Compiler techniques for mapping program code to a high performance, power efficient, programmable image processing hardware platform

A computer program product, encoded on one or more non-transitory computer storage media, comprising instructions that when executed by a processor comprising:

• (i) a two-dimensional array of processing elements,
• (ii) a two-dimensional shift-register array having a first portion of registers that are each dedicated to one of the processing elements in the two-dimensional array of processing elements and having a halo portion of registers that border the first portion of registers on one or more sides of the first portion, and
• (iii) a sheet generator configured to load sheets of image data into the two-dimensional shift-register array, wherein each sheet of image data has at least as many pixels as processing elements in the two-dimensional array of processing elements,
• cause the processor to load input data to perform a stencil function requiring data from multiple sheets of image data, wherein the instructions cause the processor perform operations comprising:
• loading a first sheet of image data into the first portion of the two-dimensional shift-register array,
• loading a portion of a second sheet of image data into the halo portion of the two-dimensional shift-register array, and
• performing a first iteration of the stencil function using the first sheet of image data loaded into the first portion of the two-dimensional shift-register array and using the portion of the second sheet of image data loaded into the halo portion of the two-dimensional shift-register array.

A method performed by a processor comprising:

• (i) a two-dimensional array of processing elements,
• (ii) a two-dimensional shift-register array having a first portion of registers that are each dedicated to one of the processing elements in the two-dimensional array of processing elements and having a halo portion of registers that border the first portion of registers on one or more sides of the first portion, and
• (iii) a sheet generator configured to load sheets of image data into the two-dimensional shift-register array, wherein each sheet of image data has at least as many pixels as processing elements in the two-dimensional array of processing elements,
• wherein the method causes the processor to load input data to perform a stencil function requiring data from multiple sheets of image data, the method comprising:
• loading, by the sheet generator, a first sheet of image data into the first portion of the two-dimensional shift-register array,
• loading, by the sheet generator, a portion of a second sheet of image data into the halo portion of the two-dimensional shift-register array, and
• performing, by the two-dimensional array of processing elements, a first iteration of the stencil function using the first sheet of image data loaded into the first portion of the two-dimensional shift-register array and using the portion of the second sheet of image data loaded into the halo portion of the two-dimensional shift-register array.