Space payloads are generating more data than ever, and increasingly, that data needs to be processed before it reaches the ground. Imaging, synthetic aperture radar, RF intelligence, and autonomous orchestration payloads can all benefit from moving substantial compute capability directly onto the spacecraft.
That is the role of the new Xiphos Q-PLC1 Payload Computer: a high-performance, space-ready computing platform designed for demanding edge-processing applications on SmallSat and other spacecraft platforms. Built around the Xiphos Q9 processor SOM and an AMD Versal AI Core adaptive SoC, the Q-PLC1 brings together general-purpose processing, real-time processing, programmable logic, AI acceleration, high-speed interfaces, and substantial onboard storage in one system.
At the center of the Q-PLC1 is the Versal AI Core VC2602 with its 152 AI Engine-ML tiles providing up to 101 TOPS and 8 TFLOPs of processing performance. These resources sit alongside dual Arm Cortex-A72 application processors, dual Cortex-R5F real-time processors, and a large programmable-logic fabric.
That heterogeneous architecture matters because spacecraft payload processing rarely consists of a single type of workload. An application might need software running under Linux, deterministic real-time control, highly parallel FPGA processing, and AI inference—all within the same payload computer.
The Q-PLC1 is intended to provide those capabilities without requiring separate computing platforms for each function.
Compute performance only matters if data can get to and from the processor fast enough. The Q-PLC1 therefore includes flexible PCIe connectivity, multiple 10 Gigabit Ethernet interfaces, and reconfigurable GPIO and LVDS interfaces for connecting high-speed peripherals, sensors, and optical payloads.
It also incorporates 16 GB of LPDDR4 DRAM and two independently powered 1 Terabyte Solid State Drives (SSD), giving payload developers significant local memory and storage for processing and buffering high-volume data.
Putting a high-performance processor in space involves considerably more than selecting a powerful chip.
Xiphos also provides a Yocto Linux BSP, standard Linux C/C++ development support, kernel drivers, and compatibility with standard Xilinx logic-development tools. Existing Linux applications can therefore be ported to the platform while FPGA developers can work within familiar Xilinx development environments.
In practical terms, Xiphos takes care of space radiation and environmental qualification, BSP, and Linux so development teams can focus on their mission software, AI models, and logic deployment.
That can be particularly important for organizations that want access to modern heterogeneous compute architectures without becoming experts in every aspect of turning commercial processing technology into a spacecraft-ready computing platform.
Another advantage of the Q-PLC1 is that a mission does not necessarily need an AI workload on day one to benefit from the robust Versal-based high-performance architecture. Its combination of CPUs, programmable logic, AI engines, high-speed connectivity, and storage provides a hardware baseline capable of supporting a variety of payload types and evolving processing requirements.
That means the same underlying computing platform can potentially support conventional payload processing today while providing the resources needed for more sophisticated onboard AI and autonomous processing later.
For spacecraft developers trying to move more processing to the edge, the Q-PLC1 represents a significant step up in available onboard compute: 101 TOPS of AI acceleration combined with FPGA, CPU, real-time processing, storage, and high-speed I/O in a payload computer designed specifically for space.