Archive for October, 2009

NVIDIA’s October GPU Computing Webinars now open for registration.
These webinars cover many topics including an introduction to C for CUDA, the OpenCL™ API, and performance optimization techniques, presented by NVIDIA DevTech Engineers with additional staff online to answer questions.

These webinars will help you get ready for the next CUDA SuperHero Challenge which will start on November 23rd, the CUDA practice servers are up and running:
http://www.topcoder.com/tc?module=Static&d1=sponsors&d2=NVIDIAOverview

Please follow the links to register for each webinar you would like to attend. Advance registration is required. Please note all times are in Pacific Time (UTC-7 hrs, British Summer time -8hrs).

Full Schedule and short abstracts can be viewed at:
http://developer.nvidia.com/object/gpu_computing_online.html?refCode=nvinvite

An Introduction to GPU Computing and The CUDA Architecture, 1.5 Hours
Monday, 10/19/09, 9pm: https://www2.gotomeeting.com/register/214702074

Introducing a new Multi-GPU Frameframe
This is an extended, technical presentation about a new framework for developing code for Multi-GPU environments.
Wednesday 10/28/09, 9am: https://www2.gotomeeting.com/register/628549827

Memory Optimization and Performance Considerations for CUDA, 1.5 hours
Wednesday, 10/21/09, 9am: https://www2.gotomeeting.com/register/723693579
Wednesday, 10/21/09, 9pm: https://www2.gotomeeting.com/register/731145515

Further CUDA Optimization Techniques, 1.5 hours
Monday, 10/26/09, 9am: https://www2.gotomeeting.com/register/406661946
Monday, 10/26/09, 9pm: https://www2.gotomeeting.com/register/953597875

An Introduction to GPU Computing and OpenCL, 1.5 hours
Thursday, 10/22/09, 9am: https://www2.gotomeeting.com/register/409536002
Thursday, 10/22/09, 9pm: https://www2.gotomeeting.com/register/580290250

Best Practices for OpenCL programming, 1.5 hours
Friday, 10/23/09, 9am: https://www2.gotomeeting.com/register/352716234
Thursday, 10/29/09, 9am: https://www2.gotomeeting.com/register/422319282
Thursday, 10/29/09, 9pm: https://www2.gotomeeting.com/register/719309611

The next Information Research Group meeting will be in E11_A3 - Arts Lecture Theatre 3 from 1 to 2pm on Friday, 30th October 2009. The focus will be on programming Graphics Processing Units for High Performance Computing.

The Graphics Processing Unit (GPU) is a recent computer hardware development providing high performance graphics rendering. These highly parallel processors are now being applied to computationally intensive problems. The GPU architecture is well suited to parallel computing with minimal inter-process communication. Tasks are divided into multiple threads that run independently on single processing elements in the same way that regions of an image are rendered.

Now that GPUs have evolved into fully programmable devices they have become an ideal resource for acceleration of many arithmetic and memory bandwidth intensive scientific applications. GPUs are typically composed of groups of single-instruction multiple-thread processing units. Parallel machines in the past failed to achieve their full performance potential due to memory access conflicts and divergence of execution paths for conditional execution operations. GPU design ameliorates these problems by using hardware multithreading, clusters of small processing units and virtualized processors.

The CUDA development environment makes programming GPUs accessible to a wide group of users. Thread allocation and memory management is simplified, code can be read from arbitrary addresses in memory and fast shared memory used as a user-managed cache, enabling higher bandwidth. The CUDA programming model is based on the decomposition of work into grids and thread blocks. Grids decompose a large problem into thread blocks which are concurrently executed by the pool of available multiprocessors. Each thread block contains from 64 to 512 threads, which are concurrently executed by the processors within a single multiprocessor. Each thread block is computed by running a group of threads, known as a warp, in lockstep on the multiprocessor.

At the meeting we will look at coding practical algorithms for high performance implementation and identify problems that will be developed further at the UNE Summer CUDA Code Camp.