Significant improvements in real-time eficiency have been obtained for plasntajluid turbulence calculatwns by m"crotasking the nonlinear jlu"d co& KITE in which they are implemented on the CRAY Y-MP C90 at the National Energy Research Supercomputer Center (NERSC).The number of processors accessed concurrently scales linearly with problem size.Close to six concurrent processors have so far been obtained with a three-dimensional nonlinear production calculation at the currently allowed memory size of 80 Mword.With a calculation size corresponding to the maximum allowed memory of 200 Mword in the next system configuration, we expect to be able to access close to ten processors of the C90 concurrently with a commensurate improvement in realtime eficiency.These improvements in pe~ormance are comparable to those expected from a massively parallel implementation of the same calculations on the Intel Paragon.