Performance differences are probably negligible. Unlikely to be anything you'd actually notice. The nforce controller has an advantage in throughput because it is not running on the PCI bus, so burst speeds are limited only by the SATA speed, rather than PCI speed, but that is unlikely to actually be anything you could see in usage. Neither one is going to be hitting the limits even in a RAID configuration during sustained transfers.
CPU usage is likely to be a bigger issue, but is still pretty much a non-issue with today's processors. The Tech Report review of the nforce3 showed the CPU usage at 11.4% with an Athlon64 3200+, while the OCW review of the 3112 had 8.3% on an AthlonXP 2700+. So the nforce3 used a higher percentage of a processor that was already running much faster. If you base it on the performance ratings of the processors, the nforce3 used 364.8 "points" while the SI used only 224.1. The SI would be only 7% on the faster processor. Of course this isn't the exact same controller on your board, so that might have changed between models. In either case, a few percentage points is unlikely to seriously affect your experience. Who knows what RAID will do to the CPU usage as well.
The nforce3 also has an advantage in that you don't need any extra drivers installed, if you didn't want to bother with the nvidia drivers and just use the standard Microsoft drivers, because it's just functioning as an integrated IDE controller as far as the OS is concerned.
Of course, you could easily do a performance test yourself. Just install the new drives and test them out, without removing your current configuration.