NVLink and NVSwitch Architecture Explained

NVIDIA NVLink and NVSwitch Architecture Explained

Network Insight / Engineering Guide 02 / Lab 03

NVSwitch Internal Fabric Path Explorer

Investigate GPU-to-GPU communication through a switched NVLink fabric. Trace ingress, switching and egress stages, observe contention and inject logical switch-plane faults.

Logical switched GPU fabric

Available logical planes —
Estimated fabric utilisation —
Modelled effective bandwidth —
Path status —
Component Inspector

Select a GPU or logical NVSwitch plane to inspect its role in fabric connectivity.

Fabric ready

Choose a source and destination GPU, then trace the switched communication path.

Engineering interpretation

A switched fabric enables communication between GPUs without requiring every GPU to have a dedicated direct physical link to every peer.

Awaiting fabric path analysis.
Network Insight / Engineering Guide 02 / Lab 05

GB200 NVL72 Rack Architecture Explorer

Explore the physical and logical architecture of a rack-scale Grace Blackwell system. Inspect compute trays, NVSwitch trays, GPU density, scale-up connectivity and liquid-cooling infrastructure.

Interactive rack-scale architecture

Blackwell GPUs 72
Grace CPUs 36
Compute trays 18
NVSwitch trays 9
Rack Component Inspector

Select a compute tray or NVSwitch tray to inspect its architectural role.

Rack operational

All logical rack components are available.

Engineering interpretation

GB200 NVL72 integrates 72 Blackwell GPUs and 36 Grace CPUs into a rack-scale system interconnected through an NVLink switch fabric.

Select a component to begin the architecture investigation.
Network Insight / Engineering Guide 02 / Lab 06

NCCL Collective Communication Simulator

Execute collective communication operations across a multi-GPU topology. Investigate how distributed data is reduced, gathered, scattered and broadcast through different communication patterns.

Animated collective communication topology

Execution step 0 / 0
Communication rounds —
Estimated data movement —
Collective state Ready
GPU Data Inspector

Select a GPU node to inspect its local contribution and current collective data state.

Collective ready

Choose an operation and execute the communication sequence.

Engineering interpretation

Collective communication algorithms coordinate data movement across GPUs. NCCL selects algorithms and transport mechanisms according to the operation, topology and environment.

Awaiting collective communication.
Network Insight / Engineering Guide 02 / Lab 07

Scale-Up vs Scale-Out Traffic Path Lab

Trace GPU communication inside an NVLink domain and across external Ethernet or InfiniBand fabrics. Examine RDMA transport, communication boundaries and the impact of scale-out bandwidth limitations.

Hybrid GPU communication architecture

Communication domain —
Modelled effective bandwidth —
Illustrative transfer time —
Communication status —
NVLink domain boundary

NVLink and NVSwitch provide high-bandwidth communication within a supported GPU scale-up domain. Communication between independent domains generally requires network adapters and a scale-out fabric.

Fabric ready

Select a communication path and run the analysis.

Engineering interpretation

Scale-up and scale-out interconnects serve different purposes. GPU placement determines whether traffic stays inside the NVLink domain or traverses a network fabric.

Awaiting GPU communication trace.
☕
Support Network Insight
Help support the development of interactive networking labs, BGP simulations, and educational content.
Support the Project
Matt Conran
Latest posts by Matt Conran (see all)
Tags: No tags

Comments are closed.