How Does DisplayPort Multi-Stream Transport Work?

DisplayPort Multi-Stream Transport (MST) enables a single video output to drive multiple independent displays through daisy-chaining or hub-based routing. Introduced in the DisplayPort 1.2 standard, MST replaces traditional point-to-point display links with a packetized, multi-node topology. By multiplexing independent audio and video streams into dynamic time slots across high-speed physical lanes, the host GPU addresses, synchronizes, and powers multiple distinct monitors over a single physical cable interface.

Architectural Transition: SST to MST

In standard Single-Stream Transport (SST) mode, a DisplayPort link operates strictly point-to-point. A single transmitter sends pixel data directly to a single receiver, consuming the full available bandwidth of the main link for that single display.

MST changes the transport layer by introducing a packet-switched routing mechanism. Instead of mapping one frame buffer to one physical link, the transmitter divides the aggregate bandwidth into 64 logical time slots per frame. Multiple independent video and audio streams are then packetized and mapped into these allocated time slots. This allows completely distinct resolutions, refresh rates, and color formats to travel simultaneously over the same differential pairs.

Topologies and Physical Connection Models

MST supports two primary hardware configurations:

• Daisy-Chaining: In a daisy-chain setup, the host GPU connects to the primary monitor's DisplayPort input. That monitor must feature an integrated MST branch device and a dedicated DisplayPort output port. The secondary monitor connects directly to the primary monitor, and additional displays can follow down the chain until the link bandwidth or hardware node limits are reached.
• MST Hubs: A standalone MST hub receives the primary DisplayPort connection from the host system and demultiplexes the incoming stream across multiple physical output ports. These downstream ports can be native DisplayPort connectors or convert to secondary standards such as HDMI or DVI using integrated protocol converters.

Link Management and Dynamic Discovery

To coordinate multi-monitor configurations without manual addressing, the DisplayPort architecture relies on sideband messaging across the Auxiliary (AUX) channel.

The DisplayPort Configuration Data (DPCD)

When an MST connection initializes, the host queries the downstream device’s DPCD registers via the half-duplex AUX channel. The host reads the topology to identify whether the connected device is a simple display endpoint (Sink device) or an MST routing node (Branch device).

Topology Enumeration via Sideband Messaging

Once an MST branch device is detected, the source initiates sideband messaging to map the downstream tree:

1. LINK_ADDRESS Requests: The source sends discovery packets asking the branch device to report all connected ports, their types, and their current link states.
2. Hot-Plug Detection (HPD) Propagation: When a monitor downstream is plugged in or unplugged, the branch device generates an HPD_IRQ pulse upstream. The source intercepts this interrupt and re-queries the topology to update its display map.
3. RAD (Relative Address) Routing: Every downstream node receives a unique hierarchical address based on its physical port routing (for example, port 2 of branch 1). This ensures control commands target the exact physical monitor intended.

Virtual Channel Payload Allocation

The distribution of video data relies on Virtual Channel (VC) payloads. The transmission mechanism operates as follows:

1. Bandwidth Negotiation: The host performs link training over the physical Main Link (evaluating parameters across RBR, HBR, HBR2, HBR3, or UHBR depending on the standard version) to establish total link capacity.
2. Payload Calculation: The source driver calculates the required bandwidth for each display stream based on its active timing parameters (horizontal and vertical resolution, refresh rate, and bits per pixel).
3. Time Slot Assignment: The source writes to the branch device's payload allocation table via AUX commands, assigning specific time slots (out of the 64 available) to each Virtual Channel ID (VCID).
4. Multiplexing and Demultiplexing: The source hardware wraps each video stream into Multi-Stream Packets (MSPs) tagged with their respective VCID. As these packets traverse the chain, each branch device inspects the header, extracts the data designated for its local sink, and passes the remaining packets to downstream ports.

Bandwidth Constraints and Display Stream Compression

Because all downstream monitors share the capacity of the original link, the aggregate bandwidth of all connected displays cannot exceed the maximum data rate of the source port. For example, a DisplayPort 1.4 connection running at HBR3 provides a maximum effective data rate of 25.92 Gbps. This pool must accommodate the sum of all individual video timings.

To prevent bandwidth saturation at high resolutions, modern DisplayPort implementations integrate Display Stream Compression (DSC). DSC is a visually lossless, low-latency compression algorithm that reduces the bandwidth footprint of each stream by up to a 3:1 ratio before packetization. By reducing the payload size per frame, DSC enables multiple 4K or 8K displays to run concurrently across a single MST connection that would otherwise exceed raw link limits.