Keynote talk II

II- “Scalloc: From Relaxed Concurrent Data Structures to the Fastest Multicore-Scalable Memory Allocator“.

Christoph Kirsch University of Salzburg, Austria

Biography of Christoph Kirsch:
Christoph Kirsch is Professor at the Department of Computer Sciences of the University of Salzburg, Austria. He received his Dr.Ing. degree from Saarland University, Saarbrücken, Germany, in 1999 while at the Max Planck Institute for Computer Science. From 1999 to 2004 he worked as Postdoctoral Researcher at the Department of Electrical Engineering and Computer Sciences of the University of California, Berkeley. He later returned to Berkeley as Visiting Scholar (2008-2013) and Visiting Professor (2014) at the Department of Civil and Environmental Engineering as part of a collaborative research effort in Cyber-Physical Systems. His most recent research interests are in concurrent data structures, memory management, and so-called spatial programming. Dr. Kirsch co-invented embedded programming languages and systems such as Giotto, HTL, and the Embedded Machine, and more recently co-designed high-performance, multicore-scalable concurrent data structures and memory management systems. He co-founded the International Conference on Embedded Software (EMSOFT), served as ACM SIGBED chair from 2011 until 2013, and is currently ACM TODAES associate editor.

Abstract:
Scalloc is the currently fastest concurrent memory allocator for multi-threaded applications that run on multicore systems. Scalloc has been demonstrated to scale to thousands of threads outperforming all other scalable and publicly available allocators while using significantly less memory. There are two key ingredients to scalloc’s performance: explorative benchmarking of memory management using a tool called ACDC and high-performance, multicore-scalable concurrent data structures with relaxed (stack-like) semantics. ACDC enables exploration of all relevant performance characteristics of memory allocators which is key in the design of a memory allocator that performs well in all dimensions. The concurrent data structures in scalloc enable efficiently reclaiming memory allocated and deallocated by different threads which is key to its multicore scalability and low memory consumption. The story of the talk is to show how the design of new, relaxed concurrent data structures leads to new ways of engineering more efficient systems software that is otherwise extremely hard to improve.
Software: http://scalloc.cs.uni-salzburg.at
This is joint work with Martin Aigner, Michael Lippautz, and Ana Sokolova.