ReviewsLeaderboardDatabaseReferenceSearch StorageReviewCheck Storage PricesDiscussionFAQsReliability SurveySearchAbout StorageReview.com Contents

Seagate Cheetah 9LP ST39102LW

  April 16, 1998 Author: Eugene Ra  
See also our Summer 1998 SCSI Drive Roundup
See also our Drive Cooler Roundup

Cheetah Since it was first announced well over a year ago, Seagate's 10k RPM Cheetah family of disks has been regarded as the highest performing units around. Indeed the Cheetah was quite swift (see 4.5GB Ultra SCSI drive roundup), outperforming all other drives available by a considerable margin. Since then, however, IBM has come up with its first 10k RPM offering, the Ultrastar 9ZX, a drive the Storage Review plans to take a look at in the near future. Seagate has not stood still either, announcing its second generation Cheetah family, the 9LP. As happens all too often in the drive industry, the Cheetah 9LP shipped 6 months after its initial announcement. I've been drooling ever since.

Seagate CheetahThe core spindle speed of the Cheetahs has not changed; these are still 10,000 RPM units. There are other important modifications, however. Seagate has increased data density, packing 1.5 gigs on each of the 9LP's six platters. This increase in platter capacity (from 1.1 gigs on the original Cheetah) allows the 9LP to sport a standard 1" high, 3.5" form-factor rather than the 1.6" height that its predecessor used. Seek time has been lowered to an astonishing 5.4 milliseconds, a full 30% lower than that of the previous line. The drive's buffer is a full meg, twice the size of its predecessor. In addition to providing these performance increases, Seagate also claims a 25% reduction in power consumption, translating into significantly less operating heat. The drives are warranted by an enterprise-standard five years. The model evaluated here is the ST39102LW, a 9.1 GB Ultra2 SCSI drive.

Upon opening Seagate's familiar brown-box packaging, I was greeted with a folded piece of paper congratulating me on my purchase of the "highest performing disc drive available in the market today!" The leaflet then went on describing how 10k RPM drives dissipate more heat due to increased motor power. I played it safe and installed it in a drive cooler. Seagate's typical pocket-size documentation, dry but thorough, was also enclosed.

Ultra2 (LVD) SCSI is most popularly cited for increasing the maximum transfer rate of the SCSI bus to 80 MB/sec. That's certainly nice, but there's no doubt that even this drive won't come close to taxing that bandwidth limit. Differential technology, however, allows for longer cable length- in LVD's case, a relatively luxurious 12 meters. The Storage Review Testbed's Adaptec AHA-2940U2W came with an LVD cable which, until recently, just sat in a closet unused. This cable show's off LVD's length-limit, stretching a good four feet or so with 5 68 pin connectors ready to go. For those of you who haven't seen one of these cables yet, take my word for it: Its -ugly-. Each of the 68 conductors run separated from each other, only coming together at each connection point. I suppose this is to allow one to snake the cable all around the inside of a case, making sharp bends without placing stress on it. A nice feature of LVD cables is the built in termination at the end so you don't have to worry about per-device termination settings. The terminator seemed rather flimsy though, with the plastic circuit-board protector easily popping off. It's a pain to pry off of a drive once it has been attached.

The 2940U2W features separate connectors for "legacy" Ultra and Ultra2 devices. Apparently if you have an Ultra SCSI device connected to the LVD bus the entire bandwidth limit is halved to 40 MB/sec. Adaptec provides these two separate connections along with a bridge chip as part of their "SpeedFlex" technology, electrically isolating the two buses to allow optimal performance. The testbed's Hawk 4XL's boot drive and UltraPlex CD-ROM were connected to the UltraSCSI connectors while the Cheetah 9LP was placed on the LVD segment.

I should note that the drive, while an Ultra2 SCSI unit, has two separate transceivers built into it allowing it to operate in either Ultra2 (LVD) or Ultra (Single-Ended, SE) modes. The drive comes configured to automatically detect and operate in the appropriate mode. Users can also shunt a jumper located on the bottom of the drive to force single-ended operation.

ZDBop's Winbench 98 along with Adaptec's Threadmark 2.0 were both run on the unit in Windows 95 OSR 2.1 and Windows NT Workstation 4.0. The drive was partitioned into a single volume of maximum size. The average of 5 trials are presented below.

 Testbed I Performance GraphsDetails... 
Windows 95, OSR 2.1 using FAT32
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 3830|
Hitachi Pegasus (9.1 GB UW-SCSI) - 3827|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 3797|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 3760|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 3600|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - Not Tested
Windows NT 4.0 using NTFS
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 4730|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 4427|
Hitachi Pegasus (9.1 GB UW-SCSI) - 4387|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 4333|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 4140|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - Not Tested
Windows 95, OSR 2.1 using FAT32
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 13567|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 13400|
Hitachi Pegasus (9.1 GB UW-SCSI) - 12167|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 11867|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 10100|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - Not Tested
Windows NT 4.0 using NTFS
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 15767|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 14067|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 12700|
Hitachi Pegasus (9.1 GB UW-SCSI) - 12400|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 12033|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - Not Tested
ThreadMark 2.0 Data Transfer Rate
Windows 95, OSR 2.1 using FAT32
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 14.50|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 11.14|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 10.75|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 10.71|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 10.41|
Hitachi Pegasus (9.1 GB UW-SCSI) - 8.91|
Windows NT 4.0 using NTFS
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - 20.35|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - 16.18|
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - 15.82|
Hitachi Pegasus (9.1 GB UW-SCSI) - 14.91|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 12.52|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 12.38|
Business Disk WinMark 98
Windows 95, OSR 2.1 using FAT32
Hitachi Pegasus (9.1 GB UW-SCSI) - 2470|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 2244|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 2126|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - Not Tested
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - Not Tested
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - Not Tested
Windows NT 4.0 using NTFS
Hitachi Pegasus (9.1 GB UW-SCSI) - 2840|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 2542|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 2522|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - Not Tested
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - Not Tested
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - Not Tested
High-End Disk WinMark 98
Windows 95, OSR 2.1 using FAT32
Hitachi Pegasus (9.1 GB UW-SCSI) - 6324|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 6194|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 5768|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - Not Tested
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - Not Tested
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - Not Tested
Windows NT 4.0 using NTFS
Hitachi Pegasus (9.1 GB UW-SCSI) - 6016|
IBM Ultrastar 18ZX (18.2 GB U2W-SCSI) - 5982|
Seagate Cheetah 9LP (9.1 GB U2W-SCSI) - 5520|
Fujitsu MAG3182LP (18.2 GB U2W-SCSI) - Not Tested
IBM Ultrastar 18LZX (18.2 GB U2W-SCSI) - Not Tested
Western Digital Enterprise 10k (18.3 GB U2W-SCSI) - Not Tested

Now that's performance! For the first time since we've started testing using two benchmarks on two operating systems, a single drive has posted record figures in all areas- a claim that even the previous Cheetah 4LP could not make. In ZD's Disk WinMarks, the Cheetah 9LP provides performance gains ranging from 7-11%. Lower seek time and a 30% increase in transfer rate were also apparent, along with a bit less CPU utilization per MB transferred.

Tests under Adaptec's ThreadMark are where the drive really shined, however. The drive was the first we've ever tested to break the 10 MB/sec barrier- and this is under Windows 95, let alone the higher-performing Windows NT. The drive scored 43-50% higher than its predecessor while again displaying slightly more efficient CPU utilization. Gains in performance should be exceptional under moderate-to-heavy multitasking loads.

The downside? As usual, heat and noise. I've read in various places how the drive is supposed to be substantially quieter and cooler than the first generation Cheetahs. Since I don't have a temperature-controlled environment where I can accurately take operating temperature readings, I have to use a subjective method (i.e., my hand on the drive) to judge heat. After operating extensively in a cooler, the drive was very warm. Touchable, but still hotter than any other drives we've tested excepting Seagate's own earlier Cheetah and the current Medalist Pro. This isn't a disk that I'd try to run without some drive fans, regardless of how well-ventilated the case is.

I can't say I've noticed any subjective lowering of noise with the new Cheetah 9LP. The piercing whine that I've learned to tolerate with the first offering is still present and noticeable. Seeks still churn away, no quieter than the original Cheetah. Noise and performance are still two virtually antithetical concepts.

What else can I say here in conclusion? The Cheetah 9LP ST39102LW is the fastest drive we've tested, and, as expected, among the noisiest, hottest and most expensive! I'm sure bleeding-edge power-users (including myself) will have no problem accepting the drawbacks and immediately plop this hotrod into their systems. For other uses, the same recommendation that accompanied the original Cheetah stands: Noise, heat and expense may very well be a concern. If they're not, this is the drive for you.

Seagate Cheetah 9LP ST39102LW
Estimated Price: $1199
Specifications
* Note: All reported test results are the average of five 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