We’ve all done it. You find yourself at the computer late at night. It’s dark out, and maybe the lights are off. You’re alone. Inevitably, your mind wanders from the task at hand to a far more carnal desire… for the perfect enthusiast’s PC. Some build dream rigs in their heads, while others prefer to fill online shopping carts with their perfect mix of components. Occasionally, I’ll consider the sort of world-beating machine I’d assemble if money were no object. But I’m a realist at heart, and most of the time I end up contemplating a sweet spot rigthe perfect mix of components that I’d actually buy if tasked with building myself a new desktop.
At the moment, my hypothetical sweet spot sits somewhere between our system guide‘s Utility Player and Sweeter Spot configs. As much as I’m tempted by the Sweeter Spot’s Hyper-Threading-equipped Core i7-860, the Utility Player’s quad-core i5-750 is plenty of CPU for the sort of things I do with my desktop on a day-to-day basis. I’d spring for the Sweeter Spot’s Radeon HD 5850 graphics card in a heartbeat, though. This perfect desktop of mine is paired with three 24″, 1920×1200 IPS panels for multitasking real estate, and I want enough horsepower to handle recent games in a three-way Eyefinity config.
I also want my ideal desktop to be as quiet as possible, which is why I’d move upmarket from the Sweeter Spot’s single hard drive to a hybrid combo that melds an SSD with a pair of low-RPM mechanical hard drives running in a fault-tolerant RAID 1 array. This sort of setup isn’t cheap, but it offers jaw-dropping performance backed by redundant mass storage that barely makes a peep.
Solid-state drives have been on the radar for a while now, but they’ve long been regarded as far too expensive to be practical, even with phenomenal performance. Prices have fallen dramatically in recent years, though, and we’re finally at a point where it isn’t so outlandish to consider seriously purchasing an SSD for a high-performance desktop PC.
There’s little doubt in my mind that increasing competition has played a role in forcing SSD prices downward. Unlike the world of mechanical hard drives, which is made up of only a small handful of well-established players, the SSD market is teeming with fresh faces looking to claim territory. New drives are popping up faster than Justin Bieber memes, and to get the lay of the land, we’ve assembled four of them for closer inspection.
In this latest batch of SSDs, we have Corsair’s Nova V128, Kingston’s SSDNow V+, Plextor’s PX-128M1S, and Western Digital’s SiliconEdge Blue. As one might expect, all four use the 2.5″, 9.5-mm mobile hard drive form factor that has become de rigueur for solid-state drives. This drive size is a perfect fit for most laptops, although the 3.5″ drive bays found in typical desktop enclosures will require an adapter. Fortunately, Corsair provides the necessary bracket with the Nova. You’ll have to seek out an adapter with the other three or resort to duct tape, which isn’t that bad of a solution considering that SSDs can be run in any orientation and require no vibration damping. Some newer desktop cases do feature 2.5″ drive bays, but those models are still few and far between.
Of course, the casing is the least interesting element of a modern SSD. The underlying controller and flash memory largely determine drive performance, making those elements the most interesting to us. As luck would have it, each of the four drives we’re looking at today features a different controller under its thin, metal skin.
Corsair’s Nova V128 sports a new spin on an old favorite. Indilinx burst onto the SSD scene last year with its Barefoot controller that promised TRIM support long before Windows 7’s release. The Nova uses a revised version of that controller dubbed the Barefoot ECO. According to Indilinx, this latest Barefoot revision should offer near-equivalent performance to the original when used with flash memory built on a 5x-nm node. The ECO also improves compatibility with newer flash memory built using 3x-nm process technology. Corsair says the chip can offer better performance than its predecessor when used with newer flash chips.
And that’s just what has been done with the Nova. Interestingly, Corsair pairs the ECO with 34-nm flash memory chips from IM Flash TechnologiesIntel’s joint venture with Micron. A total of 16 MLC flash chips populate our 128GB Nova, with eight of them appearing on each side of the drive’s circuit board. Like Barefoot implementations of old, you’ll also find 64MB of cache memory onboard.
As one might expect, the ECO controller retains the original Barefoot’s TRIM support and garbage-collection scheme. Garbage collection is actually a part of each of the SSDs we’ve assembled today, although TRIM support is not.
Although Indilinx has quickly made a name for itself in SSD circles, you don’t hear as much about Toshiba. Kingston’s SSDNow V+ series is our first glimpse of Toshiba’s T6UG1XBG controller, which is predictably paired with Toshiba’s own MLC flash memory chips. My apologies for the blank-looking chips in the picture above. The squishy rubber pad that sits between the SSDNow’s circuit board and casing has all but stripped our chips of their silk-screened markings. All that’s left are faint black-on-black letters that proved too difficult to photograph.
Few details are available on the Toshiba controller. However, we can tell you that it offers TRIM support and uses a 3Gbps Serial ATA interface. SSDs may be the only devices with a chance of actually exploiting a 6Gbps Serial ATA link, but none of the drives in this bunch employ the next-gen standard.
Unlike the other drives, which spread memory across both sides of their circuit boards, all of the SSDNow’s memory chips can be found topside. Our 128GB drive has eight 43-nm flash chips in total, and there’s a 128MB DRAM cache thrown in for good measure.
The PX-128M1S is not only Plextor’s first SSD, but also our first look at Marvell’s “Da Vinci” 88SS8014 controller. This chip has since been succeeded by a new 88SS9174 that has a 6Gbps Serial ATA interface and appears in Crucial’s RealSSD C300. However, Plextor doesn’t plan on offering an SSD with the new controller until this summer.
Although I wouldn’t write off the old Marvell controller simply because there’s a new one, the Da Vinci design has a potentially fatal flaw: it doesn’t support the TRIM command. The block-rewrite penalty associated with flash memory has long been the scourge of SSD performance, and without support for TRIM, the Plextor drive comes into this round-up at a distinct disadvantage.
Perhaps in an attempt to make up for controller shortcomings, Plextor endows the PX-128M1S with a generous 128MB DRAM cache that can be found on the underside of the circuit board. The drive’s memory chips are provided by Samsung, and their part number doesn’t reveal the process technology used to build the chips. I suspect they were fabbed on a 40-nm node. The 16 MLC flash chips are divided evenly between the two sides of the board.
With the SiliconEdge Blue, Western Digital has become the first traditional hard drive maker to get into the desktop SSD market. Don’t let the logo on the controller chip fool you, though. WD didn’t build the chip, and it wouldn’t say who did. The folks over at AnandTech have learned that the controller is a JMicron JMF612, though.
Early SSDs powered by JMicron’s old JMF602 flash controller were plagued with stuttering issues, which may be why WD was reticent to reveal the controller’s roots. Still, the JMF612 appears to be a contemporary design, complete with built-in garbage collection and TRIM support. Western Digital says it spent a great deal of time tuning the SiliconEdge’s firmware, as well. Those firmware enhancements are unique to the SiliconEdge Blue and won’t be made available to other SSD makers who use the JMicron chip.
Western Digital pairs the JMF612 with a 64MB DRAM cache and MLC flash memory from Samsung. Our drive, a 256GB model, came with a whopping 32 flash chips split evenly between the top and bottom of the circuit board. The chips themselves are double-stacked, allowing WD to increase the drive’s capacity in the same form factor without resorting to higher-density chips. Samsung’s data sheets don’t say, but I suspect these chips are 40-nm units, just like the ones on the Plextor drive.
Lining ’em up
With four drives, four controllers, and three different flash providers, there’s a lot to keep track of in this latest wave of SSDs. To help make some sense of the collection assembled today, we’ve put together a quick comparison table that covers the basics.
|Corsair Nova V128||Kingston SSDNow V+||Plextor PX-128M1S||WD SiliconEdge Blue|
|Controller||Indilinx Barefoot ECO||Toshiba T6UG1XBG||Marvell Da Vinci||JMicron JMF612|
|Warranty length||Two years||Three years||Three years||Three years|
As you can see, the drives are split between 64 and 128MB of cache. All but one packs 128GB of storage capacity, with the SiliconEdge Blue doubling that total. Each of these drives is, of course, available at different capacity points. Corsair and Plextor offer their drives in 64 and 128GB capacities. Western Digital has 64, 128, and 256GB flavors of the SiliconEdge, while Kingston takes its SSDNow range all the way up to 512GB.
Higher-capacity SSDs tend to offer better performance than lower-capacity models because they allow the drive controller to spread the load across additional chips. However, WD quotes identical performance figures for its 64, 128, and 256GB SiliconEdge Blue drives, and Kingston does the same for the different sizes of the V+. To take advantage of extra flash chips, a drive controller needs enough memory channels to use them effectively. The JMF612 may simply not have enough memory channels to exploit more than the eight chips necessary to make up a 64GB drive. By the same token, the Toshiba controller inside the SSDNow may not benefit from having more than the eight flash chips used on that 128GB model.
Rather than using capacity or chip counts to predict performance, we can get a sense of things by looking at the only common performance specifications published for all four drives: the maximum read- and write-speed ratings. Corsair has the lead on paper, with the Nova boasting faster ratings than the rest of the pack. The SSDNow and SiliconEdge sit in a sort of tie for second place; the Kingston drive claims faster writes, and the WD promises quicker reads. Then there’s the Plextor, which looks set to read and write at least 100MB/s slower than the others.
We will, of course, take a closer look at transfer rates in a momentboth in directed tests and with real-world file copies. Before digging into our performance results, it’s worth taking a moment to discuss less technical factors like pricing and warranty coverage.
Despite its anemic performance ratings and lack of TRIM support, the Plextor drive is listed at $399 at Newegg. It does have one more year of warranty coverage than the Nova, but Plextor’s three-year coverage matches what’s available from WD and Kingston. The PX-128M1S is going to have to pull off a few surprises to justify its price tag, because the free copy of Acronis True Image that’s included in the box isn’t going to be enough.
Obviously, the 256GB SiliconEdge Blue is the most expensive drive of the bunch. 128GB flavors sell for $400, so you’re looking at the same cost per gigabyte one notch down on the capacity ladder. That said, even a 128GB SiliconEdge Blue will be more expensive than the Corsair Nova, which itself runs a good $50 more than the SSDNow V+. We’ll look at each drive’s actual cost-per-capacity a little later in the round-up.
TRIM at last
For all their blazing-fast performance, flash-based SSDs live with a crippling handicap: the block-rewrite penalty. Impossible to avoid, this penalty stems from the nature of flash memory itself. Flash cells are typically arranged in 512KB blocks made up of 4KB pages. Empty cells can be written to directly in 4KB chunks. However, if any pages within a cell are occupied, the entire block must be rewritten. A block rewrite is required even if just a single page’s worth of data is being written to the block.
As you might’ve guessed, this block rewrite is the source of our performance penalty. An SSD must perform additional steps when writing to an occupied cell. The entire contents of that cell must be read into memory and modified, and then the entire block must be rewritten. Since occupied cells can only be written to in 512KB blocks, even a small 4KB write will incur a 512KB block rewrite. The performance lost due to the larger write and the additional read and modify steps is referred to collectively as the block-rewrite penalty.
The detrimental performance impact of the block-rewrite penalty is made worse by the way operating systems have traditionally dealt with deleted data. When a file is deleted in Windows XP or Vista, for example, the flash pages it occupied are marked as available but are not actually cleared. The pages may be available in the sense that they no longer contain data that’s considered valid. However, the data is still there, and thus the page is still occupied.
As more data is written to a solid-state drive over time, the SSD will eventually exhaust its supply of empty pages. After it reaches this used state, each and every subsequent write request will be slowed by the block-rewrite penalty, regardless of how much free space the drive reports to the operating system.
The TRIM command battles the block-rewrite penalty by reducing the number of unnecessarily occupied flash pages on a drive. When a file is deleted, the TRIM command lets a compatible SSD know that the associated pages are not just available, but safe to be erased. The drive is then free to clear those pages at its leisure.
Why not force pages to be erased immediately? Because flash memory, and in particular the MLC type found in consumer solid-state drives, can only endure a limited number of write-erase cycles. Forcing an SSD to clear the pages immediately might reduce the lifespan of the drive. Instead, TRIM leaves an SSD free to clear pages opportunistically, perhaps in conjunction with its own internal garbage-collection and wear-leveling mechanisms.
The TRIM command is currently supported by Windows 7, Windows Server 2008 R2, and the 2.6.33 Linux kernel released in February of this year. Microsoft hasn’t detailed plans to port TRIM back to older versions of Windows, nor has Apple announced TRIM support for any version of OS X. You’ll want to make sure that your motherboard’s storage controller drivers support the command, too. Some may not, and others do only with restrictions. For example, Intel’s latest Rapid Storage Technology drivers support TRIM for SSDs as long as they’re running as single drives, but not if they’re members of a RAID array.
Our testing methods
If you’re unfamiliar with The Twins, our new duo of storage test platforms, I recommend checking out this page from our recent VelociRaptor VR200M review. These systems pack potent hardware and have been furiously testing hard drives and SSDs for weeks now. Unfortunately, Intel still hasn’t resolved the performance scaling issue we found in its latest storage controller drivers for the P55 chipset. As a result, The Twins are still running the Microsoft AHCI driver built into Windows 7.
Before dipping into pages of benchmark graphs, let’s set the stage with a quick look at the players we’ve assembled for comparison today. Below is a chart highlighting some of the key attributes that can affect drive performance.
|Interface speed||Spindle speed||Cache size||Platter capacity||Total capacity|
|Caviar Black 2TB||3Gbps||7,200 RPM||64MB||500GB||2TB|
|VelociRaptor VR150M||3Gbps||10,000 RPM||16MB||150GB||300GB|
Our quartet of new solid-state drives will face off against Intel’s second-generation X25-M. The G2 supports TRIM and has a similar cost per gigabyte to the other SSDs, even if it does cost more than the 128GB drives due to a more generous 160GB capacity.
I know what you’re thinking: but what about newer SSDs like Crucial’s RealSSD C300 and all that new SandForce-based hotness that was on display at CES? The only SandForce-based model you can actually buy at the moment is OCZ’s limited-edition Vertex LE, which the company tells me is nearly sold out already. We’ll be taking a closer look at SandForce-based drives when mass-market models arrive. We’ll also be testing the RealSSD C300 but are currently waiting on Crucial to finish a firmware update that’s supposed to address some performance issues with the drive.
We have, however, included performance data from a trio of mechanical hard drives. Western Digital’s 10k-RPM VelociRaptor VR200M is the fastest mechanical hard drive that plugs into a Serial ATA interface, so it’s the most appropriate foil for these SSDs. We’ve also included the VR200M’s predecessor, the VR150M, which still has quicker access times than most desktop drives. Finally, we’ve thrown in a two-terabyte Caviar Black to represent the best performance 7,200-RPM mechanical drives have to offer.
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 SSD testing methods in greater detail. Before testing the drives, each was returned to a factory-fresh state with a secure erase, which empties all the flash pages on a drive. Next, we fired up HD Tune and ran full-disk read and write speed tests. The TRIM command requires that drives have a file system in place, but since HD Tune requires an unpartitioned drive, TRIM won’t be a factor in those tests.
After HD Tune, we partitioned the drives and kicked 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 deleted that file before moving onto our file copy tests, after which we restored an image to each drive for some application testing. Incidentally, creating and deleting IOMeter’s full-disk file and the associated partition didn’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. Testing drives in a used state may put the TRIM-less Plextor SSD at a disadvantage, but I’m not inclined to indulge the drive just because it’s using a dated controller chip.
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:
Intel Core i5-750 2.66GHz
|Chipset||Intel P55 Express|
OCZ Platinum DDR3-1333 at 1333MHz
Realtek ALC889A with 2.42
Gigabyte Radeon HD 4850 1GB with Catalyst 10.2 drivers
Western Digital VelociRaptor VR200M 600GB
Western Digital Caviar Black 2TB
Western Digital VelociRaptor VR150M 300GB
Corsair Nova V128 128GB with 1.0 firmware
Intel X25-M G2 160GB with 02HD firmware
Kingston SSDNow V+ 128GB with AGYA0201 firmware
Plextor PX-128M1S 128GB with 1.0 firmware
Western Digital SiliconEdge Blue 256GB with 5.12 firmware
OCZ Z-Series 550W
Windows 7 Ultimate x64
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.
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.
We’ll kick things off with HD Tune, which replaces HD Tach as our synthetic benchmark of choice. Although not necessarily representative of real-world workloads, HD Tune’s targeted tests give us a glimpse of each drive’s raw capabilities. From there, we can explore which drives live up to their potential.
Of the new SSDs, all but the Plextor are within reach of 200MB/s in HD Tune’s sustained read speed test. The SiliconEdge Blue’s read speed regularly dips below that of the SSDNow, but it just manages a higher overall average.
The Nova V128 leads our collection of new contenders and just about catches the X25-M in this test. Unfortunately, the Plextor just isn’t in the same league. The PX-128M1S’s average read speed is a whopping 60MB/s slower than the closest contender, putting the drive closer to the latest VelociRaptor than to the other SSDs.
Writes prove even more problematic for the Plextor drive, which only averages 52MB/s. Not only is that slower than all of the mechanical drives, but the PX-128M1S writes at roughly half the speed of the X25-M, which already has substantially slower write speeds than the other SSDs.
Despite the fact that it has the highest write-speed rating of the bunch, the Nova V128 must settle for third place behind the SiliconEdge Blue and SSDNow V+. Those drives maintain more consistent write performance across their entire capacity, while the Nova’s write speeds oscillate between extreme highs and lows in a sawtooth-like pattern that’s nicely illustrated in the line graph.
Next up: some burst-rate tests that should test the cache speed of each drive.
The Nova takes this one easily, but I have some reservations about the test. We’ve seen the X25-M G2 produce much faster burst speeds in HD Tach on another test platform. Installing Intel’s latest AHCI drivers didn’t alter the X25-M’s performance here, so the Microsoft drivers aren’t at fault. I suspect something about the test itself is creating problems for the Intel, Plextor, and Western Digital drives, so I wouldn’t put too much stock into their poor performances here.
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.
None of the new SSDs can match the X25-M’s random read access times. The Intel drive is quicker with every transfer size, including the 1MB test that propels all the SSDs’ access times nearly into mechanical territory.
The rest of the solid-state drives offer similar performance, with the 1MB transfer size separating the field the most. At that transfer size, the SiliconEdge Blue edges out the Nova for second place.
Intel maintains its lead when we switch to random writes, but only through the first three transfer sizes. The X25-M is slower at the 1MB transfer size than the Corsair and WD offerings.
The 1MB transfer size really spreads the SSDs out, and the Plextor drive turns in a much slower access time than the rest. I’m more interested in the results with the 4KB transfer size, though. Bucking convention, the WD and Kingston drives are both quicker at writing random 4KB chunks than they are with smaller 512-byte writes.
File Copy Test
Since we’ve tested theoretical transfer rates, it’s only fitting that we follow up with a look at how each drive handles a more realistic set of sequential transfers. File Copy Test is a pseudo-real-world benchmark that times how long it takes to create, read, and copy files in various test patterns. We’ve converted those completion times to MB/s to make the results easier to interpret.
Windows 7’s intelligent caching schemes make obtaining consistent and repeatable performance results rather difficult with FC-Test. To get reliable results, we had to drop back to an older 0.3 revision of the application and create our own custom test patterns. During our initial testing, we noticed that larger test patterns tended to generate more consistent file creation, read, and copy times. That makes sense, because with 4GB of system memory, our test rig has plenty of free RAM available to be filled by Windows 7’s caching and pre-fetching mojo.
For our tests, we created custom MP3, video, and program files test patterns weighing in at roughly 10GB each. The MP3 test pattern was created from a chunk of my own archive of ultra-high-quality MP3s, while the video test pattern was built from a mix of video files ranging from 360MB to 1.4GB in size. The program files test pattern was derived from, you guessed it, the contents of our test system’s Program Files directory.
Even with these changes, we noticed obviously erroneous results pop up every so often. Additional test runs were performed to replace those scores.
The Nova easily has the fastest creation speeds with our MP3 file set, but with the other two, it shares the lead with the SiliconEdge blue. Although not fast enough to catch the leaders, the SSDNow V+ also fares much better with our second two file sets.
As for the Plextor drive, well, the results speak for themselves. We didn’t expect the PX-128M1S to be competitive without TRIM support, and as far as file creation is concerned, it’s not. You’re actually better off here with a mechanical hard drive than the Plextor SSD.
At least the Plextor drive manages to beat the mechanicals when reading the same file sets. Heck, it even beats the Corsair and Intel drives in a couple of ’em. I should note, however, that neither the Nova nor the X25-M displayed much consistency with the MP3 file set. The Nova was consistently sluggish with the program file set, though.
Whatever creates problems for the Corsair and Intel drives doesn’t seem to be affecting the Kingston or WD ones. The SiliconEdge is the faster of the two through all three file sets, but neither bests the Corsair and Intel drives when those emerge from the back of the pack.
The results are mixed when we combine reads and writes with a set of copy tests. Here, it’s the SSDNow V+ that comes out looking the strongest overall. The Nova and SiliconEdge fare well, too, with the former trumping the X25-M with each file set.
Once more, the PX-128M1S finds itself well back of the contenders. The Plextor SSD simply can’t keep up when writes are part of the equation.
We’ve long used WorldBench to test performance across a broad range of common desktop applications. The problem is that few of those tests are bound by storage subsystem performancea faster hard drive isn’t going to improve your web browsing or 3ds Max rendering speeds. A few of WorldBench’s component tests have shown favor to faster hard drives in the past, though, so we’ve included them here.
The SiliconEdge Blue takes our first application test, which it completes nearly 10 seconds faster than the SSDNow. The Nova isn’t far behind the Kingston drive, but the Plextor and Intel SSDs are much slower.
Switching to Intel’s AHCI drivers does improve the X25-M’s performance dramatically in this test. However, I haven’t had a chance to test the other SSDs with Intel’s drivers.
WorldBench’s Nero test gives the SSDs plenty of room to stretch their legs. This time, it’s the SSDNow in first place, followed closely by the SiliconEdge and Nova. The Plextor drive simply isn’t competitive. Heck, it’s even slower than our 7,200-RPM Caviar.
All of the drives seem quick enough for WorldBench’s WinZip test. The field is only separated by a measly two seconds here.
Although source-code compiling isn’t a part of the WorldBench suite, we’ve often been asked to add a compile test to our storage reviews. And so we have. For this test, we built a dump of the Firefox source code from March 23, 2010 using Visual Studio 2008. This process writes over 22,000 files totaling about 840MB, so there’s plenty of disk activity. However, we had to restrict compiling to a single thread because using multiple threads in Windows 7 proved to be unstable.
I suspect the results would be different if we could actually compile the Firefox source on multiple threads, but that wasn’t an option. If you have any suggestions for a Windows 7 compiling test that won’t be bound by our CPU and preferably uses open-source code available to the general public, please shoot me an email.
Boot and load times
Our trusty stopwatch makes a return for some hand-timed boot and load tests. When looking at the boot time results, keep in mind that our system must initialize multiple storage controllers, each of which looks for connected devices, before Windows starts to load. You’ll want to focus on the differences between boot times rather than the absolute values.
This boot test starts the moment the power button is hit and stops when the mouse cursor turns into a pointer on the Windows 7 desktop. For what it’s worth, I experimented with some boot tests that included launching multiple applications from the startup folder, but those apps wouldn’t load reliably in the same order, making precise timing difficult. We’ll take a look at this scenario from a slightly different angle in a moment.
We only have about a two-second spread between the fastest and slowest SSDs in this test. The SSDNow is easily the slowest, with the other drives all within about a second of each other.
A faster hard drive is not going to improve frame rates in your favorite game (not if you’re running a reasonable amount of memory, anyway), but can it get you into the game quicker?
The Corsair, Kingston, and WD drives load up Modern Warfare 2‘s “O Cristo Redentor” special-ops mission much faster than the mechanical drivesand the X25-M. That said, this is another case where the Intel drive appears to be held back by Microsoft’s drivers. The X25-M’s load times with the Microsoft AHCI drivers alternate between roughly 13 and 30 seconds, but we saw consistent 12-second load times with Intel’s own drivers.
Intel’s AHCI drivers also propelled the X25-M’s Crysis Warhead load time into the 36-second range. That would put the drive up with the rest of the SSDs, which are again led by the SSDNow. Once more, though, Plextor finds itself a step behind the other contenders.
TR DriveBench is a new addition to our test suite that 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 new set of multitasking workloads that should be representative of the sort of disk-intensive scenarios folks face on a regular basis.
Each workload is made up of two components: a disk-intensive background task and a series of foreground tasks. The background task is different for each workload, but we performed the same foreground tasks each time.
In the foreground, we started by loading up multiple pages in Firefox. Next, we opened, saved, and closed small and large documents in Word, spreadsheets in Excel, PDFs in Acrobat, and images in Photoshop. We then fired up Modern Warfare 2 and loaded two special-ops missions, playing each one for three minutes. TweetDeck, the Pidgin instant-messaging app, and AVG Anti-Virus were running throughout.
For background tasks, we used our Firefox compiling test; a file copy made up of a mix of movies, MP3s, and program files; a BitTorrent download pulling seven Linux ISOs from 800 connections at a combined 1.2MB/s; a video transcode converting a high-def 720p over-the-air recording from my home-theater PC to WMV format; and a full-disk AVG virus scan.
DriveBench produces a trace file for each workload that includes all IOs that made up the session. We can then measure performance by using DriveBench to play back each trace file. During playback, any idle time recorded in the original session is ignoredIOs are fed to the disk as fast as it can process them. This approach doesn’t give us a perfect indicator of real-world behavior, but it does illustrate how each drive might perform if it were attached to an infinitely fast system. We know the number of IOs in each workload, and armed with a completion time for each trace playback, we can score drives in IOs per second.
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 in each multitasking workload.
DriveBench doesn’t produce reliable results with Microsoft’s ACHI driver, forcing us to obtain the following performance results with Intel’s 184.108.40.2064 RST drivers. The app will only run on unpartitioned drives, so we tested drives after they’d completed the rest of the suite.
Looking at our DriveBench results overall, the X25-M has a clear lead over the other SSDs. The rest of the contenders are led by the Nova, which just edges out the SSDNow. WD’s SiliconEdge Blue sits in fourth place, but it’s not too far behind the Kingston and Corsair drives. The Plextor SSD, on the other hand, is much slower.
The Plextor drive has particular problems when the background task is a file copy or a BitTorrent download. Even with the remaining workloads, though, the Plextor is still a ways behind the other SSDs.
The X25-M takes the top spot with each workload, while second place gets shifted between the Corsair, Kingston, and WD drives. The SSDNow fares best with the file-copy and virus-scanning workloads. The Nova does well with BitTorrent and video transcoding, and the SiliconEdge edges out the others when compiling.
Curious to see whether removing the multitasking element of these tests would have any bearing on the standings, I recorded a control trace without a background task.
The X25-M still comes out ahead with this easier workload. However, the SiliconEdge vaults from fourth to second place. Interestingly, the Kingston and Corsair SSDs complete this trace in exactly the same amount of time.
DriveBench also lets us start recording Windows sessions from the moment the storage driver loads during the boot process. We can use this capability to take another look at boot times, again assuming our infinitely fast system. For this boot test, I configured Windows to launch TweetDeck, Pidgin, AVG, Word, Excel, Acrobat, and Photoshop on startup.
Intel takes another one. More interesting is the fact that three of the other SSDs are locked in a virtual tie for second. The Plextor drive is still way behind, of course.
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.
The SSDNow and SiliconEdge appear to share a common weakness: random write performance. Our file server, database, and workstation access patterns are all made up of a mix of read and write operations. With the exception of an early peak in the file server workload, the SSDNow V+ and SiliconEdge deliver dismal transaction rates that are even lower than what’s offered by some of the mechanical drives. The Kingston and WD SSDs fare much better with the web server access pattern, which consists exclusively of reads.
What’s perhaps most embarrassing for the SSDNow V+ and SiliconEdge is that they’re both beaten by the Plextor drive in all but the web server access pattern. Even then, the PX-128M1S still yields higher transaction rates than the SSDNow.
On a more positive note, the Nova puts down a bit of a beating on the X25-M, at least with the workstation and database access patterns. The two are more competitive with the file server access pattern, but the Nova can’t match with the X25-M’s jaw-dropping transaction rates in the web server test.
The Intel and Corsair drives have higher CPU utilization than the others, but then they’re also completing more transactions. Let’s consider the number of transactions completed by each drive per percent CPU utilization.
These efficiency scores show the X25-M in a slightly better light than the other SSDs, but only with the file server and web server access patterns. The Nova also completes more transactions per CPU cycle with the web server access pattern.
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 the PCB facing up.
Our noise level and power consumption tests were conducted with the drives connected to the motherboard’s P55 storage controller.
SSDs are completely silent, so the noise levels you see above reflect what’s generated by the rest of the system. These results are included to illustrate the difference in noise levels between an SSD and high-performance mechanical drives, which can be quite noisy, especially when seeking.
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. Drives were tested while idling and under an IOMeter load consisting of 256 outstanding I/O requests using the workstation access pattern.
SSDs draw very little power when compared with traditional hard drives, and the SSDNow and Nova are the most efficient of the bunch. I’m not sure what’s up with the SiliconEdge Blue, though. We measure idle power consumption a couple of minutes after IOMeter hammers the drives for our load test. Surprisingly, the SiliconEdge drew less when being blitzed by IOMeter than it did idling after. These results were consistent through multiple test runs, leading me to suspect that the SiliconEdge had garbage-collection or wear-leveling routines running during this idle time.
Capacity per dollar
After spending pages rifling through a stack of performance graphs, it might seem odd to have just a single one set aside for capacity. After all, the amount of data that can be stored on a hard drive is no less important than how fast that data can be accessed. Yet one graph is really all we need to express how these drives stack up in terms of their capacity, and more specifically, how many bytes each of your hard-earned dollars might actually buy.
We took drive prices from Newegg to establish an even playing field for all the contenders. Mail-in rebates weren’t included in our calculations. Rather than relying on manufacturer-claimed capacities, we gauged each drive’s capacity by creating an actual Windows 7 partition and recording the total number of bytes reported by the OS. Having little interest in the GB/GiB debate, I simply took that byte total, divided by a Giga (109), and then by the price. The result is capacity per dollar that, at least literally, is reflected in gigabytes.
SSDs look pretty dismal if you consider their cost per gigabyte. Versus a 7,200-RPM mechanical drive, you’re looking at a difference greater than an order of magnitude. Even the latest VelociRaptor yields at least five times more storage capacity per dollar.
If we just consider the solid-state drives, the cost-per-gigabyte standings are much closer. The Kingston and Corsair drives lead the field thanks to lower prices than the other 128GB models (remember that the 128GB SiliconEdge Blue costs $400, just like the Plextor). The X25-M may be the most expensive drive by far, but it also packs 32GB more than the others, allowing Intel to maintain a competitive cost per gigabyte.
Solid-state drives are still very expensive, particularly when you consider their relatively low capacities. However, you can pick up a 128GB drive that has plenty of capacity for an operating system, multiple applications, and several games for $400 or less. That’s pricey, sure, but when you consider the silent operation, low power consumption, and staggering performance potential offered by modern SSDs, pulling the trigger on one gets mighty tempting.
But which one? If you’re on a budget, probably not the Intel X25-M. The extra capacity is nice, but laying down nearly five bills on the 160GB model might be difficult to justify. The 80GB drive is cheaper but slower, so not as appealing.
I also couldn’t justify dropping $400 on the PX-128M1S. Not only does the drive lack TRIM support, but its read- and write-speed ratings are also at least 100MB/s slower than the others. The Plextor lived up to those meager ratings, posting slower performance than the other solid-state drives nearly across the board. It’s encouraging to see Plextor getting into the SSD market, and I like the fact that the company bundles disk-cloning software with the drive. However, there’s no way the PX-128M1S should be nearly this expensive considering its performance and feature set.
SiliconEdge Blue drives have the same cost per gigabyte as the Plextor, and although our 256GB model offered much better overall performance, it stumbled in IOMeter. Hard. Random-write performance is a definite weakness for the SiliconEdge Blue, and the drive doesn’t really balance this shortcoming in other areas, making it hard to recommend. I do appreciate the extra validation work and firmware tweaking that Western Digital has done with the drive, and the Blue may prove to be more attractive than other SSDs based on the same controller as a result. Western Digital will have to roll out a high-end Black version of the SiliconEdge with a different controller if it wants to attract performance-hungry enthusiasts, though.
I suppose the SiliconEdge Blue would be more attractive if it were cheaper, like the SSDNow V+. The Kingston drive is the least expensive of the bunch at just $319 online, making it particularly appealing for budget-conscious folks. What’s more, the SSDNow is actually pretty quick, at least as long as we stay away from random writes. Like the SiliconEdge, the SSDNow tanked in our IOMeter tests, posting lower transaction rates than mechanical hard drives. That might be the drive’s only real flaw, but it’s a big one, and one that’s not shared by our pick of the litter.
Corsair’s Nova V128 wasn’t the fastest drive in the majority of our tests, but its all-around performance is at least on par with that of the X25-M G2. Unlike the Kingston and WD SSDs, the Corsair seems to be free of crippling weaknesses. Sure, the Nova got tripped up in a couple of our tests, but then so did all of the other drives.
With a revised version of the popular Indilinx Barefoot controller, the Nova may be the most mature offering of these four new SSDs. I do wish Corsair would bump the drive’s warranty from two to three years, though, if only to match what’s being offered by the competition. The Nova might be cheaper than some SSDs, but the drive rings in at $369 at Newegg (and as little as $329 in our price search engine at press time), so it’s still an expensive piece of hardware. All things considered, though, the Nova V128 is good enough to earn our coveted Editor’s Choice award.