High-end sensor fusion, real-time AI inference and advanced visualisation at the tactical edge all demand more than consumer GPUs can deliver. WOLF’s family of 6U VPX graphics cards couples dual embedded NVIDIA GPUs with high-bandwidth PCIe switching in a conduction-cooled module that drops straight into OpenVPX or SOSA-aligned payload slots. The result: desktop-class performance, military-grade reliability—and local sales, service and support from Metromatics in Australia & New Zealand.
Why Choose WOLF 6U VPX Graphics Cards?
Extreme parallel processing – up to 20 992 CUDA cores in a single slot for AI, HPEC and graphics workloads.
Latest NVIDIA architectures – options for Blackwell (GB203), Ada (AD103) or Turing (TU104) GPUs to match program timelines and power budgets.
High-speed I/O – PCIe Gen5 or Gen4 switches, GPUDirect RDMA and up to eight DisplayPort/HDMI outputs for sensor ingest and display.
Built for harsh environments – conduction-, air- or liquid-cooled variants qualify to MIL-STD-810 and RTCA DO-160.
SOSA / OpenVPX slot compatibility – factory pin-map options ensure rapid tech insertion and upgrade paths. Want to learn why OpenVPX is so popular with Defence? View this article.
Local expertise – Metromatics provides pre-sales engineering, repair logistics and long-term spares support across ANZ.
6U VPX Graphics Models
VPX6U-AD5000E-DUAL-VO (WOLF-2538) – Dual Ada
| Feature | Specification |
|---|---|
| GPUs | 2 × NVIDIA RTX 5000 Ada (AD103) |
| Parallel cores | 9 728 CUDA, 304 Tensor, 76 RT per GPU |
| Memory | 16 GB GDDR6 ECC per GPU (32 GB total) |
| Switch | PCIe Gen4, configurable |
| Video outputs | Up to 8 × DisplayPort 1.4 / HDMI |
| Typical use | ISR sensor fusion, HPEC, real-time 3D rendering |
Dual Ada GPUs deliver up to 32 TFLOPS FP32 while staying within VPX power envelopes—ideal for retrofit programmes that need maximum performance per watt.
VPX6U-RTX5000E-DUAL-VO (WOLF-2348) – Dual Turing
| Feature | Specification |
|---|---|
| GPUs | 2 × NVIDIA Quadro RTX 5000 Turing (TU104) |
| Parallel cores | 6 144 CUDA, 768 Tensor, 96 RT per GPU |
| Memory | 16 GB GDDR6 ECC per GPU (32 GB total) |
| Compute | 22 TFLOPS FP32 peak |
| Video outputs | 8 × DisplayPort / HDMI |
| Typical use | Legacy platform upgrades, AI inference, radar & EW acceleration |
Turing architecture offers proven reliability and a large software ecosystem for AI frameworks, making it a cost-effective step-up from Pascal-era cards.
VPX6U-BW5000E-DUAL-VO (WOLF-2638) – Dual Blackwell
| Feature | Specification |
|---|---|
| GPUs | 2 × NVIDIA RTX 5000 Blackwell (GB203) |
| Parallel cores | 10 496 CUDA, 320 Tensor per GPU |
| Memory | 24 GB GDDR7 ECC per GPU (48 GB total) |
| Switch | PCIe Gen5 with IPMI management |
| Video outputs | Up to 8 × DisplayPort 2.1 / HDMI 2.1 |
| Typical use | Next-gen autonomous systems, 8 K video encode/decode, large-scale model inference |
Blackwell brings fifth-gen Tensor cores, AV1 codec support and 55 % more memory bandwidth—future-proofing AI and video pipelines for the next decade.
Ready to Evaluate?
Metromatics can provide integration guidance to fast-track your 6U VPX design.
→ Contact our rugged-GPU team:
✉ sales@metromatics.com.au |
☎ Brisbane: +61 7 3868-4255 I Melbourne: +61 3 9872 4592 I Sydney: +61 2 9460 4355
Talk to us about configuration options, lead-times and long-term sustainment for your programme.
Reviews
There are no reviews yet.