Tactical edge computing requires more than high-performance processing. Mission systems must be able to ingest data from multiple sensors, move data efficiently between system resources, execute compute-intensive workloads with minimal latency, and distribute actionable information without relying solely on remote or centralized processing infrastructure.
For sensor-driven mission systems, the challenge is no longer simply providing enough processing performance; it’s integrating the compute, data movement, sensor I/O, and networking resources required to turn multiple streams of raw sensor data into actionable information within tight latency, size, weight, power, and thermal constraints.
This places greater demands on the underlying embedded computing architecture. Data must move quickly from sensors to processing resources, compute-intensive workloads must execute locally, and results must be distributed across the platform without creating bottlenecks or relying on continuous access to remote or centralized computing infrastructure. As AI inference and sensor-processing workloads continue to increase in complexity, system designers need scalable computing architectures that can accommodate higher levels of processing while remaining practical for deployment in rugged, SWaP-constrained environments.
3U VPX general-purpose computing modules provide a modular foundation for addressing these requirements at the edge. Based on the VITA 46 VPX and VITA 65 OpenVPX standards, 3U VPX combines high-speed serial connectivity, standardized module-to-backplane interfaces, and a compact rugged form factor that allows CPU processing, GPU acceleration, high-speed networking, sensor I/O, and storage resources to operate within a common system architecture.
The Condor T5-IOX 3U VPX takes advantage of this architecture by bringing the NVIDIA Jetson IGX Thor T5000 system-on-module (SoM) into a rugged 3U OpenVPX single-board computer (SBC). Powered by the NVIDIA® Jetson™ IGX Thor T5000, the SBC combines NVIDIA Blackwell GPU acceleration, a 14-core Arm® Neoverse™ V3AE CPU, up to 128 GB of LPDDR5X memory, high-speed networking, video ingest, and system I/O within a compact 3U VPX SBC. With shared system memory up to 128 GB of LPDDR5X and up to 273 GB/s of memory bandwidth, the Jetson IGX Thor provides the capacity and throughput required for increasingly data-intensive edge workloads.
NVIDIA Jetson™ IGX Thor’s unified memory architecture allows the CPU and Blackwell GPU to access the same physical memory pool, reducing the need to repeatedly copy large datasets between separate CPU and GPU memory spaces. For mission payloads processing multiple high-resolution sensor streams, AI models, imagery, and other real-time data, this can help streamline the processing pipeline and reduce data-transfer overhead as information moves between CPU- and GPU-accelerated workloads. The increased 128 GB memory capacity also enables the Condor T5-IOX to support larger AI models, larger datasets, and multiple concurrent processing workloads directly within the mission payload.
The NVIDIA Jetson IGX T5000 delivers up to 2,070 FP4 TFLOPS of sparse AI compute performance, bringing a significant level of AI acceleration into the Condor™ T5-IOX 3U VPX single-board computer. The increased AI compute density gives system designers greater flexibility when consolidating multiple processing functions within a constrained 3U VPX architecture. As sensor counts, resolutions, and AI model complexity increase, the T5-IOX provides additional compute headroom to process multiple data streams and execute concurrent accelerated workloads without immediately requiring additional GPU cards or external processing resources.
The 3U VPX single board computer allows compute-intensive workloads such as AI inference, computer vision, multi-sensor fusion, video/signal processing, and real-time analytics to be executed directly within the mission payload. Rather than sending large volumes of raw sensor data to a remote processing resource, the Condor T5-IOX can help process information closer to where it is generated. This reduces unnecessary data movement, helps minimize latency, and allows higher-level information to be distributed throughout the system. High-speed Ethernet, PCIe Gen 5, and 12G-SDI interfaces provide multiple paths for moving data into and out of the processor, while the Blackwell GPU provides the acceleration required to analyze that information in real time.
Integrating these capabilities onto a single SBC can simplify the processing architecture by reducing the number of discrete resources required to acquire, move, and process sensor data. For system designers working within a fixed slot, power, and cooling envelopes, this processing density creates opportunities to reduce slot count or reserve additional VPX slots for application-specific resources such as FPGA processing, RF interfaces, storage, networking, or additional GPU acceleration.
The Condor T5-IOX 3U VPX was developed in alignment with the SOSA® Technical Standard, allowing its NVIDIA® Jetson™ IGX Thor processing capabilities to be integrated within open, modular system architectures. SOSA builds upon OpenVPX and other open standards to further define common hardware, electrical, mechanical, and system interfaces for sensor and mission-computing systems. Rather than building a tightly coupled architecture around proprietary interfaces, the SOSA architecture supports a modular approach in which appropriately matched plug-in cards, backplanes, and system resources can be integrated around defined interfaces and profiles. The Condor T5-IOX serves as a high-performance AI and general-purpose computing resource within a larger modular VPX system, operating alongside complementary plug-in cards for GPU or FPGA acceleration, networking, sensor acquisition, storage, and other specialized processing functions.
The Condor T5-IOX supports four 12G-SDI video inputs providing high-bandwidth connectivity for integrating multiple high-resolution video sensors directly with the SBC. Each 12G-SDI interface can transport UHD/4K video at up to 60 frames per second over a single connection, enabling the Condor T5-IOX to ingest multiple high-resolution sensor streams simultaneously. For sensor-rich systems incorporating EO/IR cameras, situational awareness sensors, or other digital video sources, this provides a direct path for moving large volumes of visual data into the NVIDIA® Jetson™ IGX Thor processor for real-time processing.
Combining multi-channel 12G-SDI sensor ingest with NVIDIA Blackwell GPU acceleration enables the Condor T5-IOX to perform more of the sensor-processing pipeline within a single 3U VPX slot. Incoming video can be processed using computer vision and AI inference algorithms, correlated with other sensor data, and distributed through the card’s high-speed Ethernet or display interfaces. This helps system integrators reduce dependence on separate video-capture and processing hardware while supporting applications such as multi-camera processing, EO/IR analysis, object detection and classification, image enhancement, sensor fusion, and real-time situational awareness.
The Condor T5-IOX supports four 25GbE (25GBASE-KR) interfaces, providing up to 100 Gb/s of aggregate Ethernet bandwidth. These interfaces create high-bandwidth pathways for exchanging sensor and application data with compatible network switches and other processing resources across the VPX backplane.
Combined with PCI Express Gen 5 and RDMA technologies, the Condor T5-IOX is designed to support high-throughput data movement between sensors and processing resources while minimizing potential system bottlenecks. This creates an important relationship between compute performance and data movement. The Blackwell GPU provides the processing capability required for advanced AI and sensor workloads, while the high-speed VPX interfaces help ensure that data can reach those processing resources efficiently.
The NVIDIA Blackwell GPU within the Jetson IGX T5000 SoM introduces Multi-Instance GPU (MIG) technology to the Jetson Thor platform, allowing GPU resources to be partitioned into isolated instances with dedicated compute, cache, and memory bandwidth. This enables multiple GPU-accelerated applications to operate concurrently with greater performance predictability and reduced resource contention. For sensor-rich edge systems, MIG can help separate workloads such as AI inference and computer vision from graphics or visualization tasks, allowing each application to access dedicated GPU resources rather than competing for the entire GPU. This is particularly valuable for mixed-workload environments where latency-sensitive processing must operate alongside other accelerated applications.
For the broader system architecture, MIG allows the Condor T5-IOX to make more efficient use of its Blackwell GPU while supporting workload consolidation within a single 3U VPX slot. Instead of assigning separate GPU hardware to different processing functions, system integrators can partition available GPU resources and run workloads in parallel with hardware-level isolation. Combined with the Condor T5-IOX’s high-speed 25GbE connectivity, PCIe Gen 5, video interfaces, and 128 GB of unified memory, MIG helps turn the SBC into a flexible processing platform capable of supporting multiple independent functions within a larger modular 3U VPX system.
As mission systems incorporate higher-resolution sensors, increasingly sophisticated AI models, and greater volumes of sensor data, edge computing architectures must evolve to support both the processing performance and data movement these workloads demand. The Condor™ T5-IOX single board computer addresses these requirements by combining NVIDIA® Jetson™ IGX Thor processing, Blackwell GPU acceleration, high-bandwidth sensor connectivity, high-speed Ethernet bandwidth, and flexible GPU resource allocation within a compact 3U VPX platform. The result is a high-density computing resource capable of ingesting, moving, processing, and distributing sensor data directly at the tactical edge.
More importantly, the 3U VPX module provides system integrators with a scalable building block for the next generation of modular mission computing. Within an OpenVPX architecture, the SBC can operate alongside complementary 3U VPX plug-in cards for additional GPU processing, sensor acquisition, networking, storage, and other specialized functions, allowing capabilities to be configured around application requirements. By bringing advanced AI processing closer to the sensor while maintaining the flexibility of an open, modular architecture, the Condor T5-IOX provides a foundation for developing high-performance, interoperable, and upgradeable edge computing systems built to support evolving mission requirements.
Enabling the DoD’s AI-First Strategy Through MOSA and SOSA Aligned Architectures
EIZO Rugged Solutions is a pioneer in providing video, graphics, encoding, and recording solutions for over three decades.
