For what it is worth, I just ran into the issue of what I thought an Agent job not running as scheduled. It turned out that the job never finished...

It would be very interesting to see the differences using the profiler. In a demo I saw recently the client statistics showed a performance gain but when they ran profiler...

You might want to look into using RCSI (read committed snapshot isolation) on the Sharepoint database. It should significantly reduce the number of blocking processes.