SDF (Signed Distance Field) on GPU - 8SSEDT @ 60fps

I'm working on speeding up creating the SDF for my real time 2D ray tracer and my initial brute force method ran at 2 FPS (check every pixel against every other pixel). Moving to 8SSEDT algorithm on a single GPU thread gave me 4 fps. I figured out a way to run 8SSEDT in parallel however there is a trade off where each division cost a doubling of total computation. My Titan X (pascal) just hits 60 FPS with 256 threads and 64 total iterations of the algorithm over each pixel. I still need something faster.