Oriel — AI & technology worth noticing, every day.

Run GPU-Powered Black Hole Interferometry with NVIDIA cuPhoton

Web page · https://developer.nvidia.com/blog/run-gpu-powered-black-hole-interferometry-with-nvidia-cuphoton

Imaging black holes is challenging because they are distant and appear tiny in the sky—for example, M87*’s observed ring spans just 42 microarcseconds. The...

First seen by Oriel on .

AI summary

This tutorial explains GPU-based black hole interferometry with NVIDIA cuPhoton.xBeam, which correlates telescope recordings into visibilities and processed 10 TB 460x faster than an x86 CPU baseline.

  • xBeam ingests VDIF recordings, indexes them with .vdix sidecars, applies delay and fringe corrections, performs FFT channelization, and cross-multiplies station signals.
  • It outputs visibilities in DiFX’s SWIN format or xBeam’s xbeam-v1 format for downstream calibration and image reconstruction tools.
  • On an NVIDIA Vera Rubin NVL72 system with 64 GPUs, xBeam processed 10 TB in 70 seconds versus 31,992 seconds for the x86 CPU baseline.
  • GPUDirect Storage read bandwidth increased from 20 GB/s on 4 GPUs to 170 GB/s on 64 GPUs in the tested configuration.
  • xBeam uses delays supplied by the job and does not calculate station-position delays, trace black-hole light paths, or perform downstream imaging.

#cuphoton #gpu computing #radio interferometry #black holes

Generated by gpt-5.6-sol. Check the original for details.

Aliases

SourceExternal IDURL
NVIDIA Developer Bloghttps://developer.nvidia.com/blog/?p=125204https://developer.nvidia.com/blog/run-gpu-powered-black-hole-interferometry-with-nvidia-cuphoton/ Read discussion