WOLF-153L
Overview
HPEC with NVIDIA Ada RTX 5000 and ConnectX-7
The VPX3U-AD5000E-CX7 HPEC module includes an NVIDIA RTX 5000 Ada embedded GPU and a ConnectX SmartNIC. The NVIDIA RTX 5000 Ada embedded GPU provides the advanced processing capabilities for high performance embedded computing (HPEC) and artificial intelligence (AI) processing. The ConnectX-7 provides the Ethernet and PCIe connectivity needed to move large datasets efficiently.
The NVIDIA Ada architecture includes CUDA cores for HPEC, 4th generation Tensor cores for AI computations, and 3rd generation Ray Tracing (RT) cores for visually accurate rendering. The Ada GPU uses a new TSMC 4N NVIDIA Custom Manufacturing Process which increases efficiency. The denser Ada GPUs have more CUDA and Tensor cores operating at higher clock frequencies at the same power, delivering significantly more performance per watt compared to WOLF’s previous generation product. For example, comparing the 144L to the 153L, the 153L can achieve two to three times the GFLOPS/W (depending on power and temperature limits).
The NVIDIA ConnectX-7 SmartNIC provides PCIe and Ethernet connectivity. ConnectX-7 is ideal for the high-speed, secure, data transfer capabilities required for data-heavy tasks such as sensor data processing and other C5ISR tasks. The ConnectX-7 also provides support for RDMA over Converged Ethernet (RoCE), enabling the fastest method for transferring data across the network to the GPU.
Block Diagram
The block diagram illustrates the RTX 5000 Ada GPU integrated with a ConnectX-7 SmartNIC, highlighting the high-speed network-to-PCIe data path used for rapid ingest and distribution. It shows how Ethernet/RoCE traffic is terminated on the SmartNIC and delivered into the GPU compute pipeline through the module’s PCIe fabric interface.
Features
WOLF-153L is a rugged 3U VPX module that integrates an NVIDIA RTX 5000 Ada GPU with an NVIDIA ConnectX-7 SmartNIC to support embedded AI and HPEC processing with high-speed network connectivity. It provides up to 100GbE Ethernet and a configurable PCIe switch, with switching offloaded from the host CPU to ConnectX. The module includes 16 GB GDDR6 RAM with ECC support, supports a configurable 100W to 140W power envelope, and aligns to SOSA slot profile 14.6.11 or 14.6.13.
GPU Features
- Ada GPU parallel processing:
- CUDA Toolkit 12, CUDA Compute capability 8.9
- CUDA-X AI and CUDA-X HPEC libraries
- OpenCL 3.0, DirectX 12 Ultimate, OpenGL 4.6, OpenGL ES 3.2, Vulkan 1.2
- 184 Tensor Cores (4th Gen)
- 46 Ray Tracing cores (3rd Gen)
- NVENC (8th Gen) and NVDEC (5th Gen) with up to 8K video encoding and hardware decoding support
Connectivity / System Management
- IPMI system management
- PCIe configurable switch
- Switching is offloaded from CPU to the ConnectX
- Configurable 100GbE Ethernet switch
- ConnectX-7 provides hardware offload capabilities:
- NVIDIA ASAP2
- NVIDIA GPUDirect RDMA and RoCE support
- PCIe daisy chain support
- 40/100GBASE-KR4 data, 10GBASE-KR control on backplane
- Linux and Windows drivers
- NVIDIA driver support requires one of the following host CPUs: Intel H/HX/P/PX/S Class, AMD H/HS Class
Mechanical / Open Systems Architecture
- High level of ruggedization:
- Rugged conduction cooled
- Operating temp: CC: - 40°C to +70°C standard, operational to +85°C
- Vibration (sine wave): 10G peak, 5 - * Shock: 40G peak
- Dimensions: 160mm x 100mm x 25.4mm
- Weight (approximately): 1.3kg
- ANSI/VITA 48, 65 (VPX- REDI, OpenVPX)
- SOSA aligned 14.6.11 or 14.6.13 slot profile
Specifications
Key configuration and performance parameters for WOLF-153L 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 series of common order codes for the VPX3U-AD5000E-CX7 module. The asterisks denote characters of the part number that are defined based on common configuration options.
Some configuration options for this module include:
- PCIe Configuration Options
- Variant Locked
- Default Power Threshold
- Ethernet configuration
- P2 or P2B de-populated
- Conformal Coatings
Contact Sales for the latest Ordering Numbers and available options.
Representative ordering numbers
| Ordering Number | Description |
|---|---|
| 153L33-F***-***VPX3vA0 | 3U VPX, Conduction Cooled, NVIDIA Ada RTX5000, ConnectX-7, SOSA Payload profile with P2 depopulated |
| 153L33-F***-***VPX3vA0 | 3U VPX, Conduction Cooled, NVIDIA Ada RTX5000, ConnectX-7, SOSA Payload profile with P2B depopulated |
Related Product:
- 153833-F005-***VPX3vA0 3U VPX, Conduction Cooled, 1", NVIDIA Ada RTX5000, PCIe switch, SOSA Payload profile with P2 depopulated
Related Products
- NVIDIA RTX 5000 Ada (AD103)
- 9728 CUDA Cores, 304 Tensor Cores
- 16 GB GDDR6 VRAM
- PCIe Gen 4
- OpenVPX
- Module power 80–140 W
- NVIDIA RTX 5000 Ada (AD103)
- 9728 CUDA Cores, 304 Tensor Cores
- 16 GB GDDR6 VRAM
- PCIe Gen 4
- SOSA Payload, SOSA Legacy
- Module power 60–140 W
- NVIDIA RTX 5000 Ada (AD103)
- 9728 CUDA Cores, 304 Tensor Cores
- 16 GB GDDR6 VRAM
- PCIe Gen 4
- OpenVPX
- Module power 70–140 W
FAQ
Choose WOLF-153L when your platform is constrained by data movement, not just compute.
WOLF-153L integrates an Ada RTX 5000 GPU with ConnectX-7, which is intended for:
- High-rate ingest or distribution over Ethernet fabrics
- Architectures that benefit from RoCE and GPUDirect RDMA to reduce CPU copies/overhead
- Deployments where networking offload and security features are part of the platform requirement
If your workload is mostly local to the chassis and networking is modest, a GPU-only module may be simpler.
It is designed to reduce CPU involvement and extra memory copies for network-to-GPU data movement.
It matters most when:
- You have large streaming datasets (sensor ingest, distributed processing)
- You have multi-node Ethernet fabrics where latency and CPU overhead become limiting
- You want to keep host CPU cycles focused on mission apps rather than packet movement
In practice, it influences how you architect your ingest pipeline (buffering, CPU load, and end-to-end latency).
Confirm these early to avoid late-stage backplane changes:
- Slot profile alignment: SOSA 14.6.11 or 14.6.13 (P2/P2B depopulation impacts routing)
- Ethernet electrical interfaces: ensure your backplane supports the intended KR plane mapping
- PCIe fabric expectations: confirm how the module’s PCIe and switching behavior aligns with chassis topology
Treat the profile choice as both an electrical routing decision and a mechanical compatibility decision.
Key tradeoffs are typically integration and sustainment decisions:
- P2 vs P2B depopulation (14.6.11 vs 14.6.13): drives backplane compatibility
- Ethernet and PCIe configuration: influences port/lane presentation and throughput
- Default power threshold (100W-140W envelope): must match chassis cooling
- Program controls: conformal coating and variant locking for deployed configuration control
Contact Sales for the latest ordering numbers and available configuration options.