Hybrid computing: a user's view

Abstract

The analog computer section of the McDonnell Automation Center has been involved in hybrid com puting for some time now, leading us to think we can contribute to the dialogue now proceeding on the subject. In this article we first detail the only rational criteria for selecting a computing method, namely cost and capability, and then present the classes of problems that are prime candidates for hybrid solu tion. A brief description of our present facility is included, along with our views on selecting hybrid programmers. Looking ahead, we have outlined requirements for a future hybrid system, placing a good deal of em phasis on software needs, especially a simple, parallel language based on MIDAS, DYSAC, etc. An exam ination of hardware costs and the characteristics of the new breed of digital computers suggests that, although hybrid computation can be done more efficiently with the new generation, there is a pos sibility that neither hybrid nor analog computation in their present forms can survive the onslaught of low-cost digital computing. The outlook is not as bleak as first appears, since we believe analog con cepts can be retained in future systems, but only if SIMULATION readers and other interested parties are willing to make an issue of it.