GPU-Compression in UltraGrid Platform: Enabling Interactive Medical Applications on Commodity Academic Networks

The presentation focuses on employing massively parallel architectures of commodity GPU hardware (i.e., common graphics cards that users have in their desktop and laptop computers) to implement DXT compression and JPEG compression/decompression as a part of low-latency transmissions of high-definition (HD) and post-HD video over computer networks. Use of these compression schemes enables highly interactive video-based collaboration over networks with 1Gbps or less, while still achieving video quality that is almost undistinguisable from uncompressed video, that however requires 10Gbps networks that are still not available for vast majority of end users both in academia and healthcare. Wide adoption and low cost of consumer GPU also allows users to use very affordable hardware configuration. We will discuss application of this technology to healthcare for both research and practical purposes.



  • Petr Holub
  • Jiří Matela
  • Martin Pulec
  • Martin Šrom
  • Martin Jirman

Related documents