Automatic recognition of computational kernels for platform-dependent code optimizations

We present a novel approach that assists the task of porting code to an embedded platform. Our tool automatically identifies code segments in the input program that can be replaced with optimized kernels from a platform-dependent library. Using a C-function as a model that describes the computational kernel, the tool identifies equivalent code regardless of syntactic and computational variations. For a case study using the Texas Instruments C66x DSP library, our approach identified code replacement opportunities that resulted in runtime performance speedups of up to 2.2x.