Compilers for Fast Data
Staff - Faculty of Informatics
Start date: 5 March 2015
End date: 6 March 2015
The Faculty of Informatics is pleased to announce a seminar given by Martin Hirzel
DATE: Thursday, March 5th 2015
PLACE: USI Lugano Campus, room A23, Red building (Via G. Buffi 13)
TIME: 15.30
ABSTRACT:
Recent trends, from social media to algorithmic trading to the internet of things, have made available a deluge of "fast data". Fast data is produced continuously, and those who quickly extract insights from it gain an edge. Unfortunately, with existing systems and languages, it is hard to write efficient fast-data applications. The challenge is bridging the gap between a high-level programming experience on the one hand, and low-level incremental and parallel algorithms on the other hand. This talk describes IBM's SPL language, which bridges the gap by allowing library writers to implement compiler extensions. For illustration, this talk covers two case studies: sliding-window aggregation [VLDB'15] and streaming spreadsheets [ECOOP'14 with enhancements]. Thanks to its unique design, SPL is both a commercial success and a vehicle for ongoing research at the boundary between compilers and fast data processing.
BIO:
Martin Hirzel is a Research Staff Member and a Manager at the IBM T.J.Watson Research Center. Martin's group at IBM is the Data Languages Science group. His group focuses on languages and compilers for web data. Most of Martin's recent publications are based on the Streams Processing Language (SPL). He led the SPL language design, wrote the language specification, and implemented the compiler front-end. Before working on SPL, Martin's research revolved around programming language design, garbage collection, and optimizations for Java. Martin received his PhD from the University of Colorado at Boulder in 2004; his thesis adviser was Amer Diwan.
HOST: Prof. Matthias Hauswirth
