S150 TX4 Benchmarks
As you can see, in FCtest there is no speed difference between read1 and read2. As for the big files (ISO test), all RAID setups except for the 2xmirror read equally fast - about 125MB/s which is close to the PCI maximum (133MB/s). Creating the big files was faster on the striped arrays than on the mirrored arrays, which is pretty normal.
As for the smaller files (win/prog), differences between the setups start to occur: the less drives that are used in a striped set, the faster the files will be read. But, the more drives that are used in a striped set, the faster the files will be created.
We now can clearly see a difference between read1 & read2. Read2 was about 20MB/s faster in 2xmirror and 2xstriped. My guess is that this difference was caused by either Hard-Disk cache or memory-cache (by the OS).
It is also clear that these setups like big files: the bigger the files get, the higher the transfers become. I think it's also interesting to put some attention to the copy-test: with big files the 2xstriped setup gets about 30MB/s, which is to my opinion pretty ok. But if we then take a look at the WIN-copy test (lost of small files), none of the setups get more then 8MB/s.
Of course good old Sandra couldn't be missed in this review. Test results are pretty clear here: the more disks you use, the higher the transfer rate gets, with a maximum of about 105MB/s with a 4disk striped array. It gives a pretty good impression of how the different RAID configurations theoretically should perform.