Maybe you can look back into openCL??? Please don't kill me Andreas!! lol
Maybe you can check out the code for fractorium with GPU??
I think the fractorium code would not help, but this is not the problem. In fact, I have a lot of own ideas of how to implement stuff in openCl, and I'm even interested in the topic. But, this would take a lot of effort, and I do not have enough spare time. If some more developer(s) would join, it would be possible, but I will not do it alone (and not sure what CozyG thinks about this) - and I do not see major importance.
Faster rendering as nice feature in first place for sure, but it will come to a swarm of new problems. So many people are not even able to maintain/install a java-runtime. When it comes to OpenCl we will open Pandora's box of driver problems. And when the drivers finally works (for a few weeks, until the next update), people will complain that they can only render small images because their gfx memory is too small. Ok, then we come to stripe rendering, which will lead to other problems etc. etc.
Recently, I did not have had much contact to Matt (the author of fractorium), but what I saw some months ago, he put a LOT effort into this stuff (a "men-year"? don't know). And it is "only" a port of the classic flam3-stuff and has no innovation to the flame-algorithm itself (from what I saw the last time when I saw fractorium, please forgive if there is now more).
So, maybe we could start some Kickstarter campaign for it - but without some substantial extent of resources (people, money) there can not be an openCl-version from my side.
Thanks for your understanding
My experience with OpenCL based rendering for fractal/IFS software has not been great. I've tried several different programs, on both OSX and Windows. With my hardware setups many of them crash, either before a render completes or immediately after a render finishes. I suspect driver issues, as Andreas mentions above. I've had a little more success with CUDA based rendering software, but every system I've tried has still been problematic in some way.
So yes, it would be great to have much faster rendering in JWildfire via a GPU-based pipeline. IF it worked on most people's machines. But the evidence I've seen suggests that compatibility across graphics cards and graphics drivers is too much of an issue for this to be currently feasible.
I do like the idea of a JWildfire Kickstarter campaign though!
Even though there's already a way to donate to Andreas for JWildfire development
, I think a Kickstarter (or other online crowdfunding) campaign would have a much higher profile. I just don't think it should be tied to OpenCL rendering as a specific deliverable. There's plenty of other features to add. Maybe the goal could be JWildfire 3.0?