1 comments
Great project!<p>Are vertex insertion and deletion also supported/accelerated?<p>What compromises are keeping this constrained to 32-bit? It seems like you could cut back on quantization error by increasing bits, but if you're doing some manual SIMD magic to get this performance I can understand sticking with 32 bits.