Annual Conference: Communicating Process Architectures
Communicating Process Architectures 2018,
the 40th. WoTUG conference on concurrent and parallel systems, takes place from
Sunday August 19th. to Wednesday August 22nd. 2018 and is hosted by
Professor Dr. Rainer Spallek,
Chair of
VLSI Design, Diagnostics and Architecture
at the Faculty of Computer Science,
Technische Universität Dresden, Germany.
The conference is organised by Dr. Spallek in collboration with Oliver Knodel and Uwe Mielke
and in partnership with WoTUG.
About WoTUG
WoTUG provides a forum for the discussion and promotion of concurrency ideas,
tools and products in computer science.
It organises specialist workshops and annual conferences that address
key concurrency issues at all levels of software and hardware granularity.
WoTUG aims to progress the leading state of the art in:
-
theory (programming models, process algebra, semantics, ...);
-
practice (multicore processors and run-times, clusters, clouds, libraries, languages, verification, model checking, ...);
-
education (at school, undergraduate and postgraduate levels, ...);
-
applications (complex systems, modelling, supercomputing, embedded systems, robotics, games, e-commerce, ...);
and to stimulate discussion and ideas on the roles concurrency will play in the future:
-
for the next generation of scalable computer infrastructure (hard and soft) and application,
where scaling means the ability to ramp up functionality (stay in control as complexity increases)
as well as physical metrics (such as absolute performance and response times);
-
for system integrity (dependability, security, safety, liveness, ...);
-
for making things simple.
Of course, neither of the above sets of bullets are exclusive.
WoTUG publications
A database of papers and presentations from WoTUG conferences is here.
The Abstract below has been randomly selected from this database.
Improving TCP/IP Multicasting with Message Segmentation
By Hans Henrik Happe, Brian Vinter
Multicasting is a very important operation in high performance parallel applications. Making this operation efficient in supercomputers has been a topic of great concern. Much effort has gone into designing special interconnects to support the operation. Today’s huge deployment of NoWs (Network of Workstations) has created a high demand for efficient software-based multicast solutions. These systems are often based on low-cost Ethernet interconnects without direct support for group communication. Basically TCP/IP is the only widely supported method of fast reliable communication, though it is possible to improve Ethernet performance at many levels – i.e., by-passing the operating system or using physical broadcasting. Low-level improvements are not likely to be accepted in production environments, which leaves TCP/IP as the best overall choice for group communication. In this paper we describe a TCP/IP based multicasting algorithm that uses message segmentation in order to lower the propagation delay. Experiments have shown that TCP is very inefficient when a node has many active connections. With this in mind we have designed the algorithm so that it has a worst-case propagation path length of O(log n) with a minimum of connections per node. We compare our algorithm with the binomial tree algorithm often used in TCP/IP MPI implementations.
Complete record...
|
