SPRUJ28E November 2021 – September 2024 AM68 , AM68A , TDA4AL-Q1 , TDA4VE-Q1 , TDA4VL-Q1
Rotation and mirroring
The Video Accelerator supports 8 modes of rotation and mirroring. The PRP block can do rotations of the source picture in 90, 180, or 270 degrees and vertical or horizontal flip before starting to encode the operation.
Bit-depth and chroma format conversion
The PRP block is responsible for source format conversion in the encoder IP. 422 (planar/semi-planar) to 420 conversion can be made before encoding operation.
Frame buffer compression
To overcome bandwidth suffering and at the same time to ensure real-time encode/decode performance, TI has carefully designed lossless, great access patterned frame buffer compression technology and employed it into the WAVE IP series. WAVE521CL achieves significant reduction in bandwidth while sustaining fast processing capability.
Long Burst Write
The Video Accelerator includes burst write back (BWB) module that enables 128-byte burst write access to external memory for better bus utilization. It collects a group of short write requests from loop filter block in 128-byte unit and sends it out to the external memory more efficiently. BWB works for an external Display module or other post processing module that needs to read decoded frames in raster scan order.
The FBC/FBD block also writes or reads 128-byte burst length of compressed frame data from/to the externalmemory.
3DNR
The Video Accelerator employs temporal noise reduction technology, also called 3DNR (3D noise reduction), for better quality of image under low light. Video noise appears easily in a low light level. It is an important issue in surveillance camera because images are captured often under dark conditions or indoors, which can make identification difficult.
The Video Accelerator can detect and filter noise for each color component Y, Cb, and Cr of a frame. This achieves both good video quality and enhancement of coding efficiency. Noise reduction is done while encoding. Therefore, additional read/write bandwidth is not required.
Latency tolerance
The Video Accelerator can afford to reach real time performance under memory access latency if the number of cycles per CTU retains less than 500 cycles. The Video Accelerator is designed to be less sensitive to pipeline delay, especially at a peak bitrate, which might eventually incur performance drop. With use of decoupling techniques and inter-pipe queues, it can hide this sort of delay and deliver high performance at any situation.
Subframe synchronization
The Video Accelerator supports subframe synchronization for low latency coding. By receiving dedicated subframe-ready signals from the image process unit (IPU) or by setting host interface registers from host CPU, the Video Accelerator can start encoding process as early as the minimum set of raw video data is ready.
Programmability
The Video Accelerator embeds the 32-bit CPU and 16-bit DSP dedicated to bitstream processing and their video hardware control. Host interface registers and interrupts are provided for communication between a host processor and the Video Accelerator.
Low Power Consumption
The Video Accelerator can ensure ultra-low power operation by gating the clock sources for some internal hardware blocks in an idle state.
Trick mode
The Video Accelerator supports trick mode such as fast/slow forward playback and fast/slow reverse playback for video applications. By seeking an intra random access point (IRAP) picture as entry points of bitstream, the Video Accelerator can skip non-IRAP pictures and start decoding from an IRAP picture or decode a thumbnail of a picture.
Frame-based processing
V-CPU processor operates video operation on a frame by frame basis. While frame operation is running, there is no need for communication between the host processor and the Video Accelerator. This is the key feature for promising the lowest burden to host processor for video operations.
By issuing a picture processing, the host application can perform its own operations until it receives an interrupt from the Video Accelerator informing of the completion of picture processing.
Handling multi-instances
The Video Accelerator supports multiple instances, which can be helpful for multi-channel encoder and/or decoder applications. To support this multi-instance operation, the Video Accelerator uses an internal context parameter set for each instance. While creating a new instance and starting picture processing, a set of these context parameters is created and updated automatically in the Video Accelerator. Because of this internal context management scheme, different encoder and/or decoder tasks running on the host processor can control Video Accelerator operations independently with their own instances.
When creating a new instance, an application task will get a new handle specifying an instance if a new handle is available on the Video Accelerator. All of the following operations for the given application task could be separately handled on the Video Accelerator by using this task-specific handle. Because the Video Accelerator can only perform one picture processing task at a time, each application task should check whether the Video Accelerator is ready or not before starting a new picture operation. By calling a function for closing a certain instance, the application can easily terminate a single task of video operation on the Video Accelerator.