Moving onto a RAID 0 configuration in Windows NT. You can see that each card improved its benchmark scores by an average of 2-4%. NTFS is much more proficient at data management than FAT32 and the Windows NT kernel is faster and more stable - both these factors contribute to better data flow from the OS to the mini-port driver and that results in better scores! If you are running Windows NT as your NOS, then you will gain a slight performance increase over Windows 98 with a striped set of drives, regardless of which IDE RAID card you use. This should appeal to not only power users, but to Network administrators as well.
RAID 1 scores in Windows NT showed almost the exact same trend as the Windows 98 scores when comparing RAID 0 to RAID 1. The only thing I found slightly curious here was how these marks compared in the Sandra results – the scores actually became higher when moving from RAID 0 to RAID 1, which is totally against what we expected. Since the Sandra marks are purely synthetic, we can chalk this off and look at the Zdnet high-end disk marks to see what’s really going on, since the scores are more true to life. The high-end disk marks show only a 2-4% performance loss across the table for all cards in a RAID 0 configuration, which is a consistent and expected result.
In all the scores for Windows 98 and Windows NT in both RAID 0, and RAID 1 configurations, it is quite clear that the Promise FastTrack 66 card was the overall performance winner. But coming in just a hair under the top scores was the Iwill SIDE-RAID66 card - 2-3% is not really a very large performance gap. Lastly, the AMI was somewhat of a disappointment and gave us lower scores than what we’re used to seeing with AMI products.
forums
send article 
hardware newsletter 
article comments
Digg this!
del.icio.us