Stelline

DEPLOYMENT TARGETS

HW-01 / OBSERVATORY Allen Telescope Array Real-time radio interferometer array powered by our on-site NVIDIA RTX 6000 Blackwell cluster.

HW-02 / DEVELOPER KIT DGX Spark Compact AI supercomputer for software development and edge deployment.

SOFTWARE STACK

SW-01 / ORCHESTRATION Nexus Compute cluster orchestration and distributed execution of real-time sky observations.

SW-02 / INTERFACE CyberEther High-performance GPU-accelerated signal processing and visualization platform. Combines a familiar flowgraph-based interface with state-of-the-art heterogeneous acceleration and a unified operational layer for standard radio astronomy tools.

Stelline Telescope networking and radio astronomy tools.

SW-04 / NETWORKING DAQIRI GPU-accelerated packet ingest and processing

SW-05 / STORAGE VFD-GDS GPUDirect Storage driver for HDF5 assets

Blade GPU-accelerated DSP blocks

Correlator

Beamformer

Channelizer

Detector

CUDA

GPU Direct

Vulkan

TensorRT

SHOWCASES

DEM-001

CyberEther ingests live Allen Telescope Array data over 100 GbE, performs real-time correlation, and visualizes the resulting baselines

Developer Kit

DGX Spark

Compact AI supercomputer powered by the NVIDIA GB10 Grace Blackwell Superchip and built-in ConnectX networking, DGX Spark provides a compact, high-performance platform for developing and validating Stelline.

Getting Started Guide COMING SOON

Coyote

Each Coyote-class server combines four NVIDIA RTX 6000 Blackwell GPUs with up to 800 Gbps of fiber ingest and up to 256 TB of high-speed NVMe storage for real-time Allen Telescope Array signal processing.

Documentation COMING SOON