MxTasks: How to Make Efficient Synchronization and Prefetching Easy

Jens Teubner (TU Dortmund)

The hardware environment has changed rapidly in recent years: Many cores, multiple sockets, and large amounts of main memory have become a commodity. To benefit from these highly parallel systems, the software has to be adapted. Sophisticated latch-free data structures and algorithms are often meant to address the situation. But they are cumbersome to develop and may still not provide the desired scalability. As a remedy, we present MxTasking, a task-based framework that assists the design of latch-free and parallel data structures. MxTasking eases the information exchange between applications and the operating system, resulting in novel opportunities to manage resources in a truly hardware- and application-conscious way.

Jens Teubner heads the Databases and Information Systems Group at TU Dortmund University in Dortmund, Germany. He works on database implementation techniques for modern computing hardware (including such with new memory architectures or high degrees of parallelism; heterogeneous architectures; and novel networked architectures). Jens Teubner has previously worked at ETH Zurich, where he headed the Avalance project on FPGA-accelerated data processing. Jens holds a PhD in Computer Science from TU München, Germany (on XML database processing) and a Master’s degree in physics from U Konstanz, Germany.

Slides