State of the Art: Simulation methods for fluid dynamics (CFD) have been established in the industrial use for years. Typically they are based on pure MPI-parallelization and do not deploy SIMD-capabilities of current processor architectures. Goal of the project was to adjust two selected CFD-codes for future HPC-Many-Core-architectures. The selected method was to introduce additional levels of parallelism by a hybrid shared-memory/MPI parallelization on the one hand and on the other hand by adjusting the programs for the deployment of SIMD-registers by automatic code-transformation using a preprocessor to be developed. For analyzing the SIMD usage of the codes, extensions to the Vampir suite were developed. As a result of the project these capabilities are now available in the CFD-codes considered and the developed techniques are published. Outcome of the project is a more efficient use of future computer hardware for the CFD methods considered as well as the availability of a code-transformation and an analyzer for SIMD-
optimization of other software.