Home

ogis-modern-cxx-future-cpp:

This GITHUB-REPO is a testbed for a special kind of Modern C++ CORE LANGUAGE-extension.

NEW loop() - COMPOUND-statements s related to ITERATION

• loop(N, ...){}
• typed_loop(type, N, ...)
• named_loop_up(id, N, ...){}
• named_loop_down(id, N, ..){}

With optional opt-opration expressions. N should be of integral type. These new COMPOUNDs can be easily imlemented via the cpp-preprocessor.

Example: matrix_copy with stride

more examples: https://github.com/F-Haferkorn/ogis-modern-cxx-future-cpp/tree/master/future-cpp-loop/ogis-cpp-loop.examples/examples

template<typename TPtr, typename TRowSize, typename TColSize, typename TStrideSize >
TPtr matrix_copy( TPtr tgt, TPtr src, TRowSize nRows, TColSize nColumns, TStrideSize stride)
{
  // compiler can  mangled the loop-internal integral-types tgt and src in registers.
      loop(nRows,  tgt+=stride, src+=stride)  // apply strid eafter each row to tgt and src
            loop(nbofColumms, tgt++, src++)
                *tgt = *src;
      return tgt;
}

Advantages With recent compilers this gives no really speed advantage, but is no way slower than a full for(;;){} iteration.

And it increases readability.

Future C/C++ compilers can take advantage the reduced degree of freedom of the iteration and make e.g. heavily use of ALU-register operation of DSP-Architectures