|
|
|||||||||||||||||
Course Catalog 2011-2012
SGN-9206 Signal Processing Graduate Seminar II, 3 cr
|
Lessons
| Study type | P1 | P2 | P3 | P4 | Summer | Lecture times and places |
|
|
|
|
|
|
|
|
Description:
General Programming on Graphics Processing Units (GPGPU) is an effective method for addressing computationally intensive tasks. Modern GPU processors are sophisticated parallel computing platforms. Their capabilities transcend their traditional role in computer graphics, and entertainment software. This processor architecture is especially well suited for computationally intensive tasks in dealing with large amounts of homogeneous data, as often encountered in various areas of signal processing. GPU based algorithm implementations offer multi-fold speedups in comparison to standard CPU based approaches, even in comparison to latest multi-core processors. In addition, in recent years, GPUs have become increasingly cheap and evermore ubiquitous, as they have become available in almost every desktop and laptop PC configuration. GPU based solutions have been applied in wide variety of fields such as, image processing, computer vision, medical image processing, audio signal processing, computational systems biology, physics, chemistry, statistical modeling, computational neuroscience, data mining, machine learning, etc. The aim of this seminar is to give an introduction into the concept of GPGPU in general, and into two existing standards in this area, CUDA and OpenCL. It also aims to entice the participants to think how their own research can benefit from the parallel processing power of GPUs.
Person responsible:
Moncef Gabbouj
Atanas Gotchev
Serkan Kiranyaz
Stanislav Stankovic
Assessment scale:
Evaluation scale passed/failed will be used on the course