A network architecture for Petaflops supercomputers.

DeBenedictis, Erik P.

Sandia National Laboratories

关键词: 99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Computer Architecture;    Computer Networks;    Supercomputers;    Data Transmission;   

DOI  :  10.2172/918315
RP-ID  :  SAND2003-2954
RP-ID  :  AC04-94AL85000
RP-ID  :  918315

美国|英语

来源: UNT Digital Library

【 摘 要 】

If we are to build a supercomputer with a speed of 10{sup 15} floating operations per second (1 PetaFLOPS), interconnect technology will need to be improved considerably over what it is today. In this report, we explore one possible interconnect design for such a network. The guiding principle in this design is the optimization of all components for the finiteness of the speed of light. To achieve a linear speedup in time over well-tested supercomputers of todays' designs will require scaling up of processor power and bandwidth and scaling down of latency. Latency scaling is the most challenging: it requires a 100 ns user-to-user latency for messages traveling the full diameter of the machine. To meet this constraint requires simultaneously minimizing wire length through 3D packaging, new low-latency electrical signaling mechanisms, extremely fast routers, and new network interfaces. In this report, we outline approaches and implementations that will meet the requirements when implemented as a system. No technology breakthroughs are required.