In order to keep our country at the forefront of this area of activity, we need to provide the educational opportunities that will sustain a viable group with suitable technical skills. This will require a major investment in new curricula development that emphasizes hands-on approaches to teaching parallel computation. We need to develop and disseminate the theoretical basis of parallel computation in conjunction with practical experience, and to permeate the hands-on experience with the development of intuition into parallel computing.
There is a sustained trend on developing means and tools for education on parallel computing [18]. However, when carefully examined we may observe that while many courses we teach bear the attributes ``parallel computing" and ``high performance" in their titles, very few of them do really provide hands-on experience with HPCS. This is probably due to the fact that the existent equipment in computing laboratories is mainly based on sequential systems. On the other hand the new parallel machines are still expensive and do not generate enough profit and thus cannot be routinely available at educational institutions. It is true that NSF supercomputing centers are organized in the academic environments and offer computation time. But it is also true that they are inconvenient for small teaching jobs, are not provided with educational tools that would facilitate hands-on experience and development of intuition of parallelism and therefore while they may provide for advanced research they serve educational purposes poorly. Consequently, the critical mass of programmers for parallel machines required in order to sustain the endeavor of producing parallel machines cannot be created and the endeavor itself is threatened.