HPC : AMD News Flash

Sunday 10 February 2013

AMD News Flash

Kent Knox
Kent joined AMD in 1999, hiring into a group that helped to optimize Microsoft's DirectX runtime to run well on AMD platforms. Developing a background in graphics and software optimization, he moved to the ACML team and began work on offloading and optimizing BLAS and FFT computations onto GPU devices. The culmination of this effort resulted in the release of APPML, a set of OpenCL optimized math libraries for BLAS & FFT. Kent is currently working on the newly announced Bolt project, a library which focuses on lowering the barrier of entry of OpenCL thus expanding AMD's effort to make heterogeneous computing mainstream. You can also read Kent's blog on"Implementing Black-Scholes using Bolt".

AMD 2013 Developer Summit call for papers is in full swing.
Here is your chance to secure a position at the forefront of the heterogeneous technology movement among your peers, academics, the tech media, and the financial community. More than 140 technology professionals delivered presentations at AFDS 2012. Now it's your opportunity to share, enlighten, and educate the technology community on your latest achievements. Visit the AMD 2013 DS Call for Proposalsweb site for complete submission details. Spaces for presentations will fill up quickly, so submit your proposal now.

Introducing Bolt - A New C++ Library
Bolt is a C++ template library that eases development of applications that utilize the inherent performance and power efficiency benefits of heterogeneous computing (hc). Download an updated version of the Bolt Preview here and read Siu Chi's blog on using Bolt to compute PI using a Monte-Carlo algorithm or Ben Sander's blog on the Bolt Architecture.

ForwardPlus11 - A New Radeon™ SDK Direct3D 11 Sample
Forward+ extends traditional forward rendering to support high numbers of dynamic lights while maintaining performance. It utilizes a Direct3D 11 compute shader (DirectCompute 5.0) to divide the screen into tiles and quickly cull lights against those tiles.
Download the sample.

Supersampling Anti-Aliasing (SSAA) 11 - A New Radeon™ SDK Direct3D 11 Sample
SSAA provides high quality solutions to full screen aliasing. One SSAA option is to set up the pixel shader to execute at sample frequency. The sparse location of samples provides significantly higher quality. Additionally, this sample implements Multisampling Anti-Aliasing (MSAA) & Enhanced Quality Anti-Aliasing (EQAA).

APPML 1.8 is Out of Beta & Available
APPML 1.8 contains the complete set of L2 & L3 routines for BLAS and adds real to complex support to the FFT library. Also, clMAGMA 1.0 is available. Learn more about APPML 1.8 and clMAGMA 1.0.

AMD Boosts Data Warehouse Performance with Parallel Processing Appliance
In testing semiconductor wafers, AMD uses a data warehouse to process a terabyte of data each week. When its data warehouse began foundering, AMD switched to Microsoft SQL Server 2008 R2 Parallel Data Warehouse powered by AMD Opteron™ processors.Read more about the performance increases after the switch.

Top 10 Winners of the AMD China Heterogeneous Applications Competition
Out of a field of over 200 applicants, 10 entrants have been declared the top remaining finalists in the AMD China Heterogeneous Applications Competition. The finalists are developing accelerated algorithms and applications using GPU compute. The 10 remaining finalists will continue to improve their projects in competition for the top 2 spots, which are expected to be awarded in March 2013.