- Uses a different technique to do the multiplies, saving one
  instruction and one temporary register on SSE and AVX.

ug: v8:6020
Change-Id: I4f3ff6186dae5eb10d90cda31c7d16b651a00d7e
Reviewed-on: https://chromium-review.googlesource.com/1132522Reviewed-by: Jing Bao <jing.bao@intel.com>
Commit-Queue: Bill Budge <bbudge@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54370}