The cambrian explosion of mixed-precision matrix multiplication for quantized deep learning inference | Publicación