Accelerating tetrahedral interpolation with data-level and Thread-Level Parallel optimization