Merging rendering operations for graphics processing unit (GPU) performance

摘要

Disclosed are apparatus and methods for rendering using a graphics processing component (GPC). A computing device can receive GPC instructions, with each instruction associated with a portion of a canvas. For each instruction: a first portion of the canvas can be determined and an instruction buffer can be searched to identify a mergeable instruction for the instruction. After identifying the mergeable instruction at a mergeable position within the instruction buffer, one or more instructions in the instruction buffer that include the mergeable instruction can be searched for overlap in the canvas with the first portion. In the event that the one or more instructions do not overlap with the first portion of the canvas, then the instruction can be merged with the mergeable instruction. Each instruction in the instruction buffer can be provided to the GPC for performing the instruction.

主权项

1. A method, comprising:
receiving, at a computing device, one or more instructions for a graphics processing component of the computing device, wherein each instruction is associated with at least a portion of a canvas; after receipt of each instruction of the one or more instructions, the computing device:
determining a state of the graphics processing component, wherein the state is represented by a matrix, and wherein the matrix aggregates information about transformations for the graphics processing component;associating the instruction and the state;determining a first portion of the canvas associated with the instruction;searching an instruction buffer to identify a mergeable instruction for the instruction;after identifying the mergeable instruction at a mergeable position within the instruction buffer:
determining whether a state of the graphics processing instruction associated with the mergeable instruction differs from the state of the graphics processing component associated with the instruction;searching one or more instructions in the instruction buffer for overlap in the canvas with the first portion, wherein the one or more instructions in the instruction buffer comprise the mergeable instruction, andin the event that the one or more instructions in the instruction buffer do not overlap the first portion of the canvas and that the state of the graphics processing instruction associated with the mergeable instruction does not differ from the state of the graphics processing component associated with the instruction, merging the instruction with the mergeable instruction; and for each instruction in the instruction buffer, the computing device:
providing the instruction to the graphics processing component, wherein the graphics processing component performs the instruction.