The Barracuda ATA ST328040A (a close cousin of the Barracuda SCSI line of drives) that was tested came in Seagate's SeaShell which is a plastic storage container that protects the drive during transport. Though it isn't really the most sturdy piece of plastic made by man, it sports of enough strength to get the drive transported safely from point A to point B.

On the bottom of the drive is their SeaShield which essentially is a protective film which covers all the circuitry embedded on the PCB part of the drive. Printed on the SeaShield are a number of instructions on setting up the drive with Seagate's Disc Wizard utility. These instructions prove quite useful for those who are new to the game of installing a hard disk. In the footsteps of Quantum, this drive sports a resistance of up to 300Gs of shock. As a marketing tactic, they are not exactly the first ones to boast of shock resistance as Quantum was probably the first with their SPS or Shock Protection System.

Before moving on to benchmark results, here is a summary of the system used to perform the tests on the ST328040A drive...

Test Bed Setup

- Celeron 400 CPU running at rated clock speed
- Azza PT-810DMC Motherboard i810 chipset
- Integrated ATA/66 controller
- 40-pin, 80 conductor ATA/66 compliant cable
- 96MB PC100 SDRAM (CAS 2)

- Ziff Davis Winbench 99 version 1.1 under Windows NT 4.0 / Windows 98
- Adaptec's Threadmark 2.0 under Windows NT 4.0 / Windows 98
- SiSoft Sandra 2000 under Windows NT 4.0 / Windows 98

Ziff Davis Winbench 99 was used to test the disk subsystem performance of the Seagate Barracuda ATA ST328040A. Under this, 2 tests were used that belong to this suite. The Business Disk Winmark and the High-end Disk Winmark. To begin with, the Business Disk Winmark test hard drive subsystem performance under common business applications such as wordprocessing, spreadsheet, databases, etc.

The High-end Disk Winmark benchmarks a drive through a given series of tests under applications such as the likes of Microsoft's Frontpage, Adobe Premiere and Visual C++ among others. Let us get on to the numbers. Adaptec's Threadmark 2.0 measures multithreaded disk I/O performance under Windows NT and Windows 98. Threadmark also computes the CPU overhead required to perform these disk data transfers. The significance of CPU overhead indicates how much the CPU is involved in the process of transferring data. A lower number is better since the CPU has more time to handle other processes while the data transfers are occuring. Before testing, the hard disk was divided into 3 partitions.

One partition, FAT-16, was restricted to a size of 2GB due to the limited amount of space available under FAT-16. The second was a 12GB partition of FAT-32 and the third was of course a 12GB partition of NTFS. Fresh installations of Windows NT 4.0 and Windows 98 were used in a dual-boot configuration with a minimal amount of drivers and applications installed.