GB|
|OCZ Vertex 2||SandForce SF-1200||3Gbps||NA||NA||NA||100GB|
|Plextor PX-128M1S||Marvell 88SSE8014||3Gbps||NA||128MB||NA||128GB|
|Samsung 470 Series||Samsung S3C29MAX01||3Gbps||NA||256MB||NA||256GB|
|Samsung Spinpoint F3||NA||3Gbps||7,200 RPM||32MB||500GB||1TB|
|Samsung Spinpoint MP4||NA||3Gbps||7,200 RPM||16MB||320GB||640GB|
|Seagate Barracuda 7200.12||NA||3Gbps||7,200 RPM||32MB||500GB||1TB|
|Seagate Barracuda LP||NA||3Gbps||5,900 RPM||32MB||500GB||2TB|
|Seagate Barracuda XT||NA||6Gbps||7,200 RPM||64MB||500GB||2TB|
|Seagate Momentus 7200.4||NA||3Gbps||7,200 RPM||16MB||250GB||500GB|
|Seagate Momentus 750GB||NA||3Gbps||7,200 RPM||16MB||375GB||750GB|
|Seagate Momentus XT||NA||3Gbps||7,200 RPM||32MB||250GB||500GB|
|WD Caviar Black 1TB||NA||6Gbps||7,200 RPM||64MB||500GB||1TB|
|WD Caviar Black 2TB||NA||3Gbps||7,200 RPM||64MB||500GB||2TB|
|WD Caviar Green 2TB||NA||3Gbps||5,400 RPM||32MB||500GB||2TB|
|WD Caviar Green 3TB||NA||3Gbps||5,400 RPM||64MB||750GB||3TB|
|WD Scorpio Black 320GB||NA||3Gbps||NA||16MB||160GB||320GB|
|WD Scorpio Black 500GB||NA||3Gbps||7,200 RPM||16MB||250GB||500GB|
|WD Scorpio Blue||NA||3Gbps||5,400 RPM||8MB||375GB||750GB|
|WD SiliconEdge Blue||JMicron JMF612||3Gbps||NA||64MB||NA||256GB|
|WD VelociRaptor VR150M||NA||3Gbps||10,000 RPM||16MB||150GB||300GB|
|WD VelociRaptor VR200M||NA||3Gbps||10,000 RPM||32MB||200GB||600GB|
Although it might not seem like a fair fight, we’ve also thrown in results for a striped RAID 0 array built using a pair of Intel’s X25-V SSDs. The X25-V costs less than $100 online, making multi-drive RAID arrays affordable enough to be tempting for desktop users. Our X25-V array was configured using Intel’s P55 storage controller, the default 128KB stripe size, and the company’s latest 18.104.22.1684 Rapid Storage Technology drivers.
The block-rewrite penalty inherent to SSDs and the TRIM command designed to offset it both complicate our testing somewhat, so I should explain our methods in greater detail. Before testing SSDs, each is returned to a factory-fresh state with a secure erase, which empties all the flash pages on the drive. Next, we fire up HD Tune and run full-disk read and write speed tests. The TRIM command requires that drives have a file system in place, but since HD Tune runs on an unpartitioned drive, TRIM won’t be a factor in those tests.
After HD Tune, we partition the drives and kick off our usual IOMeter scripts, which are now aligned to 4KB sectors. 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 file copy tests, after which we restore an image to each drive for some application testing. Incidentally, creating and deleting IOMeter’s full-disk file and the associated partition doesn’t affect HD Tune transfer rates or access times.
Our methods should ensure that each SSD is tested on an even, used-state playing field. However, differences in how eagerly an SSD elects to erase trimmed flash pages could affect performance in our tests and in the real world.
To make our massive collection of results a little easier to interpret, we’ve marked the 2.5″ and 3.5″ mechanical drives in different shades of grey. The SSDs are color-coded by manufacturer, at least in the bar charts, and there’s a multi-colored rainbow to cover the line graphs. Keep an eye out for the 470 Series, which is wearing bright red today.
With few exceptions, all tests were run at least three times, and we reported the median of the scores produced. We used the following system configuration for testing:
You can read more about the hardware that makes up our twin storage test systems on this page of our VelociRaptor VR200M review. Thanks to Gigabyte for providing the twins’ motherboards and graphics cards, OCZ for the memory and PSUs, Western Digital for the system drives, and Thermaltake for SpinQ heatsinks that keep the Core i5s cool.
We used the following versions of our test applications:
- WorldBench 6
- Intel IOMeter 2006.07.27
- Xbit Labs File Copy Test 0.3
- HD Tune 4.01
- Visual Studio 2008 with 03-23-2010 Firefox source
- Call of Duty: Modern Warfare 2
- Crysis Warhead
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.
Some further notes on our methods:
- Noise levels were measured with a TES-52 Digital Sound Level meter 1″ from the side of the drives at idle and under an HD Tune seek load. Drives were run with their PCBs facing up next to our open-air test bench.
- For our power consumption tests, we measured the voltage drop across a 0.1-ohm resistor placed in line with the 5V and 12V lines connected to each drive. We were able to calculate the power draw from each voltage rail and add them together for the total power draw of the drive. The drives were tested under a load consisting of 256 outstanding I/O requests using the workstation access pattern. Power consumption was also probed while idling at the Windows desktop one minute after halting our IOMeter load.
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
We’ll begin with HD Tune, which offers a series of targeted tests that will give some insight into each drive’s raw potential. From there, we’ll proceed to more real-world measures of drive performance.
I’ve removed the mechanical drives from all of our line graphs to make them a little easier read. Plus, Excel really doesn’t have enough colors. If you’d like to see how the transfer-rate profiles of our mechanical hard drives compare, check out this page of our Caviar Green 3TB review.
Samsung gets off to a good start as the 470 Series’ sequential read rate eclipses 210MB/s. It’s only a couple of MB/s behind our single-drive leaders, the RealSSD C300 and X25-M, which is pretty good company to keep.
Before we move onto writes, scroll up and check out the line graph for a second. Note how the 470 Series maintains a consistent transfer rate across the extent of its capacity, an attribute not shared by every SSD we’ve tested.
You can see dramatic oscillations in the Force F120’s sequential write rate, but again, the 470 Series maintains a pretty even keep from start to finish. Even though the 470 Series’ write speed dips every so often, it never drops below 204MB/s, producing an overall average that puts the Samsung drive just behind a trio of SandForce-based SSDs. Those 100GB SandForce drives have much lower minimum write speeds only because they get off to a slow start in this particular test. That excuse doesn’t apply to the 120GB Force F120, though.
Next up: some burst-rate tests that should test the cache speed of each drive. We’ve omitted the X25-V RAID array from the following results because it uses a slice of system memory as a drive cache.
Dual cache chips or not, the 470 Series doesn’t fare well in HD Tune’s burst rate tests. The Samsung SSD falls to the back of the pack, and its transfer rates in this test are even lower than the minimum read and write speeds we observed in the sequential throughput tests. Given those results, I wouldn’t worry too much about these onesat least not yet.
HD Tune Transfer rates
Our HD Tune tests conclude with a look at random access times, which the app separates into 512-byte, 4KB, 64KB, and 1MB transfer sizes. Let’s start with reads.
Versus the other SSDs, the 470 Series’ access times are a little sluggish up to the 64KB transfer size. At 1MB, the field spreads out, and the Samsung SSD finds itself smack in the middle.
To get a better sense of things, let’s take a closer look at access times with 4KB and 1MB transfer sizes.
We use 4KB random access times in our overall performance index, and while the 470 Series is definitely at the back of the solid-state pack, it’s still more than an order of magnitude faster than the quickest mechanical drives. The SSDs slow down appreciably when the transfer size increases to 1MB, but even then, they remain several times quicker than traditional hard drives.
Despite lagging behind with random reads, the 470 Series is among the quickest solid-state drives with random writes. In fact, when we reach the 1MB transfer size, it’s the fastest drive of the lot.
The bar graph provides some additional perspective on just how closely matched the top SSDs are, at least with the 4KB transfer size. Things spread out with the larger 1MB transfer size, and this time we see a couple of the slower SSDs mixing in with the mechanical pack.
File Copy Test
After testing sequential transfer rates with some targeted tests, it’s time to see how things pan out with actual files. File Copy Test is a pseudo-real-world benchmark that times how long it takes to copy files in various test patterns. Because it doesn’t take advantage of command queuing, FC-Test isn’t a perfect representation of real-world file transfers.
We’ve converted FC-Test’s completion times to MB/s and taken an overall average across three file sets to make the data easier to present. You can read more about the file sets used by FC-Test and the issues some SSDs seem to have with this test on this page of our Force F120 review.
Although it offers higher transfer rates than the SandForce-based SSDs, the 470 Series is a few steps behind a number of other designs. The Samsung drive is a little slower than the X25-M, which puts it in fifth place overall and more than 40MB/s shy of the pack-leading RealSSD C300.
Interestingly, the 470 Series fared best with the program file set, which is made up of a larger number of smaller files. Our video file set sits at the other end of the spectrum, with a small number of very large files, and the 470 Series didn’t handle it nearly as well. Usually, we see the reverse: drives performing better with large files than they do with smaller ones.
File copy speed
We can better simulate real-world file transfers with an actual copy test in Windows 7. This hand-timed test copies 7GB worth of documents, digital pictures, MP3s, movies, and program files from the drive to itself. We run this test on solid-state drives in a factory fresh state immediately after a secure erase and again in a tortured used state after crunching IOMeter’s workstation access pattern with 256 concurrent I/O requests for 30 minutes. Because they don’t suffer from flash memory’s block-rewrite penalty, mechanical hard drives offer the same performance regardless of whether they’re fresh from a secure erase or just finishing an IOMeter workout.
So much for TRIM. The 470 Series is wicked-fast in a factory fresh state, nearly matching the C300 for the fastest file copy speed. However, our IOMeter torture test cuts the drive’s copy speed by nearly 40%, knocking it way down the standings. The 470 Series isn’t the only SSD to see its performance drop, but the magnitude of its fall is particularly notable. Only our X25-V RAID array sees a bigger reduction in file copy speed, and as a RAID array, it doesn’t work with TRIM at all.
I don’t doubt that Samsung has implemented TRIM in the 470 Series, but the drive doesn’t seem to be in a hurry to reclaim trimmed flash pages. SandForce makes a point of taking back flash pages lazily to minimize host writes, which is why the copy speeds of the Agility, Force, and Vertex SSDs are slower in a used state. The 470 Series appears to salvage flash pages at an even more lethargic pace, and this lack of urgency has serious implications for used-state copy speeds.
Curious to see whether the 470 Series simply needed a breather after our torture test, I let the drive idle for an hour before running the copy test several more times. When that didn’t improve copy speeds, I tried another 2.5 hours of idle time, to no avail.
WorldBench is a good way to explore system performance across a broad range of desktop applications, but precious few of its tests benefit from faster storage subsystems. We’ve included results from a couple of tests that have shown preference for faster drives in the past, and we’ve also thrown in a Firefox compile for those who requested that we examine compiling performance.
The WinZip and Photoshop tests are the only ones to really spread the field, and the 470 Series fares well in both. In Photoshop, it ties for the lead with the SiliconEdge Blue. Samsung must make do with fifth place in the Nero test, but that still puts the 470 Series ahead of the X25-M and a stack of SandForce drives, not to mention all the mechanical offerings.
Boot and load times
Our boot time test starts when the power button is pressed and ends when the mouse cursor turns into a pointer on the Windows 7 desktop. Before they even begin to load Windows, our test systems must first initialize multiple storage controllers. That takes some time, which is why your own system may boot much faster than ours. But, since all the drives are penalized equally with our setup, the results are comparable, at least amongst themselves.
At less than two seconds behind the leader, the 470 Series looks reasonably good in our boot time test. However, it’s a little bit slower than the other SSDs, beating only the Kingston drive and our RAID array, which takes a little extra time to initialize during the boot process.
When loading game levels, the 470 Series is more competitive. The Samsung SSD is only about second off the pace in Modern Warfare 2 and Crysis, which makes it faster than a number of other solid-state drives, including units based on Indilinx, Intel, and SandForce silicon.
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.
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.
Not bad, Samsung. The 470 Series sneaks into third place ahead of the Nova V128, which sports Indilinx’s Barefoot controller. The Intel and Crucial drives still have commanding leads, but the 470 Series is ahead of a fair number of other contenders.
Let’s break down the overall average into individual test results to see if anything stands out.
Although the 470 Series fares well with four of five workloads, it tanks when a file copy becomes part of the multitasking equation. Given the drive’s sluggish used-state copy speed, this isn’t an entirely unexpected result. I am, however, a little surprised at just how much slower the Samsung drive is than the other SSDs and mechanical drives.
As a control, we also recorded a trace of our foreground tasks, while nothing was going on in the background.
Since the 470 Series only has problems when files are copying in the background, the results of this control test play out much like our overall average.
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.
The 470 Series doesn’t look quite as good in this test. With our simulated boot, the drive scores lower than every other SSD short of the Plextor drive.
Our IOMeter workloads are made up of randomized access patterns, presenting a good test case for both seek times and command queuing. The app’s ability to bombard drives with an escalating number of concurrent IO requests also does a nice job of simulating the sort of demanding multi-user environments that are common in enterprise applications.
SSDs are orders of magnitude faster than mechanical hard drives in this test, making graphing them together a little pointless. If you’d like the see just how much slower the mechanical drives are, flick your mouse wheel down this page of our four-way 7,200-RPM terabyte comparison.
Making sense of these results is much easier if you know that the web server access pattern is made up exclusively of read requests, while the others combine a mix of reads and writes. When tasked solely with random reads, the 470 Series excels and is beaten only by the RealSSD C300 and X25-M. Impressive.
However, when random writes are added to the access pattern, the Samsung SSD’s transaction rates are much lower. It still manages to stay ahead of the X25-M and Nova V128, but the SandForce posse and RealSSD have a comfortable edge. Incidentally, the nature of the data in our IOMeter scripts may present a sort of best-case scenario for the write-compression mojo embedded in SandForce’s SSD controllers. The C300 has no such trickery up its sleeve, though.
The 470 Series is incredibly power-efficient at idle, drawing less than a third of a watt. Under a demanding IOMeter load, power consumption jumps to 1.6W, which puts the 470 Series in the middle of the solid-state pack. Note that the SSDs aren’t wildly more power-efficient than the 2.5″ mechanical drives. If you’re considering a solid-state upgrade to improve your notebook’s battery life, don’t expect power savings of more than a watt.
I’ve consolidated the solid-state drives here because they’re all completely silent. The SSD noise level depicted below is a reflection of the noise generated by the rest of the test system, which has a passively-cooled graphics card, a very quiet PSU, and a nearly silent CPU cooler.
A solid-state drive like the 470 Series can’t silence louder system components, but it won’t add any noise to the equation, either.
The value perspective
After seven pages of performance data, it’s time to add pricing to the mix with an analysis of the 470 Series’ value proposition. We’re using Newegg pricing for most of the drives to ensure an even playing field, but since that retailer recently increased the 470’s price from $550 to $620, we’ve used the cheaper of the two prices for the Samsung SSD. Amazon is still selling the drive for $550 through TigerDirect.
Capacity per dollar is up first. Here, we divide the total number of bytes reported to Windows 7 by a Giga (109), and then by the price of each drive.
SSDs remain expensive on a cost-per-gigabyte basis. You’re shocked, I know. At least among the solid-state field, the 470 Series doesn’t fare too poorly. It’s right in the middle of the pack with drives from Crucial and Kingston.
Overall performance is up next, and this is where things get complicated. We’ve created an overall index that compares performance to a common baseline: a lowly 4,200-RPM notebook drive from many years ago. The index is based on a sampling of test results from the preceding pages, and you can read more about which tests we used and how our value scores are calculated on this page of our SSD value round-up.
Before combining our overall performance index with street prices, it’s worth taking a moment to look at the raw performance score. As you can see, the 470 Series sits near the top of the heap in a sea of SandForce-powered drives. Although it can’t catch the Nova V128 or RealSSD C300, the Samsung SSD does score higher overall than the X25-M and a number of other solid-state offerings.
SSDs may have a huge advantage in performance, but storage capacity is equally important for, you know, storage devices. We’ve divided each drive’s overall performance score by its cost per gigabyte to get a look at overall performance per dollar per gigabyte. Try saying that five times fast.
I’ve omitted the mechanical drives from the scatter plots for the sake of readability. If you’re curious to see how they compare, consult this section of our last hard drive round-up.
The bar graph might put the 470 Series in third place among SSDs, but the scatter plot tells us much more about the landscape. Follow a vertical line up from the 470 Series and you’ll see that the RealSSD C300 offers much higher performance at roughly the same cost per gigabyte. The Nova V128 is only a little bit faster than the Samsung drive overall, but it’s substantially cheaper per gigabyte, putting the Corsair SSD in a more attractive position on our value plot.
Another way to look at this data is to divide each drive’s performance by the cost of a system built around it. The aim here is to determine whether spending a little (or a lot) more makes sense when the price premium is absorbed into the cost of a complete build. For our system price calculations, we’ve used our test rig as the inspiration for a base config, to which the price of each drive will be added. Our example system includes a Core i5-750, a P55-based ASUS P755D-E motherboard, 4GB of DDR3-1333 memory, a passively-cooled Radeon HD 4850, Antec’s Sonata III enclosure, and Windows 7. Grand total: $799.94, which happens to nicely match the prevailing cost of mid-range notebooks.
Like the other 256GB drives (the RealSSD C300 and SiliconEdge Blue), the 470 Series is at a disadvantage here. Higher capacities command higher prices, even if the cost per gigabyte stays steady, and that increases the total cost of our mythical system. The Samsung SSD’s $550 street price represents a huge chunk of the total system cost when the other parts total $800.
Once again, though, it’s clear that the RealSSD C300 offers much higher performance at essentially the same cost. I’d expect a 128GB version of the 470 Series to drift to the left on this scatter plot, but that model is still more expensive than the Nova V128, and it’s no cheaper than a like-sized flavor of the RealSSD.
In the realm of solid-state drives, Samsung’s 470 Series is unique. It’s the only one we’ve seen with Toggle DDR NAND, and the first SSD to hit our labs with dual cache chips. Throw in a fresh controller with potential ARM cores under the hood, and we’re left with more than a few questions about how the whole thing comes together. Unfortunately, most of them remain unanswered.
We do, however, have a diverse collection of benchmark results that tells us quite a lot about how the 470 Series performs. Generally, it does quite well, offering quick sequential transfer rates and competitive random access times. The drive’s strong overall performance isn’t fueled by any one test in particular, but by a reasonably strong showing throughout.
Well, almost throughout. Samsung’s latest SSD has one rather glaring flaw: slow file copy performance when the drive is in a used state. The results of our Windows 7 file copy test make this weakness apparent, and one of our multitasking workloads concurs. It’s unclear whether the root of the issue is a lazy TRIM implementation that takes its sweet time reclaiming flash pages, or if something else is affecting the drive’s copy performance after it’s been roughed up by an IOMeter torture test. Either way, occupied flash pages seem to be the culprit; when it’s fresh off a secure erase, the 470 Series has the second-fastest file copy speed we’ve ever measured.
With less expensive rivals largely impervious to whatever causes the 470 Series’ precipitous drop in file copy speed, it’s difficult to recommend the Samsung SSD. TRIM is supposed to address the poor used-state performance of solid-state drives, and this particular implementation isn’t delivering on that promise. What a shame. The 470 Series is otherwise impressive, and its low idle power consumption and tiny circuit board are sure to attract notebook makers looking for a small SSD to squeeze into their ultraportables. I just won’t be putting this one in mine.