Event data recording buffering

Abstract

Aspects presented herein may enable a device to write received data just once when the received data is consumed both by a regular real-time pipe-line and by an event data recording round-robin buffering, thereby reducing additional dynamic random-access memory (DRAM) write bandwidth consumed during nominal operation(s). In one aspect, a device allocates a first set of buffers in a plurality of buffers for a set of pre-processed frames, where the first set of buffers is associated with a first round-robin cycle with a first time-span. The device allocates a second set of buffers in the plurality of buffers for preserving a set of previously processed frames, where the second set of buffers is associated with a second round-robin cycle with a second time-span. The device enables the second set of buffers to be accessed when a triggering condition occurs.