Here are all the new additions and updates!

Core

Advection max trace steps has been exposed via a parameter in the advection settings tab. The default is a maximum of 16 steps, as it has always been hard coded to, but you can now lower or increase it to any value you see fit.
New project non-divergent algorithm options to help mitigate some artifacts that occur in specific scenarios. Common example is situations where you have velocities moving outwards in an omnidirectional pattern.
The solver now outputs VDBs on subframes. It will “floor” to the nearest subframe. i.e. if you have have 2 substeps and go to frame 1.6 it will output the volumes for frame 1.5, on frame 2.1 it will output volumes for 2.0, etc.
Influence color sourcing.
Control fields were added that allow you to source an arbitrary field to drive a control ramp on forces.
Fixed a bug where source shapes in the pump mode would clear color grid values. It now leaves all grid values untouched.
CPU performance optimizations. This should give users with very high end GPUs (3090 / 3080 / 4090 / 4080) a nice bump in performance.
Writing out VDBs is now 20% faster overall.
Solver is ~10-20% faster overall depending on GPU and CPU configuration.

API & OS Support

Updated to Cuda 12.2 form 11.5.
Axiom for Linux is now built on Rocky 8 using GCC 11.2
Metal command queue usage has been re-worked to fix a potential bug from occurring.

Fixed a bug with the fast noise type that broke when simulations were more that 1000 units away from the origin in the negative axis.
Fixed a bug with all 4D noises so they work with negative frames.