Western Digital’s 6Gbps Caviar Black 1TB hard drive

Western Digital’s 640GB Caviar hard drives have found their way into each and every one of our system guides for just about two years now. We’ve recommended one flavor or another across a wide range of systems, including our budget Econobox, the mid-range Grand Experiment, and even the modestly opulent Sweeter Spot build. Indeed, the 640GB Caviars have been one of the best examples of the proverbial sweet spot in recent memory.

As all half-dozen or so of our female readers can no doubt attest, the sweet spot can be difficult to find. You’ll know when you’ve found it, though. The original Caviar SE16 640GB debuted with all the right ingredients: two of the highest-density platters available at the time, a 7,200-RPM spindle speed, low noise levels, reasonable power consumption, an affordable price tag, and best-in-class performance across a diverse range of sequential, random, synthetic, and real-world tests.

For obvious reasons, the SE16 became an instant favorite among enthusiasts. Eventually, it was supplanted by the Caviar Black. Largely the same drive as the SE16, the Black doubled the cache to 32MB and bumped the warranty up to five years. And since the 640GB flavor was still a lower capacity than then-flagship terabytes, you could still get it cheap. That’s like taking the sweet spot, drizzling it with chocolate sauce, and then adding bacon. Even I couldn’t resist and bought two for the RAID 1 array in my personal desktop.

The uncanny ability to combine top performance with low noise levels is what really made the 640GB Caviars so special. Higher-end terabyte drives were louder at the time, likely because they needed to use three or four platters to hit the 1TB mark. 640GB only took two platters, which meant less weight for the drive motor to spin, less power consumed, and less heat generated, too.

Naturally, we’ve been eagerly anticipating the 640GB Caviar’s spiritual successor: a two-platter Caviar Black that uses Western Digital’s latest 500GB platters. This next step up the areal density ladder debuted more than a year ago inside a low-power, two-terabyte Caviar Green. Some eight months later, the very same platters spun up to a full 7,200 RPM inside the Caviar Black’s shot at the 2TB mark. And now, finally, Western Digital is putting them in a two-platter Caviar Black.

Platter capacity is so important to mechanical hard drives because it plays a large role in defining overall drive performance. What really matters is the platter’s areal density—how many bits it squeezes into a given unit area. The higher the areal density, the more data the drive head can access over a given physical distance. Platters that have the same total capacity don’t necessarily share the same areal density, though. For example, Seagate’s Barracuda XT 2TB features 500GB platters that have an areal density of 347 Gb/in², while Western Digital’s 500GB platters squeeze 400 gigabits into every square inch. Advantage, WD.

Dacking data so densely under the drive head will speed sequential transfers, but it can make seeking random sectors more difficult. To give you an idea of what kind of precision is required, consider that data points on the surface of a 3.5″ platter spinning at 7,200 RPM are moving at up to the equivalent of about 120 km/h. WD’s 500GB platters have a track width of roughly 100 nanometers—one thousand times narrower than a human hair. The drive head, which sits at the end of an actuator arm several inches in length, must be capable of darting from track to track in milliseconds, all while flying within nanometers of the surface of the platter.

To help it track quickly in such challenging conditions, the 2TB Caviar Black employs a secondary actuator stage powered by a tiny piezoelectric motor perched at the tip of the drive arm. This second stage gives the drive head 500 nanometers of fine-tuning precision beyond what’s offered by the main actuator. However, this dual-stage approach hasn’t trickled down to the 1TB Black, which uses a single-stage design. Western Digital said the 2TB Black would have been slower without the dual-stage actuator, so it seems likely that some performance has been left on the table by not including it in the 1TB model.

Of course, Western Digital’s own datasheets already concede that the new 1TB Black is going to be a little slower than its four-platter, 2TB cousin. The 1TB drive has a maximum sustained data rate of 126MB/s, while the 2TB model can sustain speeds of up to 138MB/s. That discrepancy might seem counter-intuitive given that the drives use the very same platters. However, the fastest area of the disk is the outer edge of the platter; the 2TB drive has twice the platters of the 1TB model and thus double the outer-edge area.

Although the new terabyte Black may not have the platter count or dual-stage actuator to challenge for the performance crown, it does have an ace up its sleeve with support for a 6Gbps Serial ATA interface. Well, maybe it’s more of a joker. The fact is that the drive’s 126MB/s maximum sustained transfer rate isn’t even fast enough to saturate a gen-one 150MB/s SATA link, let alone the second-gen 300MB/s SATA ports present in most modern systems.

Only short burst transfers from the Black’s DRAM cache memory stand to benefit from the drive’s faster host interface. That cache does weigh in at 64MB, which is a lot for a mechanical hard drive. However, 6Gbps SATA support didn’t do much for Seagate’s Barracuda XT, which also has a beefy 64MB cache, so we don’t expect much better from the Black.

I’d expect retailers to trumpet the Caviar’s new-fangled SATA support from the rooftops, so it shouldn’t be difficult to differentiate the drive from the old three-platter, 3Gbps Caviar Black 1TB. You can also tell the two apart based on their model numbers: the old Black is the WD1001FALS, while the new drive carries a WD1002FAEX model designation.

Like other members of the Caviar Black family, Western Digital’s latest is covered by a five-year warranty. Most desktop hard drives offer three-year warranties, which is yet another reason we’ve been recommending the Black for all this time. A longer warranty doesn’t necessarily guarantee your drive will last, of course, but if it fails, at least you’ll be guaranteed a free replacement for longer.

Our test methods
Today, we’ll be looking at the Caviar Black 1TB’s performance against its most natural rival: the two-platter, 1TB Seagate Barracuda 7200.12. Samsung and Hitachi also make two-platter drives with terabyte capacities, and while we’ll be covering both soon, they’ve yet to arrive in our labs.

In addition to letting the Black square off against a terabyte ‘cuda, we’ve thrown a couple of flagship 2TB drives from WD and Seagate into the mix. The Caviar Black 2TB uses the same platters as the 1TB model and has that fancy dual-stage actuator, but it only supports 3Gbps SATA. Seagate’s Barracuda XT 2TB has a 6Gbps SATA port and uses the same 500GB platters as the 7200.12.

All four drives were tested on the 3Gbps SATA controller inside Intel’s P55 Express PCH. Since they both support third-gen Serial ATA, the Caviar Black 1TB and Barracuda XT 2TB were also tested on Marvell’s 9123 SATA 6Gbps controller with the company’s 1.0.0.1027 drivers. The Marvell controller puts two 600MB/s SATA ports behind a single PCIe 2.0 link that offers 500MB/s of bidirectional bandwidth, so it’s not an ideal implementation. However, the 9123 is the only SATA 6Gbps controller currently on the market, and it should be plenty fast for our purposes today. Neither the Barracuda XT nor the Caviar Black is capable of fully exploiting a 600MB/s SATA link.

We used the following system configuration for testing:

Our test system was powered by an OCZ GameXStream power supply unit.

With the exception of our power consumption and noise levels, all tests were run at least twice, with the results averaged. We used the following versions of our test applications:

• WorldBench 6 Beta 2
• Intel IOMeter v2006.07.27
• Xbit Labs File Copy Test v0.3
• HD Tach v3.01
• Far Cry 2 v1.3
• Call of Duty 4 v1.4

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
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 1TB Black ties the WorldBench overall score of its 2TB Caviar counterpart when the two are connected to the Intel storage controller. However, the terabyte drive scores one point higher when it’s paired with the 6Gbps Marvell controller. We’re only looking at a two-point spread between the fastest and the slowest drives, so don’t get too excited about 6Gbps SATA just yet.

Most of WorldBench’s multimedia editing and encoding tests offer equivalent performance with the drives we’re looking at today. The Photoshop test does run notably quicker on the Caviars, though. The 2TB drive is a hair faster, but the terabyte model has a decent lead over the rest of the field, at least when connected to the Intel storage controller. When paired with the Marvell controller, the Caviar’s performance slows by about 7%.

The terabyte Black proves a little quicker in WorldBench’s Office test, but it’s trailed closely by the Barracuda XT. Scores from the web browsing and multitasking tests are too tight to call.

While there’s little difference in WinZip performance between the drives, WorldBench’s Nero test produces some interesting results. The Caviar Black 1TB is quite a bit quicker with the Marvell controller this time around, yet it’s barely slower than the 7200.12 when both are connected to the Intel chipset.

Boot and load times
To test system boot and game level load times, we busted out our trusty stopwatch.

The terabyte Black is the slowest drive to boot by a couple of seconds. It’s only about a second slower than the 7200.12, though.

Don’t pay too much attention to the Marvell scores here. The 88SE9123 appears to be in no hurry to initialize drives, which is why the Barracuda XT and Caviar Black are both slower than when connected to the Intel PCH.

Call of Duty 4 level load times don’t vary much from one drive to the next. Load times in Far Cry 2 are close, as well, with just over half a second separating the fastest configuration from the slowest one.

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. We’ve converted those completion times to MB/s to make the results easier to interpret.

Vista’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 or 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 Vista’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 Caviar Black 1TB fares well in this batch of file creation workloads, besting the 7200.12 with each test pattern. Only the video test pattern really seems to benefit from switching to the Marvell controller, at least with the Caviar—the Barracuda XT is much faster when connected to the Marvell than it is when running on the Intel chipset.

As you can see, the 2TB Caviar Black is measurably faster than the terabyte model across all three test patterns. With a sequential task like file creation, I suspect we’re seeing the 2TB drive’s greater outer-edge area at work.

Again, the terabyte Black pulls up short of its 2TB counterpart. However, the Caviar is still faster than the 7200.12 with the program files and video test patterns.

Do be wary of the Marvell controller, though. While the terabyte Black is quicker in 6Gbps mode with the MP3 test pattern, it’s actually slower with the other two.

Our copy tests combine read and write operations, and they prove a little problematic for the terabyte Caviar, which lags behind the 7200.12 with two test patterns and ties it with the third. Fire up the Marvell controller, and the Black’s performance improves slightly to eclipse that of the 7200.12.

IOMeter
IOMeter presents a good test case for both seek times and command queuing.

Western Digital drives have long offered the best performance with our IOMeter workloads, and the terabyte Black is no exception. Interestingly, the drive’s a little slower than the 2TB model. I wouldn’t expect outer-edge area to make much of a difference with the random access patterns of our IOMeter workloads, but perhaps we’re seeing the 2TB Black’s second actuator stage coming into play here.

Even though it trails the 2TB Black, the terabyte model still offers much higher transaction rates than either Barracuda. Notice also that, on the Marvell controler, the Caviar and Barracuda XT run into a performance bottleneck when the load increases beyond 32 concurrent I/O requests. When connected to the Intel controller, both drives continue to ramp up transaction rates all the way to 256 I/O requests.

Nothing to see here, folks. IOMeter uses a tiny fraction of our test system’s CPU resources with each drive.

HD Tach
We tested HD Tach with the benchmark’s full variable zone size setting.

Areal density matters, folks. The Caviars’ platters offer 53 more gigabits per square inch than those inside the Barracudas, and that translates to faster sequential transfer rates in this HD Tach drag race. The terabyte Black is 3-4MB/s quicker than the 7200.12 here.

3800MB/s from a third-gen SATA interface that’s supposed to top out at 600MB/s? Something’s fishy here, and it turns out the culprit is Marvell’s drivers, which carve out a slice of system memory to use as a drive cache. So let’s ignore those results and focus on the Intel controller, which has the terabyte Black out ahead of the 7200.12 by 9MB/s in HD Tach’s burst speed test. Oddly, the Black 1TB is 10MB/s slower than the 2TB drive, despite the fact that both use a 64MB cache.

As has become tradition, the Caviars register much quicker random access times than their Barracuda rivals. You’re looking at a difference of two-and-a-half milliseconds between the terabyte Black and 7200.12. That margin might not sound like much, but keep in mind we’re working with in the multi-GHz confines of a modern PC. In that sort of environment, milliseconds matter.

I don’t want to make too much of the 0.2-millisecond difference between the terabyte and 2TB Caviar Blacks here. However, I can’t help but wonder if this is another indication of the performance penalty associated with the terabyte drive’s lack of a dual-stage actuator.

Our CPU utilization results are within HD Tach’s +/- 2% margin of error for this test.

Noise levels
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 Tach seek load. Drives were run with the PCB facing up.

Our noise level and power consumption tests were both conducted with the drives connected to the P55 rather than the Marvell controller.

Although thankfully not as noisy as the 2TB Black, the terabyte model is a few decibels louder than Seagate’s Barracudas. At idle, the difference isn’t noticeable to my ears. However, the Black definitely chatters louder than the 7200.12 when seeking. I can easily hear the difference from a few feet away, although I should point out that we’re running this system on an open test bench rather than inside an enclosure.

Power consumption
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.

Surprisingly, the terabyte Black has similar power consumption to the 2TB drive, even with half the number of platters. The 7200.12 is the most efficient of the bunch, drawing about a watt less than our comparable Caviar.