Sample DPL Applications

This section provides NVIDIA DOCA Pipeline Language sample applications that demonstrate how to develop, compile, and deploy programs using the NVIDIA DOCA Pipeline Language (DPL). These examples are intended to help developers become familiar with DPL concepts and runtime behavior on both host and BlueField®-3 platforms.

DPL programs can be compiled on any host machine with the DPL compiler installed. Once compiled, the output binary can be loaded onto:

• The same host machine (for development/debugging purposes), or

• A BlueField-3 DPU where the DPL Runtime daemon is running

Refer to the following documents for setup and troubleshooting:

• DPL Runtime Service – For instructions on installing and configuring DPL on BlueField

• DPL Release Notes – For known issues and guidance on installation, compilation, and execution of DPL sample applications

1. Compile the DPL program using the DPL compiler.

2. Locate the generated output files in the compiler output directory:

   • .p4info – Describes the program's tables and actions

   • .dplconfig – Contains the binary pipeline configuration

3. Load the program onto the BlueField target using a P4Runtime controller (e.g., p4runtime_sh.sh, NVIDIA's controller, or a custom solution)

Once loaded, you can use the controller to populate flow tables, query runtime state, and validate pipeline behavior.

• Hello Packet Example

• GTP Parsing Example

• GTP Tunnel Encapsulation Example

• Geneve TLV Parsing Example

• VXLAN Tunnel Gateway Example

• Connection Tracking Example

• Host-based Networking Example

• Equal-Cost Multi-Path Routing

• Entropy in Hash-Based Load Balancing

• NAT64 Example