- Technology
- Computer Science
- Distributed Systems
- High Performance Computing
- Parallel Computing
- GPU Programming
- CUDA
- Technology
- IT & Networking
- Computer Networking
- Network Automation & Programmability
- Network Programming
- DPDK
- Technology
- Computer Science
- Distributed Systems
- High Performance Computing
- Parallel Computing
- GPU Computing
1.2 Tb/s Packet Processing of Radio Telescope Data on GPU
DPDK Project via YouTube
AI, Data Science & Cloud Certificates from Google, IBM & Meta
Free courses from frontend to fullstack and AI
Overview
Google, IBM & Meta Certificates – 40% Off
One plan covers every Professional Certificate on Coursera.
Unlock All Certificates
Learn about groundbreaking packet processing techniques for radio telescope data in this technical presentation from ASTRON's John Romein. Discover how the NVIDIA Grace Hopper Superchip enables unprecedented 1.2 Tb/s data processing rates by eliminating PCIe bottlenecks, allowing a single GPU to connect with multiple high-speed NICs. Explore the implementation details of streaming 1196Gb/s of Ethernet packets, including buffering, channelization, and signal correlation on GPUs. Gain insights into key technological advances like GPU support in DPDK, packet splitting capabilities, direct DMA payload transfers, and efficient pktmbuf cache utilization. Examine performance results and energy optimization strategies that achieve significant power reductions for both CPU and GPU operations, ultimately enabling more powerful radio telescope implementations.
Syllabus
1.2 Tb/S Packet Processing of Radio Telescope Data on a GPU - John Romein, ASTRON
Taught by
DPDK Project