Our tests were conducted using a Pentium III based system, using Intel’s 815E core logic with the 82801BA ICH2. Intel’s ICH2 is, in our opinion, one of the best implementations of an IDE controller currently available, and will therefore be used as a reference.
The D740X was tested at ATA-133 speeds using Maxtor’s ATA-133 controller, and at ATA-100 speeds using the on-board IDE channels. On some graphs, you may notice an additional extrapolated figure. In order to determine whether the
controller card itself was impacting performance (and
not the ATA-133 bus), we tested our older DiamondMax Plus 60 at ATA-100 speeds on both the on-board IDE channels, and using Maxtor’s controller card. This will allow us to measure the impact the controller card has on performance, and allow us to estimate what sort of performance we could expect from a on-board IDE controller that offers native support for ATA-133. Sometimes the add-on controller card improves performance, and sometimes it degrades it, as opposed to an on-board IDE controller, so sometimes this extrapolation will show higher performance than the D740X using the on-board controller, sometimes slower.**
Our IOMeter tests were conducted using an access pattern designed to simulate typical workstation use. The specifics of the Workstation index are:
-8kb Blocks
-Read/Write Percentage: 85/15
-Random/Sequential Percentage: 80/20
-Light Load Simulation: 8 outstanding I/Os
-Heavy Load Simulation: 96 outstanding I/Os
-Ramp-Up time was set to 60 seconds, and all tests were run for 10 minutes on unpartitioned disks.
This index (mostly reads, and mostly random) represents the most typical usage patterns. The light load simulation best represents light home use (such as loading Internet Explorer or MS Word), while the heavy load index best illustrates more demanding load situations (like loading Windows, or loading multiple programs simultaneously).
All WinBench99 tests were conducted under NTFS. HDTach Burst Rate tests were conducted under NTFS. Due to some extraneous circumstances, we were unable to use SYSMark2001 for our tests, and had to settle for SYSMark2000 instead. SYSMark2000 tests were conducted in Windows 98 using FAT32. Resolution was set to 1024x768x16, 85Hz.
**We realize the ideal way to test this would have been to set the controller to ATA-100 with the D740X, and not the older DM Plus 60. This was not an option, however, as the card did not have an option to reduce to ATA-100, and defaulted to ATA-133 automatically when paired with the D740X.
