To share or not to share vector registers?

Abstract Query execution techniques in database systems constantly adapt to novel hardware features achieve high query performance, particular for analytical queries. In recent years, vectorization based on the Single Instruction Multiple Data parallel paradigm has been established as a state-of-the-art approach increase single-query performance. However, since concurrent queries running often access same columns and perform set of vectorized operations, data accesses computations among different may be executed redundantly. Various have already proposed avoid such redundancy, ranging from scans via construction materialized views applying multiple optimization techniques. Continuing this line research, we investigate opportunity sharing vector registers concurrently scenarios paper. particular, our relies processing elements together within single register. As are going show, optimize can very beneficial single-threaded well multi-thread environments. Therefore, demonstrate feasibility applicability work strategy thus open up wide spectrum future research opportunities.