Wolfram Library Archive


Courseware Demos MathSource Technical Notes
All Collections Articles Books Conference Proceedings
Title Downloads

S71: Advanced GPU Programming Using Mathematica and CUDA
Authors

Abdul Dakkak
Ulises Cervantes-Pimentel
Organization: Wolfram Research, Inc.
Department: Kernel Technology
Conference

Free Online Seminars
Description

This seminar provides an in-depth overview of the new GPU programming functionality in Mathematica 8 through CUDA. Topics include how to compile CUDA code into an executable, load user-defined CUDA functions into Mathematica, use CUDA memory handles to increase memory bandwidth, and use Mathematica parallel tools to compute on multiple GPUs either on the same machine or across networks, as well as a discussion about the general workflow of CUDA programming within Mathematica.
Subjects

*Mathematica Technology > Kernel > Compiling
*Mathematica Technology > Linking Technology
*Mathematica Technology > Parallel Computing
*Mathematica Technology > Programming > 2D Graphics
*Mathematica Technology > Programming > 3D Graphics
*Mathematica Technology > Programming > Animations
Keywords

seminar, S71, Compile, LibraryLink, computing performance, parallel computing, multicore computing, NVIDIA, GPU, CUDA, Fermi, many-core processor, Tesla, ATI, AMD
URL

http://www.wolfram.com/services/education/seminars/s71.html
Downloads

Download
S71CUDAProgramming.zip (4.6 MB) - ZIP archive