High-speed Data Acquisition

Our systems reliably capture high-volume data streams (e.g. UDP packets from FPGAs) at 100s of Gb/s directly into COTS server memory (e.g. 1460 Gb/s for the SKAO Pulsar Timing Processor).
Leveraging Direct Memory Access (DMA) to bypass conventional I/O bottlenecks, we offload data directly into CPU or GPU memory, dramatically reducing CPU load and optimizing system performance.
For seamless integration with other systems (e.g. VDIF recorder), our software supports flexible output transmission protocols including TCP, UDP, and RDMA.
Facilitating high-speed ‘corner turns’ between servers, we utilize cutting-edge RDMA technologies to ensure data is routed and processed with minimal latency (e.g. Channelizer for VLBA New Digital Architecture).
Our expertise extends to meticulously tuning and optimizing server configurations to guarantee the most robust and reliable data acquisition possible.

Capability