DisplayPort Packet Priority: Video, Audio, and Data

The DisplayPort standard manages bandwidth contention on its Main Link through a deterministic, packet-based time-division architecture rather than a dynamic software-driven queue. High-priority isochronous streams—primarily active video—are given absolute temporal precedence based on pixel timing, while audio, metadata, and general data packets are multiplexed into specific framing intervals or reserved time slots. This article details the structural hierarchy and arbitration mechanisms the VESA DisplayPort specification uses to prevent packet collision and guarantee that video and audio arrive without latency or dropouts.

The Micro-Packet Architecture

DisplayPort replaces conventional raster-based digital transmission with a packetized protocol derived from PCI Express. The Main Link operates at a fixed, unthrottled link rate (such as RBR, HBR, HBR3, or UHBR) regardless of the pixel clock rate of the incoming video stream.

Because the link rate generally exceeds the native stream data rate, transmission is governed by a micro-packet framing mechanism using special control symbols. This structure allows video, audio, and secondary data to interleave predictably across physical lanes.

Active Video Priority

In Single-Stream Transport (SST) mode, active video holds strict temporal priority during the active display period:

  • Raster Synchronization: When pixels are actively being drawn across a scanline, pixel data occupies the physical link layer. Video data is mapped into data streams framed by dedicated start-of-frame and end-of-frame control symbols.
  • Stream Reservation: Video data packets cannot be interrupted or fragmented arbitrarily by auxiliary data during the active video period. The transmitter utilizes a Transfer Unit (TU) structure—typically 32 to 64 link symbols per lane—to consistently feed pixel data alongside minimal link-maintenance symbols.

Blanking Periods and Secondary Data Packets (SDP)

Because active video occupies the link during the active raster lines, all non-video traffic contends for bandwidth primarily during the Horizontal Blanking Interval (HBI) and Vertical Blanking Interval (VBI).

Data sent during blanking intervals is encapsulated into Secondary Data Packets (SDPs). DisplayPort categorizes SDPs into specific priorities based on latency sensitivity:

  1. Audio Timestamp Packets (ATP): Crucial for maintaining Audio-to-Video synchronization (lip sync). Because drift impairs audio reproduction, timestamp packets receive immediate transmission scheduling once a blanking period opens.
  2. Audio Stream Packets (ASP): Audio samples require continuous delivery with minimal jitter. Buffer sizes on the receiver side are limited, so ASPs are dispatched at predictable intervals across the HBI and VBI before lower-priority payloads.
  3. High-Priority Metadata: Dynamic High Dynamic Range (HDR) metadata, Extended Display Identification Data (EDID) updates, and Content Protection (HDCP) handshakes are prioritized next to avoid frame-to-frame rendering delays.
  4. General Control and Information Packets: Static InfoFrames (such as AVI InfoFrames, Vendor-Specific InfoFrames, and colorimetry data) utilize remaining blanking slots.

If a blanking interval lacks sufficient bandwidth to transmit both high-priority audio samples and oversized metadata, the transmitter holds the non-critical metadata in a local buffer until subsequent HBI or VBI phases occur.

Multi-Stream Transport (MST) Slot Allocation

In Multi-Stream Transport (MST) mode, multiple independent audio/video streams share a single physical Main Link. DisplayPort eliminates runtime packet contention in MST environments by implementing an explicit Time-Division Multiplexing (TDM) scheme:

  • 64 Time Slots: Every link transmission unit is organized into 64 discrete time slots per frame.
  • Virtual Channels: Each video/audio stream is assigned a specific number of time slots by the source device, determined during topology discovery and link training.
  • Guaranteed Bandwidth: Because slots are strictly allocated upstream, contention across competing video and audio streams is resolved before transmission occurs. Stream 1 cannot encroach upon the time slots granted to Stream 2.

Within each individual virtual channel's allocated slots, the same hierarchy applies: active video reserves transmission capacity, followed by audio SDPs, followed by general metadata during the channel's respective blanking portions.

At the bottom of the priority hierarchy are dummy or "stuffing" symbols (such as the FS and FE framing symbols). When neither active video nor secondary data packets are queued for transmission—or when the allocated TU capacity exceeds the data payload—the transmitter inserts stuffing symbols. These maintain link synchronization and clock recovery without taking bandwidth from functional data.