macro ukernel_generator(simd: static CPUFeatureX86; typ: untyped;
                        vectype: untyped; nb_scalars: static int;
                        simd_setZero: untyped; simd_broadcast_value: untyped;
                        simd_load_aligned: untyped;
                        simd_load_unaligned: untyped;
                        simd_store_unaligned: untyped; simd_mul: untyped;
                        simd_add: untyped; simd_fma: untyped): untyped

macro ukernel_simd_impl(ukernel: static MicroKernel; V: untyped; A, B: untyped;
                        kc: int; simd_setZero, simd_load_aligned,
                                 simd_broadcast_value, simd_fma: untyped): untyped

template x86only(): untyped