Traditional programming languages for parallel computer systems do not efficiently separate the algorithm description from the details of its hardware implementation. As a result, a significant code modification is required to port the same algorithm from one computational architecture to another. To simplify and speed up the porting procedure, an architecture-independent Set@l programming language based on the aspect-oriented programming paradigm and set-theoretical code view is proposed to be used. In contrast to conventional parallel programming tools, Set@l defines the information structure of a problem as sets, subsets of different types and relations between them. Various aspects of implementation such as processing method, parallelization, optimization and others transform an architecture-independent source code according to the certain architecture and configuration of the computer system. Set@l offers new opportunities for efficient architecture- and resource-independent parallel programming. It ensures porting of parallel applications without the source code modification, transitions between different algorithm implementations with regard to the computer system’s features or user’s preferences, and the indefinite description of calculations. This paper deals with the essential techniques and specificities of algorithm parallelization in the Set@l aspect-oriented programs. The demonstrated examples are adopted from programs in the Set@l language. The lower-upper decomposition code demonstrates the conversion of set types by parallelism during the architectural adaptation, and the Jacobi algorithm code introduces indefinite collections (classes and semisets) for the description of effective computing structure for reconfigurable computer systems. In addition, we consider several instances of memory use for the organization of parallel calculations.