[Paper Notes] Halide: Decoupling Algorithms from Schedules for High-Performance Image processing.

Abstract of the idea:
It makes abstract language of the array operations in the image processing process. Using auto optimization to achieve portability and high performance at the same time. In terms of the idea, it is very similar to Google Flume pipeline implementation, which is let user specify the steps of parallel function but do auto fuse of the steps to make it runs faster.

你可能感兴趣的:([Paper Notes] Halide: Decoupling Algorithms from Schedules for High-Performance Image processing.)