|
AMD OpenCL Reductions Case Study |
 |
 |
 |
|
Written by Douglas Andrade
|
|
Saturday, 28 August 2010 10:31 |
AMD OpenCL Reductions Case Study Vector element sum, maximum and minimum computing and some other operations are typically serial tasks. In this Case Study, AMD shows reduction algorithms to paralellize these tasks and get efficient GPU acceleration for this type of task. This Case Study is very general and applicable to a wide range of problems. Check AMD Reductions Case Study .
|
|
Last Updated on Saturday, 28 August 2010 10:40 |
|
|
OpenCL Support Vector Machine |
|
|
|
|
Written by Douglas Andrade
|
|
Monday, 23 August 2010 23:17 |
OpenCL accelerated Support Vector Machine Support Vector Machines (SVMs) is a statistical learning tool considered to be the state-of-the art classifiers for many applications today, including medical research and text categorization. CMSoft brings an OpenCL accelerated SVM implementation with source code showing classification of the MNIST handwritten database. Check the OpenCL SVM resource. 
OpenCLTemplate 1.131 Download OpenCLTemplate Changelog - Added MachineLearning Support Vector Machine algorithm.
|
|
Last Updated on Monday, 23 August 2010 23:24 |
|
Written by Douglas Andrade
|
|
Sunday, 08 August 2010 18:15 |
OpenCL 1.1 Coming Official OpenCL 1.1 support is about to come both from AMD and NVidia. Some important changes involve the possibility to read and write submatrixes of buffer elements (variables). Meanwhile, we here are working with Support Vector Machines for pattern recognition and a sparse linear system solver using OpenCL. These take some time to tune properly and we intend to release these features inside OpenCLTemplate as CLPrograms. If you want to test a preview of the above systems, feel free to contact us.
|
|
Last Updated on Sunday, 08 August 2010 18:20 |
|
|
Written by Douglas Andrade
|
|
Saturday, 14 August 2010 21:39 |
OpenCL 1.1 With the recent AMD Stream SDK 2.2 and NVidia release of OpenCL drivers, OpenCL 1.1 is now a reality. Congratulations to the manufacturers and the Khronos Group for this is another step towards an open, high-performance multi-platform GPU computing standard. Meanwhile, we will continue studying decision-making algorithms, linear system solving and performing database search using GPU.
|
|
Last Updated on Saturday, 14 August 2010 21:45 |
|
Written by Douglas Andrade
|
|
Saturday, 31 July 2010 14:21 |
OpenCLTemplate Documentation updated OpenCLTemplate Doxygen documentation has been updated and is available at http://www.cmsoft.com.br/OpenCLTemplateDocs/ Please notice that the CLPrograms that have had more in-depth testing are MarchingCubes and RK46. The other are experimental programs.
|
|
Last Updated on Saturday, 31 July 2010 14:24 |
|
|
|
|
<< Start < Prev 1 2 3 4 5 6 7 8 Next > End >>
|
|
Page 1 of 8 |