A Flow Solver for a Reconfigurable FPGA-Based Hypercomputer
A Flow Solver for a Reconfigurable FPGA-Based Hypercomputer
复制标题
用于基于 FPGA 的可重构超级计算机的流程求解器
DOI：
10.2514/6.2005-1382
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
T. Hauser
中科院分区：
文献类型：
--
作者：
T. Hauser
本文详细详细介绍了针对可重新配置的基于可重构的现场可编程栅极阵列（FPGA）在由Star Bridge Systems，Inc，Inc. Inc.该领域的程序员将针对特定问题重新配置。除了FPGA硬件的动态性质（可以优化特定应用程序）外，FPGA还固有地平行。这使程序员能够同时运行尽可能多的操作，以便使用其资源比CPU上的可能性更有效地适合FPGA。实现了基本对象，例如向量和矩阵及其相应的操作，例如Dot-rododuct，矩阵-Vector和矩阵矩阵乘法。然后，这些基本组件被用于构建一维，明确的可压缩流求解器。
This paper details initial efforts to program a basic, compressible flow solver on a reconfigurable field-programmable-gate-array(FPGA) based computer manufactured by Star Bridge Systems, Inc. An FPGA is an array of Silicon gates on a computer chip that can be reconfigured by programmers in the field for specific problems. In addition to the dynamic nature of FPGA hardware (which enables optimization to specific applications), FPGAs are also inherently parallel. This enables programmers to make as many operations run concurrently as they can physically fit on the FPGA, using its resources more efficiently than is possible on a CPU. Basic object like vectors and matrices and their corresponding operations like dot-product, matrix-vector and matrix-matrix multiplication were implemented. These basic components were then used to build a one-dimensional, explicit compressible flow solver.