Speculative Dynamic Reconfiguration and Table Prefetching Using Query Look-Ahead in the ReProVide Near-Data-Processing System

Abstract

FPGAs are promising target architectures for hardware acceleration of database query processing, as they combine the performance of hardware with the programmability of software. In particular, they are partially reconfigurable at runtime, which allows for the runtime adaptation to a variety of queries. However, reconfiguration costs some time, and a region of the FPGA is not available for computations during its reconfiguration. Techniques to avoid or at least hide the reconfiguration latencies can improve the overall performance. This paper presents optimizations based on query look-ahead, which follows the idea of exploiting knowledge about subsequent queries for scheduling the reconfigurations such that their overhead is minimized. We evaluate our optimizations with a calibrated model for various parameter settings. Improvements in execution time can be “calculated” even if only being able to look one query ahead.

COPRAO: A Rule-Based Capability-Aware Query Optimizer for an FPGA-Based Reconfigurable Near-Data Processor

“…In a situation where not enough PRs are available for the given query, a cost model is used to direct the selection of operations for the RPU. Often, if the DBMS knows about upcoming queries (which is the case if query sequences have been extracted from a query log), hints about these queries are sent to the RPU to eliminate reconfiguration overhead and table-fetching time from the execution time, as discussed in [4]. The working of ReProVide has been demonstrated in [5].…”