Model on the Cyber 205

which is governin! groups a Inherentanalysis, vectors paralleli on this vectorizal III. Seci . tlon exp~ and the q The sununa Section V A mesoscale meteorological numerical model is chosen for performing the CYBER 205 vectorization technique. Three computational modules of the model (vertical diffusion, horizontal advection, and a computation including surface con-rasts) are selected as working examples. A vectorization technique, denoted as "orthogonal vectorization," is developed which is based on the model's computational structure and the CYBER 205's computer-parallelism and vector processing. Experimental results showed that for the selected examples, a speed-up by a factor of ten of the model execution on the CYBER 205 has been achieved. Provided further experiments and developments are performed, such an increase in executional speed is expected to produce profound impacts on mesoscale modeling in particular, and on atmospheric modeling in general.