retard: > 'High performance' and 'system language' are both badly defined.
I agree, for some people 'high performance' means the automatic slicing and tiling of loops as done by advanced Fortran compilers, something that C/C++ compiler have just started to do a bit, and are far from doing well still. High performance also means using the SIMD instructions efficiently, and not even the Intel C++ compiler (that about this is better than G++) is doing that well yet. That's why high-performance kernels in GNU radio, video decoders, Golomb ruler finders, etc often need to be handwritten. On the other hand there are tools that allow you to write that maximally efficient code using Python: http://www.corepy.org/ And don't forget that today high-performance sometimes means using highly parallel code on the GPU: http://pypi.python.org/pypi/pyopencl/0.90 > For example, is LLVM a good tool for high performance code? LLVM is getting better, but it will need several more years to go be there. For high-performance numerical code it's not yet as good as GCC (it doesn't even vectorize code), and on this kind of code GCC is less efficient than the Intel compiler. Bye, bearophile
