Nesting conditions

The Radial code for the decision tree of Figure 2 is shown in Figure 3. This code was generated by RuleMaker. Experts have difficulty correctly generating a deeply nested conditional phrase like this, but they are able to inspect it for possible errors or omissions. 

FROM a most nested conditional TO a top level conditional DO... 

Compared with the above systems, the most important feature of the Cyber system is to achieve to exploit both potential concurrency and alternation at the same time in a simple and systematical way, by considering data dependencies and control structure in behavioral description. Therefore, it can synthesize more efficient control and more completely resource sharing from description containing nested conditional branchings and loops while maintaining the same ability to exploit potential concurrency with other system, and without incresise of computational complexity. Moreover, all possible paths for conditional branches are scheduled optimally. 

By applying the Condition Vector concept into a conventional scheduling algorithm which are mainly for exploiting potential parallelism, the algorithm becomes to be able to exploit both alternation and parallelism concurrently and to find out different control sequence for each path. The scheduler has only to assign CVs of nodes instead of the nodes themselves. This is because the CVs contain imformation regarding control structure (nested conditional branches) in behavioral description. 

However, each branch of conditional branches should have its individual sequence suitable for itself. This could be done by scheduling each path separatedly. Nevertheless, such approaches are weak in exploiting global potential parallelism. In addition, they cannot search for the possibilities that branches are executed before the conditionals are resolved. The Cyber system can generate such efficient control sequences in simple method for CDFGs containing nested conditional branches. 

Actually, there is not only a nesting condition in k space, but also in the dispersion relation, see Giamarchi, T. 

---