Which compiler optimization switch does specifically address optimizations for the new Intel(R) Atom(TM) Processor?
The architecture specific optimization switch is -xSSE3_ATOM. The new Intel(R) Atom(TM) processors are in-order machines. The IntelĀ® C++ Compiler models the Intel(R) Atom(TM) processor pipeline and execution flow, thus enabling it to produce code with the optimum instruction execution sequence for low-power IA. Additionally the Intel(R) C++ Compiler for Linux* has a whole set of Intel(R) Atom(TM) Processor specific optimizations that further help with getting the best performance out of your application running on an Intel(R) Atom(TM) Processor based platform. All the user needs to do to enable the in-order instruction pipeline modeling is to use the option switch -xSSE3_ATOM along with all the other compiler options they may be using to optimize their code. This allows for reordering the generated machine instructions and thus minimize pipeline stalls. -xSSE3_ATOM also implicitely enables generation of movbe instructions.