WOLF-167S
Overview

System Switch + HPC in One 3U Slot
The VPX3U-BW2000E-SWITCH rugged 3U VPX module integrates three high-performance elements in a single, slot-saving SOSA-aligned switch card; NVIDIA RTX™ 2000 Blackwell embedded GPU (GB206), Configurable PCIe Gen4 switch, and a Configurable 200 GbE network switch.
The Blackwell GPU delivers parallel compute for embedded HPC/AI workloads and includes 8 GB of GDDR7 memory with ECC support. The module can operate as both the system fabric switch and an HPC node, allowing the PCIe and Ethernet fabrics to be routed while still provisioning a dedicated PCIe path to feed the GPU.
Module power is configurable from 75W to 140W, enabling the designer to tune power/thermals to chassis cooling capability and mission profiles.
Block Diagram
The block diagram illustrates how the RTX 2000 (GB206) Blackwell GPU, PCIe Gen4 switch, and 200 GbE network switch are integrated on the 3U VPX module, including the PCIe uplinks to the VPX backplane and the Ethernet fabric paths used to route up to 8×25GbE ports (plus 1000BASE-T) across the chassis.
Features
WOLF-167S is a rugged 3U VPX system switch + HPEC module that combines an NVIDIA RTX 2000 (GB206) Blackwell embedded GPU with a PCIe Gen4 switch and a 200GbE network switch in a single slot-saving design. It supports 8 GB of GDDR7 (128-bit) VRAM with ECC and a PCIe x8 GPU data path, enabling embedded AI/HPC processing while simultaneously routing PCIe and Ethernet fabrics within the chassis. The integrated Ethernet switch supports up to 8×25GbE ports plus 1×1000BASE-T, and the module power is configurable from 75W to 140W to match platform cooling and mission needs.
GPU Features
- Blackwell GPGPU parallel processing:
- CUDA Toolkit 12, Compute capability 10.0
- OpenCL 3.0, DirectX 12 Ultimate, OpenGL 4.6, OpenGL ES 3.2, Vulkan 1.2
- 5th Gen Tensor Cores with additional new data precisions (FP4 and FP6, FP8 Gen2)
- 8 GB GDDR7 128-bit VRAM with ECC support
- NVENC (9th Gen) and NVDEC (6th Gen) with up to 8K video encoding and hardware decoding support
Connectivity / System Management
- Configurable PCIe Gen4 switch
- Configurable 200 GbE network switch:
- 8 ports up to 25GbE
- 1 port 1000BASE-T
- Time-Sensitive Networking (TSN) supported
- NVIDIA GPUDirect RDMA support
- IPMI system management
- Linux and Windows drivers
Mechanical / Open Systems Architecture
- High level of ruggedization:
- Rugged conduction cooled
- Operating temp: -40°C to +70°C standard, operational to +85°C
- Vibration Random: VITA 47.1 Class V3 (5 to 2000Hz)
- Vibration Sine: 10g peak (5 to 2000 Hz)
- Shock: 40G (MIL-STD-810H, Method 516.8)
- Dimensions: 160mm x 100mm x 25.4mm
- Weight (approximately): 1.6kg
- ANSI/VITA 48, 65 (VPX-REDI, OpenVPX)
- SOSA Aligned switch slot profile: 14.4.15
Specifications
Key configuration and performance parameters for WOLF-167S are summarized below for quick comparison. Values shown reflect the standard module configuration; contact WOLF if you need a tailored build or a specific payload profile.
Configuration Guide
The following table defines a series of common order codes for the VPX3U-BW2000E-SWITCH module. The asterisks denote characters of the part number that are defined based on common configuration options.
Some configuration options for this module include:
- Network config options
- PCIe config options
- Variant locked
- Default power threshold
- Conformal coatings
Contact Sales for the latest Ordering Numbers and available options.
Representative ordering numbers
| Ordering Number | Description |
|---|---|
| 167S33-F***-***VPX3vA0 | 3U VPX, Conduction Cooled, 1", 14.4.15 switch profile, NVIDIA Blackwell RTX 2000, 8GB GDDR7, PCIe P1 x16 and P2 x8, Network 8 ports each 25GbE |
| 167S33-F***-***VPX3vA0 | 3U VPX, Conduction Cooled, 1", 14.4.15 switch profile, NVIDIA Blackwell RTX 2000, 8GB GDDR7, PCIe P1 x16 and P2 x4 + x4, Network 8 ports each 10GbE |
FAQ
WOLF-167S combines three functions in one 3U VPX slot: an embedded GPU compute node, a PCIe Gen4 switch, and a high-bandwidth Ethernet switch fabric.
It is typically used when:
- You want a slot-saving system switch that can also run GPU workloads
- You need to route PCIe and Ethernet fabrics while still provisioning a dedicated GPU data path
- You want to reduce the number of separate switch and compute cards in the chassis
Choose WOLF-167S when slot count, weight, and fabric simplicity matter and you can accept a combined-function card.
It is most compelling when:
- You are tight on 3U slots and need both switching and compute
- You want shorter integration cycles with fewer inter-card dependencies
- Your platform can accommodate the combined power and thermal load in one slot
If you need independent scaling of switch capacity and GPU capacity, separate cards can be cleaner.
Prioritize:
- PCIe Gen4 fabric stability and correct routing for the intended uplinks and endpoints
- Ethernet switch behavior under load (port mapping, throughput, latency, TSN behavior if used)
- GPU data path validation (sustained transfers and workload stability)
- Power/thermal margin at your chosen power threshold and chassis cooling capability
Contact Sales for ordering numbers and switch configuration options.
The key ordering-time choices are primarily about fabric behavior and integration fit:
- Network switch configuration: determines port speed/port count presentation and how you partition traffic across the chassis
- PCIe Gen4 switch configuration: determines lane allocation between uplinks, GPU feed, and any downstream endpoints
- Slot profile alignment: the module aligns to a switch-slot profile (14.4.15), which drives backplane compatibility and role in the chassis
- Power threshold selection: must match chassis thermal headroom while switching at high throughput and running GPU workloads
Contact Sales for ordering numbers and available configuration options.
Choose WOLF-167S when you need to combine system switching and GPU compute in a single 3U VPX slot.
This module is typically selected when:
- Slot count is limited and you want to avoid using separate Ethernet switch, PCIe switch, and GPU payload modules
- You need a system fabric switch that can also provide embedded AI/HPC processing
- You want to route PCIe and Ethernet fabrics while maintaining a dedicated data path to the GPU
- Your platform benefits from a lower-power RTX 2000-class Blackwell GPU instead of a higher-power RTX 5000-class switch module
If your system requires independent scaling of switching and compute, separate switch and GPU modules may be the better architecture.
The integrated 200GbE network switch allows WOLF-167S to act as a chassis fabric switch while also providing onboard GPU compute.
It is most useful when:
- Multiple payload modules need high-bandwidth Ethernet connectivity across the backplane
- You need up to 8 ports at up to 25GbE plus a 1000BASE-T port
- You want to consolidate Ethernet switching and GPU acceleration in one slot
- Your system needs Time-Sensitive Networking support for deterministic traffic behavior
Treat WOLF-167S as both a switch resource and a compute resource when planning the system architecture.
Confirm these early to avoid backplane and thermal integration issues:
- Slot profile: SOSA aligned switch slot profile 14.4.15
- Ethernet mapping: confirm how the 8x 25GbE ports and 1000BASE-T port map into the chassis fabric
- PCIe topology: confirm PCIe Gen4 switch routing, GPU feed, uplinks, and any P1/P2 lane allocation assumptions
- GPU data path: confirm the PCIe x8 path to the RTX 2000 Blackwell GPU meets workload requirements
- System management: confirm IPMI integration, monitoring, and threshold behavior
- Power and cooling: confirm the chassis supports the selected 75W to 140W power configuration
This module is easiest to integrate when the chassis was already designed around a SOSA switch-slot role.
The key ordering-time choices are mainly about fabric behavior, thermal limits, and configuration control:
- Network configuration: determines port speed, port count, and how Ethernet traffic is partitioned across the chassis
- PCIe switch configuration: determines lane allocation between uplinks, GPU feed, and downstream endpoints
- Power threshold: choose a 75W to 140W setting that matches chassis cooling and mission performance needs
- Slot profile alignment: confirm the system supports the SOSA 14.4.15 switch profile
- Conformal coating: supports environmental robustness for deployed systems
- Variant locking: supports configuration control for qualified or repeat-build programs
Contact Sales for the latest ordering numbers and available configuration options.
