I have never heard of anybody doing this, so if you succeed please share your results.
Let me note that SIESTA is currently built for MPI parallel execution with large memory blocks, hence I would not suspect any real benefit with the XeonPhi coprocessing as memory could(?) be the bottleneck for large calculations. SIESTA currently lacks threading which suggests that XeonPhi should be really sub-optimal. I would be really surprised if you could get (any) performance gains on many cores on the XeonPhi coprocessing unit. I could be wrong though... 2015-10-22 14:00 GMT+02:00 Süle Péter <[email protected]>: > > The recent version of SIESTA is suitable for compiling > under Intel Xeon Phi coprocessing environment ? > If yes, what are the suggested options for installing > SIESTA ? > > Best regard, Peter Sule > -- Kind regards Nick
