Cool Compute - Harnessing More Performance per Watt to Do Cool Stuff

     Cool Compute - Harnessing More Performance per Watt to Do Cool Stuff

  PR Newswire

  LONDON, April 24, 2013

LONDON, April 24, 2013 /PRNewswire/ --

    Industry leaders to share their latest insights on developing advanced
 applications that harness the performance of parallel computing on low-power
                                GPUs and FPGAs

The LEAP conference (London, May 21-22), today announced the world-class line
up of keynotes and speakers that are ready to share their experience and
insights with developers looking to harness modern multicore mobile CPU, GPU
and FPGA architectures to build the next generation of low-energy, compute
intensive applications.

The growth in GFLOPS/Watt of mobile GPUs and FPGAs is claimed to exceed
Moore's law and the parallel architectures used to achieve this are becoming
increasingly programmable for non-graphical applications. The use of software
frameworks such as the Khronos OpenCL and Android Renderscript APIs are
enabling developers to make practical use of the energy-efficient performance
available on such devices to develop a wide range of embedded and mobile
applications and libraries.

"Advanced applications have to be delivered within the thermal constraints of
embedded and mobile form factors, which makes maximizing GFLOPS/Watt a
priority," said Tim Lewis, LEAP organiser. "LEAP provides a forum to explore
how parallel computing can be used to deliver this next generation of
applications; from enhanced computational photography and augmented reality to
advanced encryption systems and signal processing."

LEAP provides a unique forum for mobile and embedded developers to learn,
network and discuss the latest developments in L ow- E nergy A pplication P
arallelism. The event brings together researchers, suppliers, developers and
industry leaders in a unique central London location on May 21 ^st - 22 ^nd .
Registration is now open for both the one-day conference (21 ^st ) and the
tutorial and Khronos sessions on day two.


GPU Compute: A New Landmark for Energy Efficient Processing

GPUs are capable of superior computational power and energy efficiency for
massive data-parallel workloads. This keynote will cover the technology, its
benefits and most importantly the practical use cases that can take advantage
of it.

Roberto Mijat, ARM  Limited

Accelerating Extraordinary User Experiences on Mobile Devices

The growing array of sensors in your mobile are enabling a new generation of
connected, context aware experiences. This keynote focuses on the
applications enabled by this new wave of parallel processors and the APIs
required to support them.

Neil Trevett, Khronos President , OpenCL Chair &  VP Mobile Content  NVidia


Trends in Heterogeneous Systems Architectures

Simon McIntosh-Smith, Bristol University

The Impact of Low-Power GPUs to Deliver Advances in Real-World Applications

Paul Keir, Codeplay Software Limited.

Cool Compute: Strategies for Managing Power vs Performance in Mobile Devices

Doug Watt - Imagination Technologies Limited

Renderscript Accelerated Image and Video Processing

Lihua Zhang - MulticoreWare Inc

Developing High-Performance Low-Power Solutions using FPGAs and OpenCL

Craig Davis - Altera Corporation

Accelerated Computational Photography on Mobile Processors & GPUs

Mikael Sevenier - High-Performance Imaging, Aptina Inc.

Experiences with Mobile Processors for Energy-Efficient HPC

Paul Carpenter - Barcelona Supercomputing Center

Paralldroid : A Framework for Parallelism in Android

Alejandro Acosta - High Performance Computing Group, University of La Laguna

A System for Analyzing Power Consumption in a Dual Cortex-A15 Platform

Seongho Son - Samsung Electronics, Core Technology Team

Full abstracts, speaker biographies and conference details:

Tutorial: Formal Analysis Techniques for GPU Kernels; prove your code will

Wed 22nd May (am). Nathan Chong from the Department of Computing at Imperial
College will present a lively and interactive tutorial aimed at OpenCL and
CUDA kernel developers. Whether working on kernels for supercomputing,
finance or mobile applications this tutorial will help developers overcome the
common pitfalls in GPU programming such as data races and barrier divergence.
Using plenty of worked examples and demos to encourage interactive discussion
this session will highlight the practical benefits of using formal
verification techniques to prove that kernels are free from defects.

Meeting: Khronos UK Chapter Meeting

Wed 22nd May (pm). LEAP will host the inaugural meeting of the newly-founded
UK chapter of Khronos. This meeting will provide an opportunity for anyone
using or interested in any of the Khronos standards to meet, discuss and learn
about the work of the standards group; to share advice and techniques for
using the APIs and to provide feedback to vendors and Khronos on issues and
future requirements. The session will be hosted by the representatives of
Samsung Research UK who are chairing the new UK chapter, along with Neil
Trevett, President of Khronos. This session is FREE but registration is
required as space is limited.

About LEAP

LEAP 2013 (Low-Energy Application Parallelism) provides an opportunity to
learn about the latest advances in the application of high-performance
parallel computing on low-power multi-core CPU, GPU and FPGA processors. A
world-class selection of keynote presenters, speakers and  researchers  will
give their insights into the latest developments, trends and applications of
this technology to help  others  assess and exploit its potential benefits
across a broad range of applications and platforms.  LEAP is sponsored by
ARM Limited and the Khronos Group with academic support from  Bristol 
University  and  Imperial  College . A full list of partners is available
on the website. 

Contact: Press Contacts: Tim Lewis (co-organiser),,
Press spacebar to pause and continue. Press esc to stop.