view article
Figure 5
Transfer throughput from CPU and RNIC to GPU memory. The transfers from pinned memory are more efficient than from virtual memory. Using RDMA technology, full bandwidth was achieved from 1 KB. ZeroMQ transfers, which are used by the CrystFEL online reduction tool, are capped under 10 Gb s−1.

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775
Volume 27| Part 5| September 2020| Pages 1297-1306
https://doi.org/10.1107/S1600577520008140
Follow J. Synchrotron Rad.
Sign up for e-alerts
E-alerts
Follow J. Synchrotron Rad. on Twitter
Twitter
Follow us on facebook
Facebook
Sign up for RSS feeds
RSS