A Generation Ahead of Microprocessor: Where Software Can Drive uArchitecture To?

Abstract

The presentation will start with introduction of Microprocessor Technology Labs at Intel to show the Intel’s efforts in the uArchitecture and system software research areas. As all of you know, Moore’s law successfully leads Intel microprocessor business in decades. Transistor densities will continue to increase. However, device speed scaling will not follow historical trends and leakage power remains an issue. Meanwhile, new usage models and workloads will continue to demand greater performance. We can’t run business as usual. We need to develop uArchitecture features more effectively with power constraints. Software in both applications and systems will play more and more important roles in uArchitecture research and microprocessor design. Emerging applications like mining, recognition and synthesis (MRS) will become the next generation dominated applications. Study of the characterizations of these applications is significant for the next generation of microprocessor design. System software has been changing its landscape as well. There are challenges and opportunities for microprocessor designers and researchers to explore new uArchitecture features to meet the needs of the new application and system software. The presentation will show the potential direction of uArchitecture for such emerging applications also. The talk will discuss the emerging paradigms of programming systems and its impact to uArchitecture design by giving couple examples of how compilation technologies enable HW design.