At this moment warp stabilizer seems to only use one core of the CPU to "analyze" the footage, and because of that takes an extremely long time. The usual workaround is to split the video into several pieces, up to the number of cores the CPU has, but then stitching them back seamlessly is a very difficult task. Why can't Premiere virtually split the video in an optimum number of pieces of equal duration, analyze them all at the same time allocating one core for each, and them aggregate the gathered data to stabilize the whole clip in one process?