You may remember my attempt last year to port flam3 to CUDA to enable much faster GPU-accelerated rendering. You may also remember how I dropped off the face of the planet just when the results were starting to get interesting.
That's because the results turned out to be wrong. I wanted to fix them and post an explanation as to how I measured them incorrectly before recanting the figures, but life caught up with me and I never got around to it.
I'm doing a lot of work with image processing right now, and I recently hit on an idea which may make it possible to come close to the rich, salient theoretical performance figures that made me excited about the idea in the first place. I don't want to jump the gun again, so this will be the last you'll hear on the matter at this location until it's ready for testing (which may be some months away).
Those interested in reading a little more about the work I did last year and the plans to fix it can examine my spectacularly ugly blog, where I may also post updates about this project from time to time.