In the context of the CERN openlab collaboration, Eric Sedlar (Oracle) will give a computing colloquium at CERN on Monday 30 May, in the Marie Curie auditorium (40-S2-C01) at 11:00.

His talk will focus on heterogeneous computing.

For more information about the colloquium and opportunities to meet Oracle executives prior to the talk: https://indico.cern.ch/conferenceDisplay.py?confId=140879

Mélissa Gaillard, CERN openlab

=======================================================================

Description:

There are two related issues that provide the greatest challenge to computer science today: on the software side, the need for accessible parallelism; on the hardware side, the need for power-efficient computing.  There are many results that show that domain-specific software infrastructure with higher levels of abstraction (languages & runtimes like SQL and OpenGL) are much better at extracting parallelism than more general infrastructure, and that domain-specific hardware (like ASICs) are orders of magnitude more power-efficient than general purpose processors.  While trading away generality for any purpose seems foreign to most computer scientists, there are strong indications that suggest that this is the most fruitful line of research to better address the needs for parallelism and power-efficiency.  It is clear that merely providing programmers direct access to threads & locks will not result in very much good parallel software or in good usage of hardware resources.  The talk will draw primarily on examples from database systems.

Eric will also give a brief overview of Oracle Labs and its new mission prior to the talk.

About the speaker:

Eric Sedlar is Technical Director of Oracle Labs (formerly Sun Labs).  This position entails figuring out how to transfer research results from Labs research into Oracle products & services in a way that has the best impact on Oracle's business, and which new research projects are most likely to have successful impact at Oracle. 

His own research interests are in application evolution, XML (and more generally schema-later data design), and acceleration of database operations both via new hardware and using JIT compilation. He started efforts doing research inside the Oracle RDBMS product group in 2006, with two major thrusts: architecture-aware improvements for database processing, and schema-less enterprise application development. Previously, he led the effort for XML-native storage inside Oracle, starting with Oracle 9iR2. Eric has held various architecture and development management positions at Oracle since starting there in 1990. He holds over 50 patents, and has served on standards organizations for Oracle in the W3C and IETF. He co-authored the Best Paper at SIGMOD 2010 on architecture-sensitive search trees.