Using aligned memory in sound backends. The conversion from float to integer samples are vectorized, but the final packing into 16 bits is still done using scalar instructions, because simd.h does not have signed 16-bit integers implemented.
Sound engines using the old backend can either add ".getUnsafe()" on the safe pointer, or start vectorizing the sound engine using the added safety.
Using aligned memory in sound backends. The conversion from float to integer samples are vectorized, but the final packing into 16 bits is still done using scalar instructions, because simd.h does not have signed 16-bit integers implemented.
Sound engines using the old backend can either add ".getUnsafe()" on the safe pointer, or start vectorizing the sound engine using the added safety.