I notice that particle simulating is done on the GPU and this has always made me wonder. TFD got GPU capability years ago. It is in my mind a game winner. Why have all the other fluid sims not been able to adopt this ? I just find it amazing that FumeFX has been lagging so far behind. I would imagine it is not arcane knowledge and is available to all.
I can only speculate. GPU programming is somewhat more complex and you typically have to build a second version of all of your compute components (one for the CPU, one for the GPU).
Also, people don’t like to get locked into a vendor specific driver stack like CUDA or Metal - rightfully so. Yet OpenCL has still not reached a state where it makes dealing with the vendor specific stacks unnecessary (and might never do). Thus, you may have to build even more than two versions of all compute components.
One of the key components of Reactions is a custom compiler infrastructure that avoids building these separate versions. Basically, once a compute component is built using Rake (the Reactions programming language), it will run on any compute device and driver stack. Why isn’t everybody doing this? Possibly for the same reason you haven’t seen a new TFD version in several years: It’s a buttload of work
I do hope you have an Acolyte in case you get the urge to retire