I’ve coded possibly world’s first fast progressive ray tracing engine on ARM.
ARM NEON Optimization is used for QBVH traverser and ray-triangle intersection.
NEON-optimized QBVH spatial data acceleration builder will be soon.
FYI, I’ve also integrated LLVM/clang to this raytracing engine for JIT-compiled shader, but LLVM ARM backend is not yet enough to do it(assertion happens after JIT codegen). Time will be solve the problem.