Optimize products for small objects
Assigned to Nobody
Link to original bugzilla bug (#404)
Both gemv and gemm like operations could be significantly improved for small objects. First thing to do is implement an exhaustive performance test comparing various implementations and try it on different compiler/architecture.
see also "Eigen 2 to Eigen 3 performance regressions with mapped matrices" and "Blas performance on mapped matrices" threads in the ML