- debug_to_file
- the bounds query interface
- multi-output pipelines
- overriding the runtime
- lambdas
  useful tricks, e.g., `Buffer<float> image_buf = lambda(x, y, (sin(x+y)+1)/2).realize(10 * tile_size, 10 * tile_size);`
- extern functions and extern stages
- common scheduling patterns
- scheduling rvars
- tail strategies
- Wrapping existing memory in a Halide::Buffer