It may be relevant to your trouble shooting to note that no HD can saturate a SATA 1 interface so worrying about whether your drives are running in SATA 1 or SATA II mode is irrelevant. Look at the STR reported in HDTach compared to their specs. If they are performing to spec, great... if not, then you have a problem, but it's likely not the SATA link speed since neither standard can bottleneck any HD.
Another thing to note is that certain stripe sizes will improve some benchmarks and hurt others. IIRC HD Tach uses 32KB data chunks so a stripe that splits this evenly across the number of drives in your array (i.e. 16KB for 2 drives or 8KB for 4) will make it appear to perform the best. That's not to say your real world performance will necessarily benefit. A good rule of thumb is to use a stripe size that is 64KB divided by the drives in your array (32KB for 2 drives and 16KB for 4 drives).
My experience with RAID, even with using a top quality ARECA card is that at best you can expect to turn out some flashy benchmarks, but even a 4 drive RAID 0 array on an ARECA has very little real-world benefit in most windows operations since the disk system is often not the bottleneck (which may be a surprise to many). Loading Windows for example is not just about loading data from disk into memory but actually executing the code - a faster processor and slow disk is better than a slow processor and fast disk. Same with loading a game level or transcoding a video. After having spent a rediculous amount of money on RAID, I still run it, but with little real-world benefit. I think having multiple single disks would actually be more beneficial in many cases than a RAID0 array. As usual... YMMV.
Bookmarks