Speaker
Description
Bandwidth and storage limitations are a key bottleneck for many CMS physics measurements and searches. To mitigate these constraints, CMS has developed a set of techniques that increase the number of events written to disk while maintaining physics performance. These strategies remain an active area of development and are being further optimized for Phase-2.
One such technique, RawPrime, reduces event size by replacing raw strip-detector data with low-level reconstructed objects, ie. hit positions and charges. RawPrime was successfully deployed online during Run 3 PbPb collisions, yielding an event-size reduction of about a factor of two relative to the standard full raw format. Building on this, RawSecond has recently been deployed online for testing. By refining the data-format and encoding used to store the strip information, RawSecond provides an additional ~20% reduction in strip data volume while improving precision.
A similar strategy is under development for Phase-2 to reduce raw data volumes in the highest-granularity subsystems, in particular the High-Granularity Calorimeter (HGCal) and the Inner Tracker.
In this talk we present the RawPrime concept, the improvements demonstrated by RawSecond using 2025 data, and the current status of ongoing studies to apply these concepts to Phase-2 detectors.
This talk will present the RawPrime and RawSecond techniques, the latest results obtained with 2025 data, and the current status of studies to extend this approach to Phase-2 detectors.