As traditional methods of improving processor performance are reaching their limits, the high-performance computing community is exploring alternative methods in improving performance and efficiency. This has driven the development of alternative processor technologies with performance that exceeds traditional commodity processors. The most prominent examples of these technologies are the Nvidia Tesla GPU-based accelerators and the upcoming Intel MIC (Many Integrated Core).

While the raw performance of these processors is impressive, utilizing them often requires significant effort. On the other hand, evolutionary improvements in both hardware and software are constantly lowering the bar to use these technologies efficiently.

In order to track this changing landscape, CSC has been investigating them in the Tekes-funded "Finnovative Computing" (FinnoComp) project for the last two years. This seminar is intended to disseminate information learned through this project.

The topics include:

• Emerging processor technologies
  
• Novel programming languages
  
• Accessing next-generation computing resources at CSC
  
• Case study of the porting and optimization of a parallel application to an accelerator
  

The seminar is intended for anyone who is interested in current and near-future accelerator technology and how to utilize it.