S-A1760 Venus Space GPGPU
Aitech Systems
The S-A1760 Venus is a radiation-characterized small form factor space GPGPU based on the NVIDIA Jetson TX2i with Pascal architecture, offering 256 CUDA cores, 1 TFLOPS, and 8 GB LPDDR4 ECC memory in a sub-1 kg package for NEO and LEO video and signal processing.
Technical specifications
- GPU
- NVIDIA Jetson TX2i, Pascal Architecture
- CUDA Cores
- 256
- CPU
- NVIDIA Denver 2 Dual-Core ARM + Cortex A57 Quad-Core ARM
- Performance
- 1 TFLOPS
- Energy Efficiency
- 60 GFLOPS/W
- Memory
- 8 GB LPDDR4 with ECC
- Storage
- 32 GB SSD
- I/O
- Gigabit Ethernet, UART Serial, USB 2.0, Discretes, CANbus
- Form Factor
- SFF 129 mm × 129 mm
- Weight
- < 1 kg
- Radiation
- Series 300 radiation characterization
- Software
- CUDA, OpenGL, OpenGL ES, EGL
About
The S-A1760 Venus is Aitech’s smallest and most power-efficient space-rated GPGPU system, designed specifically for short-duration spaceflight, near-Earth orbit, and LEO satellite applications requiring compact high-performance compute. It integrates the NVIDIA Jetson TX2i module featuring Pascal GPU architecture with 256 CUDA cores delivering 1 TFLOPS of compute at 60 GFLOPS per watt efficiency.
The system packs 8 GB of LPDDR4 memory with ECC protection and 32 GB of SSD storage into a 129 mm square footprint weighing under 1 kg. The heterogeneous CPU cluster combines NVIDIA Denver 2 dual-core ARM and Cortex A57 quad-core ARM processors for flexible task distribution. Hardware H.264/H.265 encoding is included for video compression workloads.
Connectivity encompasses Gigabit Ethernet, UART serial, USB 2.0, discrete I/O signals, and CANbus for spacecraft bus integration. The S-A1760 Venus carries Series 300 radiation characterization and is actively used in distributed processing architectures for satellite imaging, autonomous navigation, and signal intelligence applications.
Documentation
No public datasheet yet — request the datasheet / ICD from the supplier.
