Fakultät Informatik

29.06.2017

Kolloq. Aleksei Popov, topic: Hardware acceleration of operations on graph structures, sets, associative arrays, and trees for Internet of Things and beyond

Thursday, 29th of June 2017, 11:00 am FMI 01.06.020 (MI-Building, Campus Garching)

Abstract:
The processing of data structured in form of graphs, sets, associative arrays, or trees is present in many modern research and application areas, such as motion planning for robots, computer vision, logistics, national security for communication networks, high performance computing, modeling, social networks analysis, computational biology, and Internet of Things (IoT). A variety of algorithms was developed to exploit these representations’ advantages. This yields the research and engineering challenge to develop special data processing hardware architectures to speedup processing of these data structures. In order to overcome this challenge, a novel Discrete mathematics Instruction Set Computer (DISC) architecture was introduced.
In the scope of the presentation, you will learn about this architecture and its applications (with emphasize on Internet of Things) from the leader of the development group, Assoc. Prof., Dr. Aleksei Popov. The DISC architecture consists of two processing cores with a shared memory: one of these cores has an architecture of a standard CPU, while another one, Structures Processing Unit (SPU) “Leonhard”, has a unique architecture enabling it to rapidly process data structures represented in the form of B+ trees consisting of key-value pairs with discrete mathematics instructions. The FPGA implementation of the DISC architecture has demonstrated up to 40-fold acceleration compared to common modern CPUs in respect to clock cycles. PCI-board with the soft-core is currently available for tests on a remote server.
The presentation consists of three parts:

  • Introductory part about Bauman Moscow State Technical University (BMSTU) and Chair of Computer Systems and Networks
  • BMSTU Internet of Things educational environment, Internet of Things as an application area for novel computer architectures
  • A conceptual and technical overview of Discrete mathematics Instruction Set Computer and its wide application areas

Bio:
Assoc. Prof., Dr. Aleksei Popov conducts and leads the research at the Bauman Moscow State Technical University in such areas as: heterogeneous acceleration technology, FPGA and embedded systems, discrete optimization and algorithms on graphs, Internet of Things, Cloud Computing, e-learning, virtualization. He has more than 20 years of professional industry experience (among which 8 years on a leading position at IBM) and more than 12 years of teaching and research experience. Currently, he finishes his habilitation thesis.

contact person:
--------------------
Michael Gerndt
Phone: +49.89.289.17652
Email: gerndt(at)in.tum.de

 

Kolloquientermine

« December - 2017 »
MTWTFSS
 010203
04050607080910
111214151617
18192021222324
25262728293031

Ansprechpersonen