default_pipe optimizer slowness during concurrency

Hi, I have a scenario where if I run the a SQL with 10 user concurrency in a recursive mode, the execution time is significantly slower. The SQL which completes in .8sec will end up in completing 5-10hrs during concurrency. This behavior is seen when the optimizer is set to "default_pipe" , however when set to "sequential_pipe" the timing has been drastically reduced and it end ups in seconds. How is "sequential_pipe" optimizer works here, is it to do anything with concurrency. NOTE - this happens when there is DISTINCT pattern in SQL. Regards, Sreejith