Mingsong Lu's Page


Information Exchange for Academic Purpose

Personal Information

Mingsong Lu (吕鸣松)

Ph.D Candidate

Institute of Computer Science and Engineering

Northeastern University  (110004)

3-11, Wenhua Road, Heping District, Shenyang, China P.R.

Contact Me

Email: mingsong_lv@hotmail.com

Office Phone: (86)24-2389-7811

Fax: (86)24-2391-8918

Mobile Phone: (86)138-4028-9915

Research Interests

Originally, system level design of real-time embedded systems, formal methods for real-time embedded systems.

Recently, I am very interested in desktop-available parallel computing, enabled by the emerging of Chip-Multi-Processor technologies, such as Multi-core CPU, Cell Broardband Engine, GPUs and streaming accelerators.

Since the improvement of CPU computing capacity is deathly stalled by the power-wall, we have to change our creteria of evaluating a computing system from GFLOPS/sec. to GFLOPS/sec./power, especially for desktop computers and embedded computers. Different coporations start from their original expertises to tackle the problem by providing different types of CMP chips ranging from coarse-grained multi-core chips to fine-grained GPUs. It's quite apparent that, this era is at the very beginning, since non of the above technologies are really mature. But we are confident that several parallel architectures may dominate future computing world, and we're definitely facing the coming "parallel thinking era".

The driven force of this ear is the continuous increasing demands on personal computing, either desktop or embedded. We are to provide such chips and architectures that can satisfy complex requirements within tolerable power consumptions. We can call this "Personal Super Computing", comparing to the traditional super computing, which makes us think of large scientific problems and those huge monster machines.

There is no doubt that the consumers demands are increasing, and chip providers are giving us multiple enabling technologies. But the problem is, most of the engineers are not educated to program, not to say think, in parallel. So there is a big gap between applications and CMP technologies. We think that research on software-aspects of such enabling technologies, ranging from programming model, high-level system abstraction, automatic system partitioning, low-level languages and compilers, will be the critical issues, and deserves intensive interests.

 

Publications

1  Nan Guan, Mingsong Lv, Qingxu Deng and Ge Yu: A Real-Time Scheduling Algorithm with Buffer Optimization for Embedded Signal Processing Systems. The 4th IEEE International Symposium on Embedded Computing (SEC'07).

Qingxu Deng, Mingsong Lv, Ge YuSelecting a Scheduling Policy for Embedded Real-Time Monitor and Control Systems. ICESS 2004: 494-501, 2004

3  Dan Wang, Ge Yu, Mingsong Lv, Baoyan Song, Derong Shen, Guoren Wang: An Ant Algorithm Based Dynamic Routing Strategy for Mobile Agents. APWeb 2003: 453-464, 2003

Links

Zonghua Gu's Page at HKUST.

My Hobbies

Football, Table tennis, Tourism, Consumer electronics - especially SONY MiniDisc Players and Cellphones, and a lot yet to be added on...