February 2nd, 2006 by eugene
SR's 250 GB Drive Roundup
When StorageReview launched back in 1998, we made a conscious decision to generally review what we coined as "flagship" drives, the largest hard drive in a given family. In turn, a "family" consisted of a drives that featured identical specs; only capacity differed through the use of a varying number of platters and/or platter sides. These platters themselves, however, featured the same areal density.
Why such an arbitrary decision?
In the good old days, new families (in other words, new densities) hit the market as often as every four months! Reviewing more than one representative drive from a family while addressing products from all major manufacturers proved too daunting a task. Balancing timeliness and breadth of coverage required that SR address just one product from each family.
Secondly, not all drives from all manufacturers were available at matching capacities. If manufacturer A offered 40-, 30-, and 20-gigabyte drives while manufacturer B presented products of 30 and 25 gigabytes with C offering 40-, 35-, and 25-gigabyte units, then which capacities would be "fair" to review? Always sticking with the largest drive allowed SR a clearly defined, impartial method to choose evaluation units.
Finally, drives within a given family differed only in capacity. While capacity can exert a slight effect on high-level performance (in the past, informal tests have suggested that a doubling of capacity results in a performance increase of about 7% due to lessened actuator seek distances), it was easily argued that a flagship drive's performance results consistently translated downward to smaller members within the same generation.
Times have changed
This system worked quite well in the early years. More recently, however, the traditional definition of a drive family has fragmented. Though drives may all fall under the same "family" as specified by a manufacturer, in reality these units could feature differing densities, seek times, and buffer sizes. The rationale that "all drives within a given family perform similarly" has lost ground.
Further, some manufacturers often choose to sit out in pushing the bleeding edge of the capacity envelope and instead choose to devote resources to bringing more practical (and higher-volume) products to the market. A recent example rests with Samsung's 250-gigabyte SpinPoint P120. A modest two platter design, the flagship 250 GB SpinPoint was forced to go head-to-head with 500 GB monsters from Hitachi and Seagate (not to mention WD's slightly smaller but no less formidable Caviar WD4000). Our standalone review of the drive thus presented a Samsung unit that sometimes lagged competing drives by considerable margins. The other drives, however, enjoyed a capacity advantage significant enough to impact performance measures. Inversely, we praised the SpinPoint's low noise and power levels... but the Samsung was a two-platter design going up against drives incorporating up to five platters.
Lastly, advancements in areal density (and thus capacities and families) have also dramatically slowed in recent years. As a result, the industry now enjoys the "luxury" of windows of up to a year or more between product refreshes. This permits closer looks at more than one product from each firm's lines. While reviewing all drives from each family remains out of the realm of practicality, combining our traditional approach of individually reviewing each family's flagship with following articles that examine units catering to popular capacity points is a realistic goal.
Hence, we are pleased to be able to present our first ever look at a roundup of smaller, wallet-friendly offerings. Let us move on to StorageReview's 250 gigabyte drive roundup!
As of this writing, the 250 gigabyte capacity point represents a sweet spot at which consumers receive the most space per dollar spent. All five major SATA manufacturers offer a 250 GB model with a street price that hovers around $100.
All drives in this roundup feature two or three platters and incorporate a native SATA interface running at either 1.5 or 3.0 Gb/sec (remember, however, that this figure represents a ceiling rather than achieved rates). Some manufacturers outfit their budget-oriented drives with 8 megabytes of buffer while two have doubled cache to roomier 16-megabyte standard.
Two firms offer 4-pin molex style power connectors in addition to the 15-pin SATA standard while the other three have gone 15-pin only.
Access Time and Transfer Rate
For diagnostic purposes only, StorageReview measures the following low-level parameters:
Average Read Access Time- An average of 25,000 random read accesses of a single sector each conducted through IPEAK SPT's AnalyzeDisk suite. The high sample size permits a much more accurate reading than most typical benchmarks deliver and provides an excellent figure with which one may contrast the claimed access time (claimed seek time + the drive spindle speed's average rotational latency) provided by manufacturers.
Average Write Access Time- An average of 25,000 random write accesses of a single sector each conducted through IPEAK SPT's AnalyzeDisk suite. The high sample size permits a much more accurate reading than most typical benchmarks deliver. Due to differences in read and write head technology, seeks involving writes generally take more time than read accesses.
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 Hitachi Deskstar T7K250 weighs in with a measured access time of 12.8 ms and continues the line's tradition of leading the pack in this low-level measure. A bit more surprising, however, is the Seagate Barracuda 7200.9's placing at 13.3 ms. Traditionally, Seagate's drives have lagged somewhat in the access time department. Here, however, it appears that relieving the drive of an arm or two brings the 250-gigabyte unit in line with that of the competition. When it comes to write accesses, the Maxtor DiamondMax 10 mimics the high score turned in by the MaXLine III (not shown, see the SR Performance Database). Even so, its important to note that write accesses are heavily mitigated by buffering- the Maxtor's lagging score here should not significantly affect high-level performance.
Unsurprisingly, the drives featuring 125-gigabyte media lead the pack when it comes to transfer rates. Samsung's SpinPoint P120 is the only unit to breach the 70 MB/sec barrier, managing a 3.5 MB/sec lead over the Barracuda. Given its 83 GB platters, it is also no surprise that WD's Caviar brings up the rear... 10 MB/sec short of the Samsung. For a more complete picture of transfer rates, look over the graphs found here.
It is important to remember that seek time and transfer rate measurements are mostly diagnostic in nature and not really measurements of "performance" per se. Assessing these two specs is quite similar to running a processor "benchmark" that confirms "yes, this processor really runs at 2.4 GHz and really does feature a 400 MHz FSB." Many additional factors combine to yield aggregate high-level hard disk performance above and beyond these two easily measured yet largely irrelevant metrics. In the end, drives, like all other PC components, should be evaluated via application-level performance. Over the next few pages, this is exactly what we will do. Read on!
StorageReview uses the following tests to assess non-server use:
StorageReview.com Office DriveMark 2006- A capture of VeriTest's Business Winstone 2004 suite. Applications include Microsoft's Office XP (Word, Excel, Access, Outlook, and Project), Internet Explorer 6.0, Symantec Antivirus 2002 and Winzip 9.0 executed in a lightly-multitasked manner.
StorageReview.com High-End DriveMark 2006- A capture of VeriTest's Multimedia Content Creation Winstone 2004 suite. Applications include Adobe Photoshop v7.01, Adobe Premiere v6.5, Macromedia Director MX v9.0, Macromedia Dreamweaver MX v6.1, Microsoft Windows Media Encoder 9.0, Newtek Lightwave 3D 7.5b, and Steinberg Wavelab 4.0f run in a lightly-multitasked manner.
For more information, please click here.
The Caviar WD2500KS leads the pack with 769 I/Os per second in StorageReview's Office DriveMark 2006, our bellwether non-server performance measure. WD's lead is minimal, however, with Maxtor's DiamondMax 10 and Samsung's SpinPoint P120 nipping at the Caviar's heels. Resting another notch back is Hitachi's Deskstar T7K250. The Seagate Barracuda 7200.9 brings up the rear- at just 510 I/Os per second, the gap between the 7200.9 and the competition is quite signicant.
Hitachi and Samsung tie to claim the top slots in the SR High-End DriveMark 2006 with NCQ-disabled scores of 546 I/Os per second. WD's Caviar trails slightly, followed by the DiamondMax. Once again, the Barracuda 7200.9 holds the floor. Overall, however, differences in the High-End DriveMark are slight, with just 11% separating the Barracuda from the Deskstar and SpinPoint.
Three decidedly different entertainment titles cover gaming performance in StorageReview's test suite.
FarCry, a first-person shooter, remains infamous for its lengthy map loads when switching levels.
The Sims 2, though often referred to as a "people simulator," is in its heart a strategy game and spends considerable time accessing the disk when loading houses and lots.
Finally, World of Warcraft represents the testbed's role-playing entry; it issues disk accesses when switching continents/dungeons as well as when loading new textures into RAM on the fly.
For more information, please click here.
The Deskstar and DiamondMax jointly lead the pack in our trace of FarCry's level-loading and gameplay disk access. WD's Caviar comes in just behind the Hitachi and Maxtor. Seagate manages to climb out of the cellar here, besting Samsung's SpinPoint to claim fourth place.
Hitachi's drive manages to put just a bit of distance between itself and the Maxtor in our Sims 2 suite. WD again occupies the middle while Samsung's drive makes a comeback and wrests the fourth slot from the Barracuda.
In our World of Warcraft test, the Caviar claims the number one spot. Always a contender, Hitachi's Deskstar trails a bit. The DiamondMax lands a notch behind the leading pair with the offerings from Seagate and Samsung coming in ever so slightly behind.
Unlike single-user machines (whether a desktop or workstation), servers undergo highly random, non-localized access. StorageReview simulates these multi-user loads using IOMeter. The IOMeter File Server pattern balances a majority of reads and minority of writes spanning requests of varying sizes.
IOMeter also facilitates user-configurable load levels by maintaining queue levels (outstanding I/Os) of a specified depth. Our tests start with the File Server pattern with a depth of 1 and double continuously until depth reaches 128 outstanding I/Os.
Drives with any sort of command queuing abilities will always be tested with such features enabled. Unlike single-user patterns, multi-user loads always benefit when requests are reordered for more efficient retrieval.
For more information click here.
Drives featuring NCQ clearing outgun WD's offering, the sole unit in the roundup that fails to offer SATA's command reordering functionality. Leading the group at very high queue depths is Maxtor's DiamondMax 10... no other contender comes close when concurrency exceeds 32.
More practically, however, any 7200 RPM drive regularly experiencing queue depths greater than 8 is being overworked. Under such a purview, Seagate's Barracuda 7200.9 offers the best overall performance, competitive right out the gate and scaling relatively well. The Deskstar T7K250's NCQ implementation, mimicking that of the Deskstar 7K500, remains unimpressive.
Noise and Power Measurements
Idle Noise- The sound pressure emitted from a drive measured at a distance of 3 millimeters. The close-field measurement allows for increased resolution between drive sound pressures and eliminates interactions from outside environmental noise. Note that while the measurement is an A-weighted decibel score that weighs frequencies in proportion to human ear sensitivity, a low score does not necessarily predict whether or not a drive will exhibit a high-pitch whine that some may find intrusive. Conversely, a high score does not necessarily indicate that the drive exhibits an intrusive noise profile.
Operating Power Dissipation- The power consumed by a drive, measured both while idle and when performing fully random seeks. In the relatively closed environment of a computer case, power dissipation correlates highly with drive temperature. The greater a drive's power draw, the more significant its effect on the chassis' internal temperature.
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.
Samsung's SpinPoint and Seagate's Barracuda offer the lowest idle noise floors of the five drives in the roundup, offering objectively-measured scores of about 39 A-weighted decibels measured from a distance of 3 millimeters. The DiamondMax and Deskstar weigh in about 1.5 to 2 decibels higher while WD's Caviar brings up the rear with a measured score significantly greater than the competition. Subjectively speaking, however, the WD2500KS's idle noise is hardly something worth sounding an alarm over. Its 43.6 decibels remains inaudible over most fans.
Again subjectively, seek noises for the five drives are all dull and muted. The Deskstar and DiamondMax's noises are, perhaps, just a wee bit more substantive that the others.
Despite its three-platter design, the Caviar WD2500KS leads the pack in dissipating just 9.4 watts under a full load. At idle, the WD's 6.8 watts more or less ties that of the Barracuda 7200.9 and trails only the Deskstar T7K250. The Deskstar, however, swallows up significantly more power when seeking. The SpinPoint P120 offers the second-best overall power profile, though the Barracuda makes a convincing case. Maxtor's DiamondMax 10 rests on the floor, with the highest marks in both idle and seek power draw measures.
Only the Deskstar and Barracuda stand out in our peak power draw assessment. Hitachi's unit continues the firms category-leading tradition while the Seagate continues to draw significantly more current than other drives.
How do 250 GB drives compare to their respective flagships?
As mentioned before, our long-held rule that "all drives within a given family generally perform the same" is no longer the rock that it has been in the past. Drives within the same generation can now differ in capacity, density, specified seek times, transfer rates, and buffer size- all factors that may significantly impact high-level performance. Hitachi and Seagate have for several months been shipping 500-gigabyte behemoths that double the capacity of the drives reviewed in this roundup. Western Digital offers a 400 GB unit that, while not quite twice the size of a 250 GB drive, nonetheless offers a considerable capacity advantage. Maxtor's flagship 300 GB drive, on the other hand, is just 20% larger than these review units while Samsung's 250 GB SpinPoint P120 is the firm's readily-availble flagship at the time of this writing. Here we will examine the differences in performance between the 250 gigabyte units represented in this roundup and the larger flagship offerings available from four of the manufacturers.
Note: To optimize performance, NCQ was disabled in all cases except for IOMeter's tests.
Despite its larger, heavier actuator, the 500-gigabyte Deskstar 7K500's access times do not suffer when compared with those of the Deskstar T7K250's. The T7K250 enjoys a 3 MB/sec lead over the 7K500 over most of its relative capacity, however, thanks to its denser (125 GB to the 7K500's 100 GB) platters.
Thanks to its larger buffer and size, however, the 7K500 resoundingly defeats its smaller brother in all of our single user measures. The difference is particularly noteworthy under the office test where the 500-gigabyte unit blows away the T7K250 in the SR Office DriveMark 2006 by a margin of 27%.
Mirroring the lack of difference in access times, both drives yield virtually identical scores under our multi-user suite... to be expected with the test's highly-random, non-localized nature.
Given the 7K500's higher platter count (five versus the T7K250's two), it is unsurprising that the larger drive generates significantly more noise and heat than the 250-gig unit.
Note: To optimize results, NCQ was enabled in all tests. The MaXLine III and DiamondMax 10 lines are physically identical, differing only in post-construction testing and warranty coverage.
Maxtor's 250 GB and 300 GB DiamondMax 10 both feature three identically zoned platters. Despite effectively being short stroked by a small amount, however, the 250-gig drive posts slightly higher access times than the 300-gig unit.
The real story here, however, is the superior performance that the smaller 250 GB drive delivers when contrasted with the 300. Differences are not huge, but remain consistent in the 250 GB drive's favor, the opposite of what one would expect. Also of note is the 250-gig unit's better showing in our noise and power measurements.
Our 300 GB MaXLine III sample was manufactured in June of 2004 while the 250 GB DiamondMax 10 was made in October of 2005. Presumably, Maxtor's code has not stood still since then and has enjoyed revision similar to, say, those we have covered in our recent revisit of WD's 74-gig Raptor.
Therefore, while we have chosen to present these results side by side for consitency's sake, we should note that more recently made 300-gigabyte DiamondMax 10s would likely exhibit slightly higher performance than the sample we have presented here.
Note: To optimize performance, NCQ was disabled in all cases except for IOMeter's tests.
The 250-gig Barracuda enjoys a small but real access time advantage over the 500-gig unit. This discrepancy likely arises from smaller drive's lighter actuator. Also of note is the difference in zoning between the two drives. Though both units feature 125-gigabyte platters, the 250 GB drive reaches nearly 68 MB/sec while the larger disc tops out at just 62 MB/sec. Given the latter's increased head count and associated alignment issues, our guess is that Seagate had to choose a more conservative zoning pattern for the larger sample.
Transfer rates and access times, as we have stated many time in the past, have relatively little to do with performance under today's applications. Rather, it is the buffer and its associated caching strategies that matter. The larger 7200.9 features a 16-megabyte buffer compared to the 250-gig unit's 8. Thus, despite its inferior low-level measurements, the 500-gigabyte 7200.9 enjoys a considerable performance advantage over the 250-gigabyte unit in single-user applications, with the most marked difference (20%) arising under the Office DriveMark.
The smaller drive's access time advantage strikes back, however, under IOMeter's tests. Note, however, that IOMeter ideally runs on an unpartitioned drive. Hence, the 250 GB drive experiences random accesses across all 250 GB of its space while the 500 GB unit likewise accesses sectors all over its 500 GB of capacity. Were the 500-gig drive limited to a real data set of 250 GB, it would enjoy a performance advantage over its smaller brother.
Finally, as was the case when comparing Hitachi's drives with each other, the two-platter drive possesses an acoustic and thermal advantage over the four-platter entry.
Note: To optimize the WD4000KD's performance, NCQ was disabled in all cases except for IOMeter's tests. The WD2500KS does not feature NCQ.
Though it utilizes four platters to the WD2500KS's three, the WD4000KD manages a slight lead in random read access time. Its newer 100-gigabyte platters also permit it a slightly higher overall sequential transfer rate.
Both drives feature 16-megabyte buffers. The WD2500KS, WD's first all-native SATA drive, ships with a 3.0 Gb/sec interface. The WD4000KD, leveraged from the firm's Raptor design, comes equipped with a more conservative 1.5 Gb/sec signaling scheme. Even so, however, the WD4000KD leverages a fundamentally newer design and its higher capacity to deliver better across-the-board scores than the WD2500KS, highlighted by the former's 18% advantage over the latter in the SR High-End DriveMark 2006.
The WD2500KS lacks NCQ, an omission that hurts the drive under IOMeter, a suite that simulates a server load. Here the WD4000KD, equipped with NCQ, delivers a markedly better showing.
Finally, as one would expect, the WD4000KD's increased media count draws more power and thus produces more heat. When it comes to noise, however, the larger drive's objectively-measured acoustic floor weighs in significantly lower than that of the WD2500KS's. In reality, though, both units feature the same unobtrusive operation.
Long-time readers no doubt recognize the tables above that compare two drives with each other from the StorageReview Drive Performance Database. This unique feature allows readers to sort all drives reviewed in our current testbed by a variety of metrics. After choosing a sort, readers may select individual drives for an in-depth, custom, side-by-side comparison. We invite those who have yet to experience the power and versatility of the Performance Database to try it out!
Much of the data compiled for this review underscores some trends already discerned by knowledgeable readers - features such as a faster 3 Gb/sec interface and Native Command Queuing do not necessarily translate into a real, measurable performance gain.
That said, a summary of what we have found in putting each of these five drives to the test:
Hitachi Deskstar T7K250: The Deskstar 7K500 is easily the fastest 7200 RPM drive around. A bit gets lost in translation, however, when moving down to the 250-gigabyte model. The T7K250 lags the WD Caviar WD2500KS by a bit in the Office DriveMark and places fourth overall. Hitachi's drive finishes first or second in all other performance tests, however, and as a result is the fastest overall drive featured in this roundup. Its seek noise and active power draw, however, remain on the high side when contrasted with the competition.
Maxtor DiamondMax 10: This aging design seems to have a lot of pep remaining, especially evidenced by the smaller but more recently revised 250 GB drive's besting of the slightly larger (but older) 300 GB flagship. Maxtor's drives tend to enjoy a cost advantage... at the time of this writing, it is the only unit in the roundup with a best-case price tag resting just under $100. A decent performer, the DiamondMax 10's power dissipation levels and seek noises, like the Deskstar, sit slightly above the rest of the pack.
Samsung SpinPoint P120: Samsung's flagship no longer stands contrasted against competing units twice its size. In our productivity suite, the drive offers decent performance, placing near the top in both the Office and High-End tests. For gaming purposes, however, Samsung's design languishes near the bottom of the charts. On the plus side, even when going up against same-size competition, the SpinPoint's noise levels and power consumption are among the best around.
Seagate Barracuda 7200.9: In its flagship 500 GB incarnation, the Barracuda 7200.9 is a middling performer at best. The loss of half its capacity and half its buffer leaves the 250-gigabyte 'Cuda a relatively poor performer. Though it offers environmental and gaming performance on par with the SpinPoint, Seagate's drive does not match Samsung's strong productivity scores. The Barracuda's server scores, on the other hand, lead the pack. No discussion of the 7200.9 is complete without mentioning that Seagate remains the only manufacturer to back its SATA drives with the same robust 5-year warranty that protects its SCSI line. While we do not necessarily believe that a longer warranty automatically translates into a more reliable product, it is clear that many readers do- if you are looking for peace of mind, get a Barracuda.
Western Digital Caviar WD2500KS: The WD2500KS is built on an older platform, evidenced most clearly by its less dense, 83-gigabyte platters. It nonetheless offers above-average performance and manages to capture the number one slot in the Office DriveMark as well as our World of Warcraft trace. The drive's idle noise floor is not nearly as loud as objective measurements may suggest. It also offers the coolest overall operation of all drives in the roundup. The Caviar is the only review drive that does not offer NCQ - while this is no drawback in single-user applications, the drive's scores suffer under multi-user patterns.
We would be remiss without noting that, overall, performance differences between all five contenders remain relatively small. These days, 250 gigabyte SATA drives are commodities; all players have honed their offerings down to levels where distinguishing between them on the basis of performance, environmental factors, and price can be quite difficult. Absolute fastest? Hitachi. Least expensive? Maxtor. Quietest? Samsung. None of these factors matter? Choose your favorite brand.
As demonstrated by drives from Hitachi, Seagate, and Western Digital, one does give up more than sheer space when moving down to a commodity 250 gigabyte drive from a capacious flagship. Also, Maxtor and Samsung both have larger units right around the corner. Keep this in mind when allocating your budget towards various components... a bit more dough gets you more capacity and more speed.