Abstract: The subject technology discloses configurations for receiving, by a first process, a set of input events from an application in which the set of input events includes a set of input update commands. The first process writes the set of input update commands into a low-latency graphics pipeline. The subject technology dispatches, by the first process, the set of input update commands from the low-latency graphics pipeline to a second process. The second process receives the set of input update commands from the low-latency graphics pipeline. The subject technology then writes, by the second process, a set of input data into a shared graphics processing unit (GPU) texture.

Abstract: Processing an input event within an application includes detecting an input event within an application executing on a first thread, the input event being associated with an event handler. A separate execution corresponding to a current state of the application is performed on a second thread based on the event handler associated with the input event. Within the separate execution, a determination is made whether the event handler modifies at least one of a document associated with the application or a default behavior of the application. In a case where the event handler does not modify at least one of the document or the behavior, the subject technology refrains from invoking the event handler on the first thread.