AL_INVERSE_DISTANCE_CLAMPED versus AL_INVERSE_DISTANCE for deterministic audio culling

When authoring 3D audio with OpenAL 1.1, developers must select a distance attenuation model that determines how source gain diminishes with listener distance. A common design constraint requires predictable audio culling: sources beyond a gameplay-defined radius must be muted or processed inactive without audible artifacts.

Two approaches compete: AL_INVERSE_DISTANCE_CLAMPED with an explicitly set AL_MAX_DISTANCE produces a hard gain cutoff at the specified radius, ensuring deterministic culling but potentially introducing an audible discontinuity if the cutoff radius does not align with the source's perceptual threshold.

Alternatively, AL_INVERSE_DISTANCE (unclamped) paired with a large finite AL_MAX_DISTANCE allows gain to asymptotically approach inaudibility, avoiding a sharp cutoff but risking floating-point precision loss and unpredictable effective distance at extreme values, especially where default maxDistance is FLT_MAX. The choice prioritizes a clean, testable culling boundary versus a mathematically smooth falloff dependent on runtime distance calculations.

Does the clamping radius align with the application's culling logic, and how does the choice interact with per-source rolloffFactor? Can developers predict effective gain at a given distance without reimplementing the model, and what precision safeguards are needed for unclamped large maxDistance? Is the hard cutoff perceptually acceptable, or does it introduce audible artifacts that outweigh culling simplicity?