Winbench — continued
Finally, it's time for some real-world testing, starting with Winbench's Business tests, which run thought a series of scripts for typical 'Office' software.
This is certainly interesting, for a couple of reasons. First, it looks like FAT32 is quite a bit faster than NTFS. However, while performance scales up as it should with NTFS, FAT32 is actually slower under RAID 0. The slower RAID 0 results for FAT32 are puzzling, but Winbench 99 is rather old, so we'll have to see if other more recent tests exhibit the same kind of results. FAT32 is, however, most definitely faster than NTFS.
Business applications are well and good, but they don't really stress your storage all that much, at least not as much as high-end applications potentially can, and to measure that we have Winbench's high-end suite.
Here things start to make more sense. FAT32 retains its respectable lead over NTFS, which is considerably slower for single drive ATA100 and RAID 1 configurations. It would seem that NTFS has more to gain from RAID 0 configurations as it sees a much bigger jump in performance from ATA100 and RAID 1. Relatively speaking, the performance gap between NTFS and FAT32 is smaller for the high-end tests than it is for the business benchmark.
One of the nicer things about Winbench 99 is that it breaks down its high-end score into individual applications so we can take a look at which programs are seeing the most, or least benefit. First up is AVS/Express.
Odd. There it is againFAT32 sees a decrease in speed going from RAID 1 to RAID 0. Otherwise, peformance seems to scale reasonably, with NTFS again seeing a large gain from RAID 0. Interestingly enough, FAT32 is able to reap some benefits in performance from RAID 1, despite the fact that the purpose of RAID 1 is data redundancy. The likely explanation is that the controller reads data from both drives simultaneously, increasing read scores and thus the overall RAID 1 score.