Nvidia Orin Edge

The Consystence edge runtime runs on an Nvidia Jetson Orin in our workshop today. It is the same edge runtime that targets the PLCnext VL3 (a single-silo Orleans application on .NET 10), and it talks to controllers through the same driver layer. What differs is the hardware it runs on and the way it reaches its controllers.

Workshop host, not a released install path

In our workshop, the Orin edge reads a physical ControlLogix directly over EtherNet/IP CIP and a virtual PLCnext over gRPC. That is workshop evidence, not field experience. The edge runtime is not yet publicly released; Consystence provides it through beta engagements. Treat this page as the target setup, not a released install procedure.

If you have read the VL3 setup guide, there is very little here that is new. This page exists to answer one question precisely: what is specific to running the edge on an Orin? The short answer is very little — a different controller link and a couple of host-specific config values — and specifically, an Orin is not a standalone GPU inference appliance.

Same runtime, different hardware

There is one edge runtime, built from the same code for either host: the same .NET 10 single-silo Orleans runtime, the same store-and-forward buffer, and the same driver layer.

flowchart TD
  RT["Consystence edge runtime<br/>.NET 10 · single-silo Orleans"]
  RT --> VL3["PLCnext VL3<br/>(target platform)"]
  RT --> ORIN["Nvidia Jetson Orin<br/>(workshop host)"]
  VL3 -->|"gRPC over UDS"| BG["Co-located PLCnext"]
  BG -->|"EtherNet/IP CIP"| FIELD["Field controllers and IO"]
  ORIN -->|"EtherNet/IP CIP<br/>(direct)"| CLX["ControlLogix"]
  ORIN -->|"gRPC over TCP<br/>(default port 41100)"| VPLC["Remote or virtual PLCnext"]

Because the runtime is hardware-agnostic, most of what is documented for the VL3 applies unchanged on an Orin — the one structural difference is the controller link:

• PLC communication — the VL3’s gRPC-over-UDS path (/run/plcnext/grpc.sock) does not apply on an Orin: a Unix domain socket is machine-local, and an Orin has no co-located PLCnext runtime. An Orin can speak EtherNet/IP CIP directly to a ControlLogix (the ethernet-ip driver) and can reach a separate PLCnext over gRPC on TCP (PlcNext:GrpcHost, with PlcNext:GrpcPort, default 41100). The workshop Orin does both at once. See PLC communication.
• Driver selection — the same Driver values (plcnext-grpc, ethernet-ip, stub; Modbus TCP is roadmap); no Orin-specific driver exists.
• Store-and-forward — SQLite buffer over gRPC (HTTP/2), surviving upstream disconnects and power cycles.
• IT/OT separation — keep the IT (upstream) and OT (field) interfaces on separate network segments. The network security reference policy applies to an Orin as to any edge host.
• Identity and licensing — the identity sent at registration is the configured Edge:Id, required in the edge’s local configuration on any host (bootstrap fails without it; the target design is an activation flow that enrols the device by its write-once install identity, with Edge:Id remaining a display label — not yet shipped). That write-once install ID, minted on the device at first boot, is also the licensing identity to which the account-signed offline licence artifact is bound. An edge install draws an Edge seat from the organisation’s licence pool, and licence state never gates control.

Tip

Follow the VL3 edge setup page for an Orin as well — there is no separate Orin install path, and the systemd unit, configuration model and registration handshake are the same. Two config values are host-specific: the PLC drivers (see above) and the interface names. The reference network policy defaults to eth0 (IT) and eth1 (OT), so on an Orin set Firewall:ItInterface and Firewall:OtInterface, and the NetworkMonitor interface names, to its actual interface names.

The Orin’s GPU

The distinguishing property of an Orin is its GPU. Today the edge runtime does not use it for anything: there is no on-edge TensorRT or ONNX inference shipping on any edge host. The GPU matters only as forward provisioning for Edge ML.

Edge ML is on the roadmap

On-edge machine learning is not yet available. The planned model is cloud-trained ONNX, executed on the edge in an advisory capacity. CPU inference is the baseline, so it does not depend on a GPU, and GPU-compiled inference (TensorRT) on an Orin is the design option for GPU-dependent workloads such as vision models and complex autoencoders. None of this ships yet: the Orin’s GPU does not enable any inference feature today. Treat it as future capacity, not a current feature.

AI is advisory only

Advisory AI runs off site today: a site server relays its requests through the cloud, or, where its administrator has enabled a direct route in the server’s settings, sends them directly to a model provider. Site-local inference is roadmap. Advisory AI and any future edge ML only read, explain and recommend. They never command equipment. Control actions remain with the controller logic and hard interlocks in the IO modules and safety relays. An Orin GPU does not change this boundary. See AI architecture.

What Orin is not

Orin in Consystence is a general-purpose edge node that happens to have a GPU. It is not:

• a standalone TensorRT inference appliance — it runs the full edge runtime (tag polling, store-and-forward, upstream streaming), not just a model server;
• a PLCnext host — an Orin does not run the PLCnext control runtime. It can connect directly to an existing ControlLogix over EtherNet/IP CIP, and can reach a separate PLCnext (such as a PLCnext border gateway or a virtual PLCnext) over gRPC on TCP; control logic stays in the controllers either way;
• a different software product — there is no “Orin edition”, only the one edge runtime on different silicon.

Setup

Use the VL3 edge setup guide as the procedure. It describes the target installation, and its systemd unit, registration handshake and network-separation approach apply unchanged on an Orin (the workshop Orin runs the edge as a systemd service). Set the required Edge:Id in local configuration exactly as you would on any host, choose the drivers for your controllers, and set the interface names to the Orin’s own (see the tip above).

For where the Orin and the VL3 fit alongside site servers and the border gateway, see deployment topologies and the edge overview.