So I went back and tested the PERC 5/i in CrystalDiskMark using the 100MB test size (I had been using the 1GB size) and the results were vastly different.
RAID5
Sequential Read : 753.755 MB/s
Sequential Write : 714.859 MB/s
Random Read 512KB : 718.944 MB/s
Random Write 512KB : 654.487 MB/s
Random Read 4KB (QD=1) : 74.223 MB/s [ 18120.7 IOPS]
Random Write 4KB (QD=1) : 64.773 MB/s [ 15813.6 IOPS]
Random Read 4KB (QD=32) : 145.747 MB/s [ 35582.7 IOPS]
Random Write 4KB (QD=32) : 139.644 MB/s [ 34092.8 IOPS]
RAID0
Sequential Read : 784.340 MB/s
Sequential Write : 732.173 MB/s
Random Read 512KB : 706.846 MB/s
Random Write 512KB : 680.394 MB/s
Random Read 4KB (QD=1) : 56.651 MB/s [ 13830.8 IOPS]
Random Write 4KB (QD=1) : 56.779 MB/s [ 13862.0 IOPS]
Random Read 4KB (QD=32) : 149.119 MB/s [ 36406.0 IOPS]
Random Write 4KB (QD=32) : 137.637 MB/s [ 33602.7 IOPS]
Hmm...I thought to myself - that seems like phenomenal performance, but what does it mean for me? I tried the only thing I really could do with what i had on hand - I dropped a few larger files (1-4GB in size) in a folder on the desktop and copied them. I got 200MB/s which isn't horrible, but it's not much faster than a single drive doing the same thing and it's certainly no 700MB/s achievement either. I can say that staying with files smaller than the cache certainly makes a big difference, but I've gotten much better copy results when messing around with the other RAID card (that I will try to test with this weekend if at all possible).
In other news, I thought I would post a few pictures of the server that we've got running
You can see my slight modification to get the SSDs mounted in the factory sleds - works pretty well if you ask me.