One current trend in software development for PDEs, and here especially for
FE approaches, clearly goes towards very sophisticated
hierarchical techniques and adaptive methods in any sense. In contrast,
the employed data and solver structures
are mostly chosen as `globally defined` types which neglect the very specific performance
facilities of modern hardware platforms. As a result, the observed computational efficiency
is often far from the expected peak rates of (potentially available) several GFLOP/s per processor.
These discrepancies, between numerics and software concepts and the available hardware,
often lead to unreasonable calculation times for `real world` problems as can be easily seen from recent
benchmark comparisons for commercial as well as research codes.
Hence, strategies for massive efficiency enhancement are necessary, not only from the mathematical
(algorithms, discretisations) but also from the software side of view.
To realise some of these aims our FEM package FEAST (`Finite Element Analysis
/& Solution Tools`) is under development. Recent results on JUMP,
including applications from CFD and CSM, are given.