APART FROM COOLING FANS, hard drives are the only mechanical devices inside a modern PCthe last bastion of an era lost in a sea of silicon teeming with billions of transistors running at gigahertz clock speeds. The mechanical internals of modern hard drives are hardly low-tech, of course, but they do carry a number of inherent penalties we’d rather avoid.
For example, hard drives feature platters that take a fair bit of energy to keep spinning at thousands of rotations per minute. Power consumption is of course a greater concern for mobile applications, but it’s become a key metric for desktops, as well. Those spinning platters also introduce a measure of rotational latency that severely impedes seek performance. The drive head can’t just go and grab data from a given target; it has to wait for that target to come ’round on the platter. And then there’s the not-so-trivial matter of fragility. Hard drives have become considerably more robust over the years, but moving parts are still prone to damage from jostling and other physical abuse.
Solid-state drives (SSDs) aim to solve the problems associated with hard drive mechanics by replacing them with memory chips. On paper, it looks like a great idea. The first batch of SSDs did offer low power consumption and quick seek times without moving parts, however, actual throughput was dismal and capacities were quite limited.
Now a new wave of SSDs is upon us, led into our labs by Super Talent’s SATA25. This 2.5″ drive packs a stunning 128GB of total capacity and claims sustained read and write speeds of 60MB/s and 40MB/s, respectivelyhuge improvements over previous solid-state drives. But how does it hold up in the wild? We’ve run the drive through our comprehensive suite of performance, power consumption, and noise level tests to find out.
The big squeeze
Limited storage capacity has always been a problem for solid-state drives, so the SATA25’s 128GB capacity makes it rather special. Traditional hard drives still have a significant storage advantagethe latest 2.5″ mobile drives are available up to 320GBbut 128GB probably more than enough for most folks. In fact, I’m barely using 128GB on my primary desktop, and most of that space is consumed by junk I could easily live without.
Solid-state drives typically top out around 32GB, so the SATA25’s 128GB capacity is particularly impressive when compared with its peers. Super Talent achieves this capacity with 32 Samsung SLC NAND Flash chips, each of which tips the scales at 32Gbits, or 4GB.
Cramming 32 memory chips into a 2.5″ hard drive form factor is no easy task. Super Talent manages to get the job by spreading chips over a pair of PCBs that are stacked on top of each other. This makes the SATA25 a little thicker than typical mobile drives, though.
Most mobile drives measure 9.5mm thick, but the 2.5″ drive standard also allows for thicknesses of 12.5mm, 17mm, and 19mm. Fortunately, the SATA25 hasn’t completely let itself go; at 12.5mm thick, it’s at best packing a spare tire. Of course, that spare tire is one that the drive can’t suck in to slide into notebooks designed exclusively with 9.5mm drives in mind.
Super Talent is looking into stacking chips on a single PCB to reduce the SATA25’s thickness. However, the company is unsure whether it will be able to slim things down enough to reach the next notch in the belt.
If you don’t fancy squeezing the SATA25 into a notebook, the drive sports a standard Serial ATA interface and will work in regular desktop systems. It also has an “ultra rugged” metal casing that nicely complements the inherent shock tolerance and durability of solid-state drives.
A lack of moving parts makes Flash memory chips much more durable than mechanical hard drives, but longevity can be an issue. The SLC NAND memory used in the SATA25 is typically good for 100,000 write/erase cycles, and Super Talent quotes drive life at greater than 140 years. Of course, like most mobile drives, the SATA25’s warranty runs out in three years.
To help extend drive life, the SATA25 employs wear-leveling algorithms that spread write/erase cycles evenly across the disk to ensure frequently changed memory blocks don’t fail before their neighbors. There is no limit on read cycles for SLC NAND Flash memory, so wear leveling isn’t necessary for frequently read memory blocksjust those that are changed.
Today we’ve pitted the SATA25 against 7,200-RPM mobile drives from Seagate and Hitachi, and Super Talent’s old IDE Flash solid-state drive. We’ve also thrown in the latest 3.5″ drives from Hitachi, Maxtor, Samsung, Seagate, and Western Digital. These 3.5″ drives aren’t natural competitors for the SATA25, but we’re curious to see how SSD performance compares with the fastest desktop drives on the market.
Our testing methods
All tests were run three times, and their results were averaged, using the following test system.
Pentium 4 Extreme Edition 3.4GHz
Asus P5WD2 Premium
|North bridge||Intel 955X MCH|
|South bridge||Intel ICH7R|
|Memory size||1GB (2 DIMMs)|
Micron DDR2 SDRAM
RAS to CAS delay
Radeon X700 Pro 256MB with CATALYST 5.7 drivers
Western Digital Caviar SE16 750GB SATA
Seagate Barracuda 7200.10 750GB SATA
Western Digital Raptor X 150GB SATA
Western Digital Raptor WD1500ADFD 150GB SATA
Western Digital Caviar RE2 500GB SATA
Seagate Barracuda ES 750GB SATA
Samsung SpinPoint T 400GB SATA
Maxtor DiamondMax 11 500GB SATA
Hitachi Deskstar 7K1000 1TB SATA
Seagate Momentus 7200.1 100GB SATA
Hitachi Travelstar 7K100 100GB SATA
Super Talent IDE Flash 8GB ATA
Super Talent SATA25 128GB SATA
Windows XP Professional
|Service Pack 2|
Thanks to the folks at Newegg for hooking us up with the DiamondMax 11 we used for testing.
We used the following versions of our test applications:
- WorldBench 5.0
- Intel IOMeter v2004.07.30
- Xbit Labs File Copy Test v1.0 beta 13
- TCD Labs HD Tach v3.01
- Far Cry v1.3
- DOOM 3
- Intel iPEAK Storage Performance Toolkit 3.0
The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests.
All 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.
WorldBench overall performance
WorldBench uses scripting to step through a series of tasks in common Windows applications. It then produces an overall score. WorldBench also spits out individual results for its component application tests, allowing us to compare performance in each. We’ll look at the overall score, and then we’ll show individual application results.
The SATA25 is a huge improvement over Super Talent’s previous generation solid-state drive in WorldBench, turning in an overall score 13 points higher than that of its predecessor. That score is nearly good enough to catch Hitachi’s Travelstar 7K100, although desktop drives prove faster.
Multimedia editing and encoding
Windows Media Encoder
VideoWave Movie Creator
The SATA25 scores a win in WorldBench’s Windows Media Encoder workload, but the results from the Premiere test are far more interesting. There, the SATA25 is nearly twice as fast as the older IDE Flash drive and even quicker than the Travelstar. The SATA25 also shows a marked improvement over the IDE Flash in the Movie Creator test.
WorldBench’s ACDSee workload creates huge problems for the IDE Flash, but the SATA25 is unfazed, finishing the test almost four times faster. The SATA25 isn’t quick enough to catch either 7,200-RPM mobile drive, however.
Multitasking and office applications
Mozilla and Windows Media Encoder
The tables turn briefly in WorldBench’s Office XP test, where the older IDE Flash takes top honors and the SATA25 languishes at the back of the pack. The drive redeems itself somewhat in WorldBench’s multitasking test, but it’s not much faster than the competition there.
WinZip doesn’t look good for either solid-state offering, with even the SATA25 taking nearly twice as long as the closest 7,200-RPM mobile drive to complete the test. Fortunately, the SATA25 springs back in Nero, where it’s more than four times faster the IDE Flash. In that test, the SATA25 proves faster than the Momentus and Travelstar, as well.
Boot and load times
To test system boot and game level load times, we busted out our trusty stopwatch.
Load times are very fast with the SATA25. The drive edges out the rest of the pack in our system boot time test and easily posts the fastest Doom 3 and Far Cry level load times. Not even the speediest 3.5″ Serial ATA drives on the market can keep up with the SATA25 here.
File Copy Test
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. File copying is tested twice: once with the source and target on the same partition, and once with the target on a separate partition. Scores are presented in MB/s.
To make things easier to read, we’ve busted out our FC-Test results into individual graphs for each test pattern. We’ll tackle file creation performance first.
File Copy Test’s file creation workload probes the greatest weakness of solid-state drives: write performance. Here, the SATA25 is decimated, failing to even come close to the performance of 2.5″ drives spinning at 7,200-RPM. To be fair, however, the SATA25 is between two and three times faster than the original IDE Flash drive.
Now that we’ve created the files dictated by all these test patterns, let’s see how fast the drives can read them.
Solid-state drives prefer to read, and the SATA25 does rather well when FC-Test turns its attention to read performance. The SATA25 easily outdistances the fastest 2.5″ drives, and with a couple of test patterns, it dusts all or most of the 3.5″ drives we’ve assembled, too. Perhaps even more striking is the huge performance improvement over the IDE Flash drive, which is roughly a fourth as fast across the board.
File Copy Test – con’t
Next, File Copy Test combines read and write tasks in some, er, copy tests.
Copy tests stress the SATA25’s strength and its weakness, giving us somewhat mixed results. In four of five test patterns, the SATA25 is much slower than our 7,200-RPM notebook drives but still significantly faster than the IDE Flash. However, in the ISO test pattern, which is made up of a small number of very large files, the SATA25 actually manages to pull even with the Momentus and Travelstar.
FC-Test’s second wave of copy tests involves copying files from one partition to another on the same drive.
The results from the partition copy tests closely mirror transfer rates from the standard copy tests. Plagued by its poor write performance, the SATA25 is slow here, but comparatively a little faster in the ISO test pattern.
We’ve developed a series of disk-intensive multitasking tests to highlight the impact of command queuing on hard drive performance. You can get the low-down on these iPEAK-based tests here. The mean service time of each drive is reported in milliseconds, with lower values representing better performance.
The SATA25 scores a rare overall win in one of our iPEAK workloads and easily stays ahead of our other 2.5″ competitors. What’s most interesting is how much the SATA25 improves upon the mean service time of Super Talent’s IDE Flash. Again, we see the latest generation SSD beat the performance of its predecessor by nearly a factor of four.
Our second batch of iPEAK workloads presents more mixed results. In one test pattern, the SATA25 is actually a little slower than the 7,200-RPM Travelstar. However, throughout the rest of the tests, it’s the fastest 2.5″ drive in the pack.
Again, we see the SATA25 mopping the floor with its IDE Flash ancestor. Solid-state drives have come a long way in just a year.
IOMeter – Transaction rate
IOMeter presents a good test case for seek latencies with a demanding suite of multi-user load simulation patterns.
IOMeter’s file server, database, and workstation test patterns hammer drives with a mix of read and write operations, so they’re less than ideal for solid-state drives. Still, the SATA25 manages a good showing in the file server and workstation test patterns, particularly because the drive’s throughput is consistent regardless of whether there’s a light or heavy load.
Things turn in the SATA25’s favor with the web server workload, which is exclusively made up of read ops. There, the SATA25 decimates the field, leaving even Western Digital’s 10K-RPM Raptors in its wake. The web server test pattern results are even more striking because the SATA25 and IDE Flash are relatively evenly matched across IOMeter’s other workloads, yet the SATA25 is close to three times faster when it’s only asked to read.
IOMeter – Response time
The SATA25 doesn’t look quite as hot when we look at IOMeter response times; the drive is largely bunched with the rest of the field. However, in the web server test pattern, the SATA25 is clearly much more responsive.
IOMeter – CPU utilization
CPU utilization stays low across three of four IOMeter workloads. With the web server test pattern, though, the SATA25’s near-biblical throughput does require additional CPU resources. Even the IDE Flash consumes a few extra CPU cycles here, but neither drive takes up much more than 4%.
We tested HD Tach with the benchmark’s full variable zone size setting.
Super Talent claims 60MB/s read and 40MB/s write speeds for the SATA25, and they’re not being overly optimistic. The SATA25 just eclipses those numbers in HD Tach’s sustained read and write speed tests, in both cases leaving 7,200-RPM mobile drives behind. Again, we see the SATA25 simply outclassing Super Talent’s previous generation SSD.
Without a fast DRAM cache, the SATA25 doesn’t have a chance in HD Tach’s burst speed test. At least the drive’s burst performance is more than four times quicker than that of the IDE Flash.
Random access times are where solid-state drives really shine, and this synthetic test shows just how brightly. The SATA25 seeks in just 0.4 milliseconds20 times faster than Western Digital’s fastest Raptor and even a little quicker than the IDE Flash.
CPU utilization results are within HD Tach’s +/- 2% margin of error for this test.
Noise levels were measured with an Extech 407727 Digital Sound Level meter 1″ from the side of the drives at idle and under an HD Tach seek load. Drives were run with the PCB facing up.
Without moving parts, the SATA25 is utterly silent. The noise levels you see here for it and the IDE Flash refer to the noise level of the system as a whole, effectively minus any noise created by a hard drive.
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. Through the magic of Ohm’s Law, we were able to calculate the power draw from each voltage rail and add them together for the total power draw of the drive.
In a rare moment, the IDE Flash tops the podium, drawing less juice than any other drive. The SATA25 understandably draws more power here; it’s packing 16 times the storage capacity of the IDE Flash, and that’s a lot more memory to power.
Likely as a result of its extremely high capacity, the SATA25 doesn’t make as compelling a case for lower power consumption as one might expect. At least at idle, our 7,200-RPM mobile drives barely consume more power. However, those drives more than double their wattage under a seek load while the SATA25 barely moves the needle.
Super Talent’s SATA25 is a testament to just how far solid-state drives have come in a very short while. SSDs have always enjoyed blistering random access times, but as the IDE Flash illustrates, transfer rates have been dismal. The SATA25 is a huge improvement on the transfer rate front, in some cases offering more than four times the performance of its predecessor. That’s a huge leap in performance from one generation to the next, and one that allows the SATA25 to be competitive across a much wider range of tests.
Of course, the SATA25’s particularly jaw-dropping performances are confined to a narrow range of tests. Random access time is clearly the drive’s strongest suit, and as our IOMeter results suggest, the SATA25 is an absolute monster when crunching web server workloads. The drive’s performance in IOMeter’s web server test pattern is nothing short of stunning, particularly when you consider that its 128GB storage capacity should be more than enough for most web server needs.
128GB of fast flash doesn’t come cheap, though. When the SATA25 hits Newegg and other online retailers at the end of the month, it’s expected to sell for a whopping $4600, or roughly $36 per gigabyte. For web servers, $4600 for a 128GB solid-state drive is a lot cheaper than the equivalent in DRAM, so the SATA25 may actually be a pretty good deal. Obviously, though, it’s still deep in luxury territory for the rest of us. And that’s fine, because despite massive improvements in transfer rates, solid-state drives don’t yet offer the kind of all-around performance that most of us seek.
They are getting close, though, and that’s the most encouraging thing about the SATA25. This drive may have mouth-watering potential for workloads that stress random access times, busy web servers, or environments that demand high storage capacity with rugged shock tolerance or absolute silence. What’s perhaps more important is how quickly solid-state drives are catching up with their mechanical counterparts. Solid-state drives won’t replace mechanical drives for the masses overnight, but if the SATA25 is any indication of what’s to come, SSDs may chip away the mechanical monolith quicker than anyone expected.