SPRUJ28E November 2021 – September 2024 AM68 , AM68A , TDA4AL-Q1 , TDA4VE-Q1 , TDA4VL-Q1
The VISS operation is controlled through a single HTS thread. For details of the HTS operation refer to Section VPAC Hardware Thread Scheduler (HTS). The VISS configuration is valid for a single or multiple frame. External-host manages VISS configuration at end of each frame, if required. The HTS controls image processing between lines and handles managing shared memory for VISS inputs/outputs. Sub-frame processing is not natively supported. SW must configure VISS and HTS to start initialization process before enabling sensor capture. After the initialization sequence, the HTS will generate a start trigger to VISS. The generation of the start trigger depends on the availability of H3A data out buffer and FCP data out buffer inside the SL2 memory.
At this stage, the LSE waits for CSI RX video port (VP) to start sending valid pixels. As and when VP data starts streaming in, it is routed to the RFE video port. As RFE, CAC/NSF4V/GLBCE, and FCP are running in streaming mode, the FCP indicates to LSE about completion of one line of operation . After receiving valid data out, the HTS will trigger UTC for data store into DDR. Each 'thread done' accompanies output mask for all output buffers. Mask bits indicate to HTS about validity of output buffer as for each thread done all output buffers do not need to be produced. The HTS will issue next start after checking all output buffers availability. The streaming data from CSI RX is stallable upto 2KB storage inside VP memory of CSI RX. Any temporary stall can be absorbed, but prolonged stall is buffer overflow inside CSI RX. In VPAC, VISS produced data is transferred using Real Time fabric of NAVSS/MSMC and low predictable latency needs to be guaranteed.
The following OTF operation steps are illustrated in Figure 6-53:
The following assumptions are made for the OTF operation described above: