https://hgpu.org/?p=12116
Vector Quantization: A Many-Core Approach