Reviews Leaderboard Database Reference Search StorageReview Discussion Reliability Survey Search About StorageReview.com Contents

Hitachi Pegasus DK3E1T-91WS

  December 7, 1998 Author: Eugene Ra  
Evaluation unit provided by Hitachi Corp.

Over the last several years, Seagate Technology has been regarded as the pioneer when it came to increasing spindle speeds. Their Barracuda line was the first to hit 7200 rpm. The Cheetah was the first to break the 10,000 rpm barrier. So, with an announcement back in April, Hitachi surprised everyone by announcing the first 12,000 rpm disk. It's December now, nearly 8 months later. Hitachi also seems to be rivaling Seagate in the time-to-market arena . I've received conflicting reports from various Hitachi representatives on the status of the drive. A few at Comdex said the drive was already shipping (I have yet to see it on sale, however). Others estimate a 1st Quarter 99 delivery. As intriguing as the drive is, it would be a sad situation if the drive started out the gate out of date. Though no other manufacturer has risen above the 10k rpm level yet, increasing data densities and refinements of venerable lines promise drives that may perform every bit as well as a year-old 12k design.

The major claim-to-fame of the DK3E1T, or "Pegasus" (I'm not sure how firm the name is, but it sure beats calling it DK3E1T twenty times over in an article, so I'm going to stick with it) is, of course, its 12,000 rpm spindle speed. The twin benefits of a higher rotation speed, lower seek time and higher transfer rates, are both brought to bear with this new plateau. Combined with a blazingly fast seek time of 5.0 milliseconds, such low latency (2.5ms) results in an amazingly low random access time rating of 7.5ms. Like other high-performance units, however, the drive is somewhat of a "cheater" when it comes to platter diameter. To shave time off of seeks, 3.5" form factor drives have been incorporating physically smaller platters. The Pegasus uses platters 3" in diameter. This results in a rather low areal density (considering the upcoming Cheetah 18LP's 3 gigs/platter) of 1.1 gigabytes per platter. To pack 9.2 gigs in a drive, the assembly features 9 platters. This is a 1.6", half-height unit. The drive's buffer, on the other hand, is a bit forward looking: a spacious 2048k. The particular pre-production evaluation unit tested here, the DK3E1T-91WS, is a standard UW-SCSI design. A U2W version will also be available.

The drive itself features heatsink fins on its front and left sides. A little hint of what was to come, for sure. Remember, this is a 12k rpm drive . Installation into our standard SCSI test bed went without a hitch. Expecting reference-quality results from this drive, we've decided to debut tests under ZD's WinBench 99. Over the next three months or so, we'll run WinBench 98 and WinBench 99 side-by-side to establish a reference baseline between the two suites before moving exclusively to the 99 version. WB99's Disk Transfer test is much more comprehensive, testing data transfer rate over all parts of the drive, and hence takes much more time to complete than WB98 did. As a result, we've decided to standardize on three trials instead of five.

[an error occurred while processing the directive]

Yes, we've got a new champion here. In its prerelease form, the Pegasus outran the Seagate Cheetah 9LP in most tests. The drive fell just short of breaking 2500 in its Business Disk WinMark 98 figure under Win95, a full 16% increase over the Cheetah. The increase in the High-End WinMark was a bit more modest 10%. The results carried over somewhat to tests under NT, with the Hitachi scoring 13% and 9% higher than the Seagate in the Business and High-End tests respectively. ThreadMark, as usual, presented some interesting results. In Windows 95, the Pegasus actually lagged behind the Cheetah by 14%. Yet in NT, it leapt ahead by a margin of 19%. As usual, however, use in my own personal system once again proved that ZD's WinBench test correlates far more accurately to performance delivered than ThreadMark does.

With a new level in spindle speed, we can expect new frontiers in heat and noise. While being tested in our tesbed's PC Power & Cooling BayCool drive cooler, the drive ran rather cool, just slightly warm to the touch. With the BayCool's fans disabled, the drive ran rather warmly. Interestingly, however, it didn't reach the too-hot-to-touch temp levels delivered by the Cheetah line. Nevertheless, you need a cooler for this drive.

When it comes to noise, the Hitachi is bar-none the loudest drive I've ever used. The high-pitch whine that one expects from high spindle speed it present, though once again a bit less intrusive than the Cheetah drives. Seek noise, however, is in a class of its own. Even twenty feet away at the other end of the room, the drive's seeks churned, rumbled, and rattled away quite loudly. But what did you expect?

Despite its noise, the Pegasus delivers enviable performance. It truly has the potential to be crowned the world's fastest drive. But time and the competition never stand still either. Though it produces record-breaking figures, the Pegasus design in many ways is falling behind the times. The drive's low capacity-per-platter results in the unwieldy 1.6" form factor. That's huge these days for a mere 9 gigs of space. Such areal density also limits what could have been a stellar sequential transfer rate- instead of pushing the envelope here, the Pegasus merely matches the Cheetah 9LP. Keep in mind that other state of the art 10k rpm designs are on the way from Seagate, Quantum, IBM, and even Hitachi itself. I think what we're seeing here is a bit of a "concept car" paradigm. If the drive makes it out the door soon, it can enjoy a short reign of supremacy. If not, it'll merely serve as a research stepping stone to even faster products. Good luck, Hitachi.

Hitachi DK3E1T-91WS
Estimated Price: TBA
Specifications
[an error occurred while processing the directive]
* Note: Threadmark 2.0 and WinBench98 test results are the average of five trials.
WinBench99 test results are the average of three trials.


HOME | ARTICLES | LEADERBOARD | PERFORMANCE DATABASE | REFERENCE GUIDE
COMMUNITY | RELIABILITY SURVEY | SUPPORT SR! | ABOUT SR |

Copyright © 1998-2005 StorageReview.com, Inc. All rights reserved.
Write: Webmaster