xmath-usage

To eliminate per-frame allocations in Defold, adopt xmath for zero-allocation in-place operations. It writes results into the first argument using pre-allocated scratch buffers, preventing transient vector and quaternion objects from triggering garbage collection pauses.

Defold games stutter during physics loops because standard math operations allocate transient objects, triggering garbage collection pauses. Using xmath for in-place mutation eliminates these per-frame allocations by reusing pre-allocated scratch buffers for vector math.

To use pre-allocated scratch buffers for vector3 and quaternion math, initialize your buffer objects with vmath, then apply xmath operations to mutate them in place. The first argument serves as the output target, allowing buffer reuse across frames.

Yes, xmath works with vmath by using it to initialize scratch objects at module scope or during initialization. You then apply xmath operations to perform in-place mutations on those pre-allocated objects within your performance-critical loops.

The first-argument output pattern in zero-allocation math writes the computation result directly into the first argument passed to the function. This enables in-place mutation of existing pre-allocated objects rather than returning new instances that allocate memory.

You should use in-place mutation for Defold math operations in performance-critical hot paths like physics nudges, animation tweaks, and gameplay logic. This zero-allocation approach prevents garbage collection spikes during intensive per-frame calculations.