NxtGenUG - Event View

	Search

Latest Headlines

local Indie IronSun Studio's needs you Kickstarter backing to get their exciting Fathom game off the ground

More...

Modus IT has an urgent requirement for an Analyst Programmer to help deliver the first version of a revolutionary Business Intelligence product to the market.

Coming Events

Monday, June 03, 2013

Hereford - GIT

Tuesday, June 18, 2013

Birmingham - Making Windows Store Dance

Wednesday, June 19, 2013

Manchester - TDD, where did it all go wrong

Monday, July 08, 2013

Hereford - Windows Phone 8 WP8

Monday, August 12, 2013

Hereford - Intro to Angular JS

Event Calendar...

Member Quotes

Event Resources

Conferences

Mix11

Partner Showcase

Grey Matter has provided independent and trusted software supply and advice since 1983.Grey Matter offers a wide range of software to cover all your business needs and specialises in delivering licensing and support advice on all your software requirements, no matter the size of your organisation. Grey Matter is a Microsoft Gold Partner.

Event: CUDA

The way we process data is shifting, more CPU cores means the processing and data can be streamed across CPU cores or across many machines in the case of scale out technologies like Hadoop and Cassandra.

Within the scientific community its long been established that GPGPU (aka your graphics card) can be used to process streams of data with massive parallelism.

In this talk we look at the Compute Unified arena, the principals of then focusing on CUDA from NVIDIA.

Pizza of course will be served!

If you are not a member of NxtGenUG, you are more than welcome to attend. We just ask you register on the site, to help us to plan numbers and refreshments. Once registered on the site, you can then register for this event from this page.

Monday, October 08, 2012
Region:	Hereford
Location:	Shire Hall Shire Hall Union Street Hereford HR1 2HX
Time:	7:00 PM - 9:00 PM

Event Breakdown

Tony Rogerson	CUDA
The way we process data is shifting, more CPU cores means the processing and data can be streamed across CPU cores or across many machines in the case of scale out technologies like Hadoop and Cassandra. Within the scientific community its long been established that GPGPU (aka your graphics card) can be used to process streams of data with massive parallelism. In this talk we look at the Compute Unified arena, the principals of then focusing on CUDA from NVIDIA. • What interests me? • Parallelism for Data Processing – CPU cores, GPU cores • RBAR or Set processing? • SISD, SIMD, SIMT, MIMD (Single/Multiple Instruction Single/Multiple Data/Thread) • The General Purpose Graphics Processing Unit (GPGPU) Paradigm • Landscape – CUDA, OpenCL • Hybrid Programming – mix CPU and GPU execution • The can’t get data onto device quick enough myth • Using CUDA • Nvidia SDK and toolset • Programming from .NET (CUDA.NET and CUDAfy) • Grids, Blocks and Threads – the threading mechanism • Kernels • Memory types • Example