News

Work on Co-scheduling Will be Presented at ICPP

FAST is going to present a paper at the 11th International Workshop on Scheduling and Resource Management for Parallel and Distributed Systems which is held in conjunction with International Conference on Parallel Processing (ICPP'15) in Beijing, China.

Title: Case Study on Co-Scheduling for HPC Applications

Authors: Jens Breitbart, Josef Weidendorfer, Carsten Trinitis

Abstract: Most applications running on supercomputers achieve only a fraction of the peak performance of the system. In this paper we analyze the performance and energy efficiency of co-scheduling one memory bandwidth bound and one compute bound application on the same node. We present autopin+, a tool designed to monitor and optimize co-scheduling of applications. Our analysis shows that co-scheduling can improve both energy efficiency and overall throughput of a supercomputer. At best, runtime can be decreased by 28% and the energy consumption by 12%, respectively, compared to best case dedicated execution. The overall efficiency however strongly depends on the ratio of jobs available in the queue. We furthermore present a simple adaptive strategy depending on the available jobs in the queue.

Poster Presentation at NAS Conference

FAST is going to present a poster at the 10th IEEE International Conference on Networking, Architecture, and Storage  (NAS2015), which will be held at the Boston Park Plaza Hotel in Boston, USA, August 6 − 7, 2015.

Poster title: Evaluation of a Hash-Compress-Encryptfor Storage System Applications Pipeline

Authors: Matthias Grawinkel, Michael Mardaus, Tim Süß, André Brinkmann

Abstract: Great efforts are made to store data in a secure, reliable, and authentic way in large storage systems. Specialized, system specific clients help to achieve these goals. Nevertheless, often standard tools for hashing, compressing, and encrypting data are arranged in transparent pipelines.
We analyze the potential of Unix shell pipelines with several high-speed and high-compression algorithms that can be used to achieve data security, reduction, and authenticity. Furthermore, we compare the pipelines of standard tools against a house made pipeline implemented in C++ and show that there is great potential for performance improvement.

Talk About FAST Project at HPCS 2015

A talk about the FAST project will be given at the "International Conference on High Performance Computing & Simulation" (HPCS 2015) in Amsterdam, Netherlands. The talk includes an introduction to FAST as well as a more detailed insight into migration techniques and co-scheduling. It is scheduled for Tuesday, the 21st July.

Live LAMA Demo at ISC’15 in Frankfurt

Fraunhofer SCAI is going to give a live demonstration at the International Supercomputing Conference (ISC 15) in Frankfurt. They will show how a Computational Fluid Dynamics (CFD) application performs on a low-energy cluster which is composed of four NVIDIA Jetson TK1 boards and utilizes the GPUs as well as the ARM multi-core processors while consuming only 25W in total. The application is based on LAMA which is one of the test applications within FAST.

laurettaTegraClusterLAMA

FAST Project Presented at EASC

In the third week of April the Exascale Applications and Software Conference (EASC) took place in Edinburgh, Scotland. Organized by the Edinburgh Parallel Computing Centre, the SIGHPC ACM conference covered recent work on the challenge to prepare applications for the upcoming exascale computing infrastructures.
During the course of the event the FAST project was presented with a lightning talk and a poster. Both were very well received and interest was expressed by many participants.

Meeting of the FAST Consortium in Cologne

The FAST project partners held a project meeting in Cologne in which the current work was discussed. Key performance indicators were determined on whose basis the initial placement of the jobs can be improved.
Migration techniques for the rescheduling were discussed and selected. Other topics include the agent-based monitoring framework, the analyses of the example applications and the simulation environment which will be used for the evaluation of the scheduling algorithms.

Posted on | Posted in News

Talk About Migration Techniques in HPC Environments at VHPC’14

We are glad to announce that the FAST project will present a research paper at the 9th Workshop on Virtualization in High-Performance Cloud Computing (VHPC'14).

The 9th Workshop on Virtualization in High-Performance Cloud Computing
(VHPC'14) will be held in conjunction with (Euro-Par 2014), on August 25-29, Porto, Portugal.

The talk "Migration Techniques in HPC Environments" will be given on the 26th of August, at 9:50 am.

Posted on | Posted in News