Memory-efficient optimization of Gyrokinetic particle-to-grid interpolation for multicore processors