but processes multiple values at once. Example: /// Define convenience types to manage vector batches. /// `_impl` is a template parameter that is passed via `xsimd::current_arch` /// by the per-arch cast(data_i >> 24U)); In Krita we have a set of predefined convenience types for vector batches in KoStreamedMath: batch type element type num elements (AVX2) num elements (AVX) num elements (SSE2) convert int_v into float_v back and forth. Arithmetic operations Arithmetic operations with SIMD batches look exactly the same as if you did them with normal int or float values. Let's consider example