ARCHITECTURAL CHOICES FOR THE COLUMBIA 0.8-TERAFLOPS MACHINE

被引：5

作者：

ARSENIN, IV

机构：

[1] Physics Department, Columbia University, New York

来源：

NUCLEAR PHYSICS B | 1995年

关键词：

D O I：

10.1016/0920-5632(95)00416-7

中图分类号：

O412 [相对论、场论]; O572.2 [粒子物理学];

学科分类号：

摘要：

We discuss the hardware design choices made in our 16K-node 0.8 Teraflops supercomputer project, a machine architecture optimized for full QCD calculations. The efficiency of the conjugate gradient algorithm in terms of balance of floating-point operations, memory handling and utilization, and communication overhead is addressed. We also discuss the technological innovations and software tools that facilitate hardware design and what opportunities these give to the academic community.

引用

页码：902 / 904

页数：3