Western Digital Raptor WD1500

A Tale of Two Markets, a Tale of Two Drives Since its inception, the Raptor has always officially been positioned by WD as an inexpensive alternative to the SCSI-based drives that dominate the server landscape. Thus far, however, the Raptor has met with only limited success in its declared target market. Instead, it is the enthusiast/gamer community that has carried the line over the past two years. Given the intricacies of its distribution network, WD claims that it is impossible to offer precise figures regarding where Raptor drives actually end up. Until now, WD has never overtly acknowledged the Raptor’s penetration into the enthusiast market. Undoubtedly, however, the company has noticed the family’s revered status among gamers. The firm briefly flirted with distribution of the relatively expensive drive through retail superstores. Further, the latest versions of the WD740GD sport tweaks to single-user speed at the expense of a little server performance. When planning the third-generation Raptor, WD sought a way to acknowledge the gaming market while still preserving the brand in the enterprise. Regarding the latter, the company hopes the third time truly is the charm. This time around, the Raptor will enjoy some advantages not accorded to its predecessors. The industry is gearing up for the inevitable transition from the parallel Ultra320 interface to the new Serial Attached SCSI (SAS) standard.

SATA’s interoperability with SAS (both use the same physical connector and the latter’s HBAs can control the former’s drives) opens up new doors for WD’s fledgling entry. The Raptor is no longer beholden to dedicated SATA infrastructure and instead may be positioned as an inexpensive alternative that remains directly compatible with the SAS world. In addition, SATA’s maturation continues to bring more credibility to the interface as Seagate, Maxtor, and WD all move to bring high-capacity, 24/7 nearline-style SATA drives to the market.

Western Digital’s solution to the conundrum is two separate SKUs. The “Raptor 150” (WD1500ADFD) continues the line’s tradition as a low-cost alternative to the SCSI drives that dominate the server world. The “Raptor X” (WD1500AHFD) caters to the enthusiasts and gamers that have carried the family since its debut.

What’s the difference? Cosmetically speaking, the Raptor X pays homage to the DIY crowd that enjoys things such as clear PC cases with… well, a transparent cover. Though the actual top plate remains a jet-black aluminum compound, the portion that actually sits atop the spindle assembly, constructed from polycarbonate, permits a clear view of the drive’s interior. To further the product’s unique appearance, WD investigated incorporating an LED into the device but ruled it out due to engineering constraints. The Raptor 150, on the other hand, features the more staid, standard look that debuted with WD’s latest 7200 RPM units.

The Raptor X is officially specified with a 600,000 hour MTBF while the 150 features a 1.2 million hour claim. Both units receive 24 hours of factory burn-in are backed by 5-year warranties. As attractive as the concept may sound to enthusiasts, the two different versions do not feature significantly differing firmware. Creating and maintaining two code bases would increase overhead and product prices that would not proportionally translate into better performance. Here at least, differences really do extend only skin deep. Thus, given that the drives are mechanically and electronically identical, it should come as no surprise that the unit featuring the more complexly-engineered cover commands a price premium: WD pegs the Raptor X’s street price at $350 while the standard unit will run about $300.

The latest Raptor, of course, features a 10,000 RPM spindle speed. Claimed average read seek time also remains unchanged at 4.6 milliseconds. Buffer size has doubled to 16 megabytes, matching the size found in today’s latest SATA units. The Raptor maintains a 2-platter design- WD has increased the line’s capacity by doubling platter density to 75 GB, matching the per-platter data stored by contemporary 10K SCSI drives.

Previous iterations of the Raptor utilized a bridged design to mate the drive’s PATA-style electronics with SATA’s physical connector. With the WD1500, the manufacturer finally moves the Raptor series to a native SATA implementation complete with NCQ. Like the server-oriented Caviar RE2, the Raptor features a standard first-generation 150 MB/sec interface rather than the 300 MB/sec conduit found on many newer units. WD feels that electronics capable of negotiating at 300 MB/sec are sufficiently proven for its standard consumer-oriented drives but not quite mature enough to mate with enterprise-class products. The firm indicates that this may change as the relevant ICs continue to develop. For the most part, of course, a 150 MB/sec or 300 MB/sec figure simply represents a ceiling, a bandwidth figure that today’s speediest drives, including this new Raptor, have yet to approach. As we have noted in the past, WD’s Caviar RE2 WD4000YR shares much in common with the Raptor platform. In fact, as time goes by, it becomes more and more apparent that the Raptor WD1500 and Caviar WD4000 spring from the same basic design. Thus, the Raptor enjoys Time Limited Error Recovery (TLER, a feature that prevents false drive dropouts on RAID levels 1 and greater) and Rotary Acceleration Forward Feed (RAFF, electronics that sense and compensate for the vibration commonly found in multi-drive installations that may adversely affect performance), two features that debuted with the Caviar RE. This time around, thankfully, TLER ships disabled by default and may be enabled via a utility for installations that would benefit from the feature.

As previously mentioned, the two “versions” of the WD1500 only differ cosmetically. As a result, the tests that follow on our “Raptor WD1500ADFD” sample represent results for both the gamer-marketed Raptor X and enterprise-oriented Raptor 150. As WD ambitiously targets both the server and the high-end desktop markets, however, the tests that follow will pit the Raptor against all current SCSI and SATA drives.

Some notes on the following tests:

• Unless otherwise noted, all references to the “Raptor WD740GD” refer more specifically to the latest widely available revision, the WD740GD-00FLC0.
• When applicable, all presented results are drawn from the drives operating with command queuing appropriately toggled on or off to yield the best performance:
  • For all SATA drives (accepting Maxtor’s MaXLine III, which performs best with NCQ enabled), NCQ/TCQ is disabled in all single-user tests.
  • NCQ/TCQ is enabled for SATA drives under multi-user tests.
  • For all SCSI drives, TCQ remains enabled in all tests.
• Seagate’s SCSI offerings are toggled to “Desktop” mode for single-user tests and “Server” mode for the multi-user tests.

Come now and join us as we see where Western Digital’s third-generation Raptor lands when compared against every other enterprise and desktop hard disk!

 

 

WB99 Disk/Read Transfer Rate – Begin– The sequential transfer rate attained by the outermost zones in the hard disk. The figure typically represents the highest sustained transfer rate a drive delivers.

WB99 Disk/Read Transfer Rate – End– The sequential transfer rate attained by the innermost zones in the hard disk. The figure typically represents the lowest sustained transfer rate a drive delivers.

For more information, please click here.

 

The Raptor WD1500 weighs in with a measured average read access time of 8.0 milliseconds. Subtracting 3.0 ms to account for the latency of a 10,000 RPM spindle yields a measured seek time of 5.0 milliseconds. WD claims 4.6 ms… our sample misses its claim by 0.4 ms. The older WD740GD, by contrast, manages a 4.7 ms seek. Both drives use the same actuator; the difference likely arises from increased settle times associated with the WD1500’s denser platters.

Random writes average out to 9.0 milliseconds.

Although these scores place WD’s latest squarely at the bottom when contrasted with SCSI drives, it is clear the Raptor nonetheless continues its tradition of integrating SCSI-class mechanics with the SATA interface. When compared against SATA units, of course, the WD1500 (along with its predecessor) remain at the top of the heap.

 

Due to its 75 GB platters, the WD1500 achieves an outer-zone (maximum) transfer rate of 88.3 megabytes per second, an improvement of 23% over the WD740GD. This score essentially matches those turned in by the Maxtor Atlas 10K V and the Fujitsu MAT3300 and leads the Hitachi Ultrastar 10K300 and Seagate Cheetah 10K.7. Against SATA drives featuring higher densities but lower spindle speeds, the newest Raptor easily leads the pack.

Scores decay gracefully as data moves towards the inner tracks thanks to an aggressive zoning pattern. At 60.2 MB/sec, the WD1500 bests all other 10,000 RPM drives and comes close to the figures delivered by 15K RPM beasts.

 

 

 

 

 

WD’s latest delivers 1010 I/Os per second in the StorageReview Office DriveMark 2006 and stands alone as the only drive to burst through the 1000 IOps mark. Representing a 17% improvement over that of its predecessor, the WD1500’s score essentially leapfrogs past a three-way tie previously held by Fujitsu’s MAU3147, Hitachi’s Deskstar 7K500, and WD’s previous Raptor.
 

The WD1500’s 775 I/Os per second improves upon its forerunner by 21% in the SR High-End DriveMark 2006, a suite that features significantly higher average transfer block sizes than the Office test. As a result, the Raptor wrests first place from the MAU3147, besting Fujitsu’s potent SCSI offering by nearly 11%.

 

 

 

 

 

935 I/Os per second places the Raptor WD1500 within a stone’s throw of Fujitsu’s 10,000 RPM MAT3300 and represents a 12% improvement over the score turned in by the WD740GD in our FarCry capture. This gain grants the Raptor a formidable 23% margin over the Deskstar 7K500. The 15K RPM MAU3147, however, maintains a dominant 17% lead over the new Raptor.
 

WD’s latest strikes back when it comes to replaying a trace of The Sims 2’s disk access. With a 21% improvement in delivered IOps over the older WD740GD, the new Raptor tops the charts and blows past the pair of 15,000 RPM titans from Maxtor and Fujitsu. Such a margin, of course, gives the WD1500 a comfortable lead over all other SATA drives.
 

The WD1500’s 775 I/Os per second improves upon the WD740GD by about 15%. Though the Raptor’s showing again easily dominates the SATA field, Seagate’s Cheetah 15K.4 and Fujitsu’s ever-present MAU3147 remain out of reach. The Fujitsu, in particular, boasts a 14% margin over the Raptor.

 

 

 

When it comes to SATA drives, the old meets the new in WD’s implementation of TCQ in the Raptor WD740GD and NCQ in the Raptor WD1500. Out of the gate with just a single operation outstanding, the newer drive enjoys a slight lead over its predecessor despite the latter’s access time advantage, presumably due to firmware-level TCQ’s higher overhead. The WD740GD closes the gap as queue depths reach 2. Moving beyond, however, it is clear that the WD1500’s NCQ execution scales much better at “lower” queue depths. It is not until concurrency hits 16 and above that the WD740GD overtakes the newer drive.

While the older Raptor with TCQ performs better at high queue depths, most single-drive enterprise deployments experience much lower concurrency. Under such depths, the newer Raptor holds an advantage over its predecessor.

When contrasted with NCQ-equipped 7200 RPM drives, the WD1500 stands in a league of its own. Regardless of load, the new Raptor offers significant performance advantages over other SATA drives in the highly-random access that servers undergo.

Despite enjoying a mechanical design similar to 10,000 RPM SCSI drives, the NCQ-equipped Raptor does not scale as well as the competition under increasingly heavy loads. Though the Raptor features access times competitive with these SCSI units, it is clear that SCSI’s robust, mature queuing features combine with their multi-user orientation to delivier significantly more I/Os per second than the SATA drive.
Disparities increase, of course, when the Raptor goes up against 15,000 RPM units.

 

 

Startup (Peak) Power Dissipation– The maximum power dissipated by a drive upon initial spin-up. This figure is relevant when a system features a large number of drives. Though most controllers feature logic that can stagger the spin-up of individual drives, peak power dissipation may nonetheless be of concern in very large arrays or in cases where a staggered start is not feasible. Generally speaking, drives hit peak power draw at different times on the 5V and 12V rails. The 12V peak usually occurs in the midst of initial spin-up. The 5V rail, however, usually hits maximum upon actuator initialization.

For more information, please click here.

 

The Raptor WD1500 posts an objective sound pressure measurement of 42.5 dB/A and matches the level turned in by its predecessor, the WD740GD. Subjectively speaking, the two drives’ idle noise profiles are quite similar… totally unobtrusive and virtually silent. Though on paper drives such as Samsung’s SpinPoint P120, Seagate’s Savvio 10K.1, and WD’s own WD4000KD all appear quieter, in practice, the difference is negligible and will be overshadowed by the noise produced by other fans in the system.

Seek noise remains similar to that of the WD740GD’s. When churning away, the Raptors are noticeably louder than other SATA drives. Perhaps due to its lower platter count (all SCSI drives tested by SR are four-platter flagships contrasted with the Raptor’s two-platter design), WD’s seeks are lighter and more hollow than those produced by the enterprise competition.

 

WD’s latest dissipates approximately one watt more than its predecessor when idle. Under a full seek, the WD1500’s 11.8 watts weigh in just slightly higher than the WD740GD’s 11.6. In either case, the new Raptor consumes less power than the capacious three- to five-platter flagships of the competition. It is only Samsung’s SpinPoint P120, also a two-platter design, that boasts a more modest thermal profile.

Given the drive’s lower platter count, its no surprise that the WD1500 operates much cooler than any SCSI unit save Seagate’s diminutive Savvio 10K.1.

 

With Testbed4’s launch, we chose to draw peak power figures from a standalone AT power supply to minimize interaction with multiple components that may have impacted results. Increasing compliance with SATA-IO’s staggered spinup specification, however, has caused us some difficulty. Drives such as the Deskstar 7K500 and the new Raptor (as well as SAS drives from Seagate and Maxtor not yet formally reviewed) do not power up without a data signal. As a result, to properly assess the Raptor’s startup power requirements, we had to measure the WD1500’s dissipation when operating off of the Testbed’s power supply. This in turn required new measurements for all the drives compared in this review. Thus, the figures presented above, while directly comparable with each other, can not be compared with the peak power numbers found in the SR database.

The WD1500 is relatively current-hungry when contrasted with other SATA drives. Though the older WD740GD manages to obscure itself somewhere in the middle of the pack, the newer Raptor dissipates nearly 33 watts total upon spinup… only Seagate’s Barracuda fares worse.

Such high consumption is par for the course when it comes to SCSI drives, however. Here the WD1500 indeed finds itself in the middle of a relatively power-greedy pack.

 

 

 

 

Differences are particularly pronounced with the Raptor WD1500. While scores from other drives may drop by 3%-5%, NCQ’s presence on WD’s latest hampers its single-user scores by margins of 6%-10%. Though the WD1500 remains a speedy unit when NCQ is enabled, we recommend disabling NCQ in non-server settings to obtain the best possible performance.
NCQ truly shines, of course, in highly-random, multi-user scenarios. Under any level of concurrency, NCQ’s hardware-level reordering delivers considerable advantages. In this particular situation, the disparity is particularly great as the SI3124’s driver seems to innately defeat host-side reordering, perhaps to maximize the gains delivered by the controller hardware itself.

 

 

Let us take a look at how the Raptor family’s performance, noise, and power characteristics have evolved over the line’s 3-year history. Each of the following graphs commences with the original WD360GD, then moves on to the first widely available version of the WD740GD (00FLA1) and the latest WD740GD available at the time of this writing (00FLC0), then concludes with the mighty WD1500. TCQ/NCQ is disabled for single-user tests and enabled for multi-user trials.

The WD360GD came out the gate specified with a 5.2 millisecond seek time and as a result lags its successors when it comes to measured read access times. Interestingly, however, it remains competitive with random writes. The swiftest accesses do not belong to the new WD1500; rather, it’s the WD740GD-00FLC0 that takes the crown, presumably featuring updated code quite similar to that of the WD1500’s… when combined with less dense 37 GB platters, the 00FLC0’s heads take less time to settle and in the end get wherever they need to go a bit faster.
Initial samples of the WD360GD featured an impressive (at the time) 63 MB/sec outer zone transfer rate. As the product hit the channels, however, it became clear that such a transfer rate was overly ambitious- shipping units topped out at 57 MB/sec, igniting consternation across the enthusiast community. As irrelevant as transfer rates were to current applications, WD nonetheless followed up with a 72 MB/sec maximum in the WD740GD, a figure that matched the best 10,000 RPM SCSI units of the day. Today’s WD1500 features a rate of nearly 90 MB/sec, one that is just as fast as contemporary 10K (and even 15K) units… but in the end one that is just as irrelevant today as the WD360GD’s STR was three years ago.
Thanks to its superior size, swifter mechanics, larger buffer, and newer code, the WD1500 commands a 65% performance lead over the original WD360GD in the SR Office DriveMark. Just as notable, however, is the difference between the older and newer WD740GDs. A 12% improvement is nothing to sneeze at.
The WD1500 leaves the WD360GD in the dust by a 68% margin this time around. Again of note is the enhancement the newer WD740GD’s firmware enjoys over the older unit. Here, however, the margin is not as significant when contrasted with the improvement the newest Raptor delivers over either 74 GB unit.
FarCry’s largest increase in performance arises when moving from the WD360GD to the WD740GD-00FLA1. Note, however, that FarCry’s aggregate disk accesses span beyond the 36 GB drive’s inner most zones and in effect “wrap” back to the beginning of the drive, perhaps inducing some broad-stroke seeks that a larger drive (remember, Testbed4’s traces were drawn off of a 40 GB partition) does not encounter. These cases are nonetheless rare and should not have significant bearing on the drive’s results.
As was the case with FarCry, The Sims 2 makes full use of the 40 GB partition from which its accesses were recorded. Hence, the WD360GD may have encountered an occasional broad-stroke seek not codified by the application itself. The WD1500, at any rate, enjoys a sizable lead over all of its predecessors.
Both versions of the WD740GD become identical twins in our World of Warcraft by turning in duplicate scores of 671 I/Os per second. Moving from the WD360GD to a WD740GD yields a slightly greater improvement than going from the WD740GD to the WD1500.
The WD360GD’s lack of command queuing remains glaringly obvious when placed against its successors. Of note here is the newer WD740GD’s prowess compared to the older version under lower queue depths… at the expense of high-concurrency performance. The graph also clearly demonstrates the WD1500’s more consistent NCQ implementation against the WD740GD’s more disparate scaling achieved via legacy TCQ.
While the 00FLA1 revision of the WD740GD jumps out as the only outlier in objective measurements, it is actually the WD360GD that features the most annoying noise profile. When idle, the 36-gigabyte drive exhibits an annoying upper-midband whine reminiscent of an older SCSI drive. By contrast, the other three units, all equipped with FDB motors, deliver a remarkably unassuming noise floor.
Though it features only one platter, the older WD360GD dissipates the most power at both idle and under a full seek. The family hits its coolest point with the WD740GD-00FLC0, with the WD1500 regressing upwards ever so slightly.
Perhaps due to the electronics associated with its all-new native SATA design, the WD1500 draws more peak current than older units when spinning up from a cold start.

 

 

According to filtered and analyzed data collected from participating StorageReview.com readers, a predecessor of the Western Digital Raptor WD1500, the Western Digital Raptor WD740GD , is more reliable than of the other drives in the survey that meet a certain minimum floor of participation.

Note that the percentages in bold above may change as more information continues to be collected and analyzed. For more information, to input your experience with these and/or other drives, and to view comprehensive results, please visit the SR Drive Reliability Survey.

 

 

The WD1500’s standing in the enterprise world remains less clear. 10K RPM 147 gigabyte Ultra320 SCSI drives sit squarely within the Raptor’s price bracket. Unencumbered with the concern of delivering blazing desktop performance and featuring time tested, mature command queuing implementations, these offerings from Seagate, Maxtor, Fujitsu, and Hitachi all significantly outdistance the WD1500 under even the lightest of concurrent loads. To compete in the market that WD covets so much, the Raptor must seek out some sort of advantage such as price or ease of integration with the budding SATA nearline field. As of this moment, the relative infancy of SAS’s supporting infrastructure when contrasted with even SATA’s nascent offerings may give the Raptor the nod in this regard. Such a lead, however, is temporary at best… SAS (and its compatibility with SATA drives) is the ordained successor to parallel SCSI and will inevitably level the playing field within the WD1500’s product lifespan.

Thus, as of right now, the Raptor’s primary advantages remain in the non-server field. Here the WD1500 once again demonstrates that efficient code coupled with a large buffer is the dominant factor when it comes to solving today’s localized data access patterns. Manufacturers of 10K and 15K RPM SCSI drives simply do not take the time (and indeed, do not really care one way or another) to cater to the non-server market by aggressively coding and continuously refining firmware for single-user access paradigms.

If high capacities are relatively unimportant and money is no object, it is the Raptor WD1500, not a 15,000 RPM SCSI unit, that is the drive to get for single-user setups. In other cases where space is a concern and/or funds remain limited, the WD1500 nonetheless succeeds in creating an appetizing conundrum for those torn between massive storage and the fastest performance money can buy.

Western Digital expects to ship the WD1500ADFD (the standard unit) in volume by the end of this month with the Raptor X (clear cover) following by winter’s end. StorageReview will, of course, bring readers any developing news on the new Raptor as soon as possible.

  Review Discussion

Update 4/6/10 – Western Digital VelociRaptor 600GB Review