We are very excited to announce the DirectCompute Lecture Series, which starts today, June 14th, and runs for the next three weeks. A component of DirectX, DirectCompute is Microsoft’s GPGPU programming solution that enables developers to utilize the GPU for general purpose computing on Windows. In today’s opening video lecture, Robert Hess hosts a round table discussion of thought leaders in parallel computing. Six lectures will follow about DirectCompute, developing DirectCompute applications, and tips for optimizing performance and integrating the DirectCompute data into the graphics pipeline.
Each week we will release two new lectures, Mondays and Fridays, that are presented by developers experienced with parallel computing and DirectCompute. Several lectures will address key parallel programing scenarios. The series will begin with high-level topics common to most DirectCompute scenarios, and continue into deeper lectures that examine more complex problems and algorithms.
With the realization that GPUs are useful for purposes other than graphics, the need for APIs to harness their potential has grown rapidly over the last several years. Microsoft developed DirectCompute in response to this need. Its benefits for mass-market applications are numerous. Because it can run on all DirectX 11 hardware and most DirectX 10.1 hardware, you can safely use DirectCompute for core application functionality across many customer segments. Additionally, as part of the Direct3D API, DirectCompute can be seamlessly integrated with games, CAD tools, and other graphics applications.
The integration with Direct3D makes DirectCompute an attractive choice for any scientific computing applications that require data visualization. The Compute Shader Extensions library provides access to high-performance implementations of common computing algorithms, such as parallel prefix reduction and multi-dimensional fast Fourier transform.
If you are interested in doing some GPU computing, look for our video lectures at http://go.microsoft.com/fwlink/?LinkId=194362, which will point to the lectures as they become available.
Currently, the following lectures are planned:
Introduction to DirectCompute
DirectCompute Memory Patterns
Basics of DirectCompute Application Development
DirectCompute GPU Optimizations and Performance
DirectCompute GPU Accelerated Physics
DirectCompute - Integration with the Graphics Pipeline