The High Throughput Computing Collaboration (HTCC) investigates the use of Intel technologies in trigger and data acquisition (TDAQ) systems. The HTCC will study the extent to which custom-made real-time triggers can be complemented by off-the-shelf Intel technology, the use of Intel OmniPath for high-bandwidth data acquisition and the acceleration of complex, CPU-intensive algorithms with Intel accelerator technology Xeon Phi and Xeon FPGA concept.
For data acquisition high-bandwidth local area networks are required, which can work well in unusually bursty traffic conditions in a many-to-many traffic. Using custom-made benchmark programs, we are evaluating what percentage of the nominal bandwidth can be realized using RapidIO under these challenging conditions.
In the first level of the LHC trigger systems hard real-time guarantees are required. We are investigating if the latency and realtime features of RapidIO are sufficient for potential use in the first level-trigger systems.
CERN openlab holds 2015 technical workshop |
High-throughput computing at the LHCb experiement |
Follow the CERN openlab Open Day online |