Fast queries in a multithreaded queue of a graphics system

摘要

According to one aspect of the invention, a query request is received from a client application at a multithreaded graphics framework. The multithreaded graphics framework including a first thread configured to place graphics commands from the client application into a command queue and a second thread configured to dispatch the graphics commands from the command queue to a graphics processing unit (GPU) for execution. In response to the query request, the first thread is configured to store query information of the query request in a query block of memory that is separated from the command queue and notifying the second thread. In response to the notification, the second thread is configured to issue a query command to the GPU based on the query information retrieved from the query block, prior to dispatching remaining graphics commands pending in the command queue.

主权项

1. A computer-implemented method for processing queries of a graphics system, the method comprising:
receiving a query request from a client application at a multithreaded graphics framework, the multithreaded graphics framework including a first thread configured to place a plurality of graphics commands from the client application into a command queue and a second thread configured to dispatch the graphics commands from the command queue to a command buffer of a graphics processing unit (GPU) for execution; in response to the query request, the first thread storing query information of the query request in a query block of memory that is separated from the command queue, wherein the query request is to query an operating status of the GPU, wherein query information identifies a type of query associated with the operating status of the GPU to be queried; notifying the second thread that the query information of the query request has been stored in the query block; in response to the notification, the second thread retrieving the query information from the query block; and the second thread issuing a query command to the command buffer of the GPU based on the query information retrieved from the query block, prior to dispatching remaining graphics commands pending in the command queue, wherein the query command is executed by the GPU prior to executing the remaining graphics commands in the command queue.