Pipelining Computational Resources in General-Purpose Graphics Processing Units

Abstract

This disclosure describes techniques for extending the architecture of a general-purpose graphics processing unit, GPGPU, with parallel processing units to allow efficient processing of pipeline-based applications. The technique includes configuring local memory buffers connected to parallel processing units operating as stages of a processing pipeline to hold data for transfer between the parallel processing units. The local memory buffers allow on-chip, low power, direct data transfer between the parallel processing units. The local memory buffer may include a hardware-based data flow control mechanism to enable transfer of data between the parallel processing units. In this way, data can be passed directly from one parallel processing unit to the next parallel processing unit in the processing pipeline via the local memory buffer, effectively transforming the parallel processing unit into a series of pipeline stages.

Description

technical field [0001] The present invention relates to processing data, and more particularly, to processing data using a general purpose graphics processing unit (GPGPU). Background technique [0002] A general-purpose graphics processing unit (GPGPU) is a generalized version of a graphics processing unit originally designed to handle 2D and 3D graphics. GPGPU extends the high-power parallel processing of GPUs to general data processing applications beyond graphics processing. As one example, a GPU may be configured to process data according to the OpenCL specification, which gives certain applications access to a graphics processing unit for non-graphics computing. The "OpenCL Specification, Version 1.1" was published in June 2010 and is publicly available. [0003] GPGPUs consist of programmable processing units arranged in a highly parallel architecture that does not allow data sharing or synchronization between processing units. Instead, individual processing units ...

AI Technical Summary

Problems solved by technology

This limitation extends to 2D and 3D graphics processing that uses parallel processing at each processing stage but requires computational resources to be pipelined between stages

Images