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

One Last Time: The Quantum Atlas 10k II
  August 24, 2000 Author: Eugene Ra  
The following article remains as a reference only. As of November 9th, 2000, figures from Quantum's second Atlas 10k II sample, evaluated here, have been entered in the database to represent the line. As a result, the dynamically-generated bar graphs found in this article will feature figures representing the second sample (DA40) rather than the third sample (DE46) as the article text indicates. DE46's full results, however, may be found here. DA40 is the firmware revision that ships with all Atlas 10k II drives.

Atlas 10k II evaluation unit provided by Quantum Corporation
See also Quantum's Atlas 10k II vs. Seagate's Cheetah 36LP: A Grudge Match!
See also Quantum Atlas 10k II review

Introduction

Quantum's Atlas 10k II has followed a topsy-turvy path in the eyes of StorageReview.com readers. Originally announced early November 1999, the 10k II's specs captivated SR readers... 10k RPM, 4.7 ms seek time, 7.3 GB/platter and an 8 MB buffer. What wasn't to like? Things hummed along right smoothly as Quantum managed to deliver us one of these hotly anticipated drives in February. Reviewed in our original testbed under our original methodology, the Atlas easily posted record-breaking scores in all low-level and high-level tests.

March 12th, 2000 marked StorageReview.com's second anniversary. We took the opportunity to commemorate the event with the unveiling of our all-new testbed and IOMeter benchmark suites. Some interesting upsets ensued... the much-heralded Maxtor DiamondMax Plus 40, for instance, turned in decidedly mediocre scores. And the Atlas 10k II, though maintaining (even increasing) it's massive WinBench margins, didn't fare as well when compared to competing drives such as the Seagate Cheetah 18XL in IOMeter.

Eventually drives from IBM and Fujitsu arrived. These units posted top scores in the important IOMeter Workstation Access Pattern tests. Quantum's unit increasingly started to fall by the wayside. As June rolled around, Seagate observed that we tested a prerelease version of the Cheetah 36LP and offered to submit a revised version of the drive featuring release-level firmware. In the interest of providing SR readers with the most accurate performance figures available, we accepted. We also contacted Quantum about a newer Atlas 10k II. After all, we had originally reviewed the drive in February- considering the extensive delays the drive experienced, the unit's firmware surely had been revised. A couple weeks later, another Atlas 10k II had arrived.

In June we published our look at the revised Cheetah 36LP and Atlas 10k II. Interestingly, though Seagate's unit recorded significant increases in its IOMeter (as well as WinBench) scores, the Quantum drive actually exhibited slight but measurable decreases. As a result, despite the Atlas' superior WinBench scores, we felt that the Cheetah 36LP clearly exhibited itself as the superior drive.

A month later, we were contacted by Quantum regarding the results of the second Atlas 10k II sample. Apparently there were some mishaps in communication both between Quantum and its public relations firm as well as between Quantum and StorageReview.com. Quantum claims that the 2nd sample received by SR was not to be reviewed as the final product. Instead, Quantum's PR firm shipped us (as well as other industry analysts) the drive to facilitate advanced looks and qualifications of the product. On the surface one may very well think: "Hey, Quantum's sore since they came out 2nd to the Cheetah 36LP in SR's latest tests." We are, however, are inclined to believe the explanation. Recognizing SR's position, Quantum was eager to get us an initial sample ASAP... hence the first evaluation unit back in February. Then Quantum's PR agency, not realizing that we had already received a prerelease unit (yes... these things happen believe it or not), included us in the batch for their prerelease shipments. The fact that we had written in requesting another sample seems to have been pure coincidence. This is corroborated by the total lack of response that we received from Quantum after requesting a revised unit... apparently our request never reached the appropriate party at Quantum.

The bottom line? Quantum has requested tests on a third sample that they claim is indicative of shipping units' performance. We've agreed to do so; this review presents the results. We realize this decision will not be without controversy. Indeed we had to consider the request very carefully before accepting.

StorageReview.com readers are a demanding lot... and we would not have it any other way! On one hand, they expect SR to maintain its reputation of being the first source to get its hands on a hot new drive. On the other, however, readers expect SR to maintain a database of performance figures that are as accurate as possible. These two expectations are inherently at odds with each other. Pressing a manufacturer to send us an evaluation unit ASAP will result in a review (preview) of a prerelease unit. Declining prerelease units will result in SR's inability to deliver the scoop on new drives: we'd have to leave it to sites which, modestly put, do not feature test suites as comprehensive nor methodologies as rigorous as ours.

In the end, declining to review this third sample would in effect punish Quantum for the initiative they demonstrated in providing SR with an evaluation unit far in advance of any others. It was, after all, the first sample that allowed us (and therefore readers) to get such an early look at the drive. On the opposite end, accepting Quantum's offer of a "final unit" permits SR to satisfy the antithetical issues of timeliness and accuracy as best possible. This policy stands for all manufacturers.

For our part, we'll be more careful in our communication with drive manufacturers and PR firms. We'll do our best to clarify whether a given article covers a prerelease or release product. And we'll continue striving to be the first and the best in bringing readers storage reviews.

This latest sample of the Atlas 10k II features firmware revision DE46, as opposed to the 2nd sample's DA40 and the original's D620. On paper, the drive has always remained the most impressive 10k RPM drive of 2000: 7.3 GB/platter, a 4.7 millisecond seek time, and an 8-megabyte buffer.

Let's see if the third time's the charm!

WB99/Win2k Low-Level Measurements

 Testbed II  Low-Level Measurements Details... 
Windows 2000 Professional using NTFS
Quantum Atlas 10k II Third Sample - 7.8 |
Quantum Atlas 10k II Second Sample - 7.9 |
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) - 8.5 |
IBM Ultrastar 36LZX (18.3 GB Ultra160/m SCSI) - 8.5 |
Seagate Cheetah 18XL (18.4 GB Ultra160/m SCSI) - 8.9 |
Seagate Cheetah 36LP (36.7 GB Ultra160/m SCSI) - 9.0 |
Windows 2000 Professional using NTFS
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) - 42767 |
Quantum Atlas 10k II Second Sample - 41467 |
Quantum Atlas 10k II Third Sample - 38400 |
Seagate Cheetah 36LP (36.7 GB Ultra160/m SCSI) - 36167 |
Seagate Cheetah 18XL (18.4 GB Ultra160/m SCSI) - 35967 |
IBM Ultrastar 36LZX (18.3 GB Ultra160/m SCSI) - 34800 |
Windows 2000 Professional using NTFS
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) - 28300 |
Quantum Atlas 10k II Third Sample - 25100 |
Quantum Atlas 10k II Second Sample - 25000 |
Seagate Cheetah 18XL (18.4 GB Ultra160/m SCSI) - 25000 |
Seagate Cheetah 36LP (36.7 GB Ultra160/m SCSI) - 24100 |
IBM Ultrastar 36LZX (18.3 GB Ultra160/m SCSI) - 22800 |

Click here to examine the STR graph for this drive

An interesting anomaly arises when examining the sequential transfer rates achieved by the third sample. The two previous samples both managed to break the 40 MB/sec barrier in their outer zone. This unit, strangely enough, does not. The 38.4 MB/sec figure listed above is a bit misleading... a look at the STR graph shows the drive missing 40 MB/sec several tracks inwards by just a hair. We contacted Quantum, wondering perhaps if there had been some physical changes to the drive... perhaps a lower sector-per-track count in the outer zone. Quantum states that all physicals have remained the same, however. An engineer hypothesized that the drive may have negotiated an ultra-wide rather than ultra160 connection with our host adapter. A check of the drive in my personal system, however, confirmed the drive, even operating in LVD mode, failed to break 40 MB/sec. Apparently the drive achieved an excess of 41 MB/sec before being shipped out to us (as all of them should). We weren't able to get to the bottom of the problem; Quantum maintains that this is an aberration and that all shipping 10k II units will maintain their impressive outer-zone STRs. Since the small reduction in STR doesn't affect high-level results, both Quantum and StorageReview.com agree that the high-level figures that follow are representative of a full-STR unit.

This third sample maintained the impressive access time that we witnessed in previous units. It's score of 7.8 milliseconds is right in line with its advertised 4.7 ms seek time.

WB99/Win2k WinMarks

 Testbed II  WB99/Win2k WinMarks Details... 
Windows 2000 Professional using NTFS
Quantum Atlas 10k II Second Sample - 8763 |
Quantum Atlas 10k II Third Sample - 8607 |
IBM Ultrastar 36LZX (18.3 GB Ultra160/m SCSI) - 7233 |
Seagate Cheetah 36LP (36.7 GB Ultra160/m SCSI) - 7040 |
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) - 6577 |
Seagate Cheetah 18LP AV (18.2 GB Ultra160/m SCSI) - 6073 |
Windows 2000 Professional using NTFS
Quantum Atlas 10k II Third Sample - 24667 |
Quantum Atlas 10k II Second Sample - 23033 |
Fujitsu MAJ3xxx (18.2 GB Ultra160/m SCSI) - 20100 |
Seagate Cheetah 36LP (36.7 GB Ultra160/m SCSI) - 19667 |
IBM Ultrastar 36LZX (18.3 GB Ultra160/m SCSI) - 19200 |
Seagate Cheetah 18LP AV (18.2 GB Ultra160/m SCSI) - 15600 |

The Atlas 10k II easily defends its margins in the ZD Disk WinMarks. Its Business score of 8.6 MB/sec is 19% higher than the next fastest unit,IBM's Ultrastar 36LZX. The Atlas' High-End result of 24.6 MB/sec is even more impressive, leading the second-place Fujitsu MAJ3xxx by 23%.

Regular readers are likely aware that we're not very high on WinBench these days. It should be noted, however, that Quantum believes WinBench 99 is still the most accurate tool to measure workstation performance of various disks. This in itself would probably surprise no one since the Atlas 10k II is by far the leader in the Disk WinMarks, easily besting even the Seagate Cheetah X15. What's interesting is that Quantum thus joins Seagate in backing WB99's accuracy. With the current state of affairs, Seagate has far less of an interest in propping up WB99 as #1. After all, the ATA IBM Deskstar 75GXP matches the Cheetah 36LP in the High-End Disk WinMark and beats it in the Business test. At any rate, Quantum and Seagate both acquiescing to WB99 is definitely food for thought.

IOMeter Performance

 Testbed II  IOMeter - Workstation Access Pattern - Total I/Os per second Details... 
Windows 2000 Professional using NTFS
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|

Our former Atlas 10k II evaluation units both scored well in our File Server and Database IOMeter access patterns. This third sample modestly improves those already-decent figures. It's in the area of Workstation performance that the newest unit posts spectacular gains. The previous powerhouse in the IOMeter Workstation profile was Fujitsu's MAJ3xxx. The DE46 firmware allows the Quantum to power past Fujitsu's offering in Linear, Very Light, and Light loads by more than 6%. The margins are thinner under Moderate and Heavy loads where the Atlas edges by a margin of one or two percent.

The figures posted by this Atlas are considerably better than its DA40 predecessor... the latter's scores are improved on by 9% - 18%. This coup places the Atlas 10k II at the top of the 10k heap in IOMeter under all three access patterns.

Conclusion

Noise levels remain unchanged with this sample. The drive grinds away when seeking... significantly louder than, say, the Fujitsu MAJ3xxx. One should keep active cooling in mind when incorporating the Atlas 10k II in to a case that isn't well ventilated.

In conclusion, the Atlas 10k II's improvements in its IOMeter scores over its prior units allow it to lay claim to just about every SR benchmark, low or high-level, as the 10k champ. We realize the drive has suffered a somewhat maligned history due to our testing of multiple samples as well as the drive's scarce availability... nonetheless, sweeping figures like these are hard to ignore. The Atlas 10k II is the fastest 10k RPM drive around.

Speaking of availability, the drive has been shipping "in volume" since early June. Unfortunately, a general scarcity of parts that have been plaguing the industry as a whole combined with strong demand from OEMs has kept the Atlas 10k II from appearing in mainstream channels alongside offerings from Seagate and IBM. The rather steep estimated price of $949 likely reflects this shortage. Hopefully the supply problem will ease up as time passes. Once its pricing is in line, the Atlas 10k II is the drive to get.

[an error occurred while processing the directive]

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