Again, the ICH10 does not provide performance advantages over ICH9 and older parts: 301.6 MB/s in RAID 0 with four hard drives is less than the 311.5 MB/s we measured on ICH9 with exactly the same hard drives—although the decrease is small, it is still there. However, Intel’s core logic remains the fastest choice for enthusiasts who want to work with RAID arrays.
The same results apply to RAID 0+1: ICH9 remains the fastest Intel part when it comes to SATA RAID throughput. Unfortunately, Intel still does not use both mirrors of a RAID 1 (or in this case RAID 0+1) to speed up read performance. AMD and Nvidia do this, and achieve much improved read transfer rates over a single hard drive in RAID 0+1.
Although the maximum throughput in RAID 5 did not improve with ICH10, the throughput on a degraded RAID increased from 51.8 MB/s with ICH9 to 57.2 MB/s with ICH10. However, write performance is still ridiculously slow. It is even slow enough to prevent any efficient work on a system running a degraded RAID 5 array.