论文部分内容阅读
Skein算法提供一种树形哈希模式,方便了并行硬件资源的充分利用.提出一种并行计算平台,用以评估树形哈希算法与传统线性哈希算法相比的性能提升.该平台通过开关网络连接多个运算单元和多个存储单元,允许并行运算和并发的存储器访问.平台中包含一个控制器,负责动态地向运算单元调度任务,支持算法参数的灵活配置.为了确定最优配置参数,根据平台特征建立了性能模型.实验结果显示,加速比趋近于理论上限,即平台中运算单元的数量.该平台的硬件原型通过台积电65 nm工艺进行综合,工作频率达到833 MHz,对应吞吐率为38.091 Gbps.
Skein algorithm provides a tree-like hash pattern to facilitate the full utilization of parallel hardware resources.A parallel computing platform is proposed to evaluate the performance improvement of tree-hash algorithm compared with the traditional linear hash algorithm, The switching network connects multiple computing units and multiple storage units to allow parallel computing and concurrent memory access.The platform includes a controller that is responsible for dynamically scheduling tasks to computing units and supporting flexible configuration of algorithm parameters.In order to determine the optimal configuration The experimental results show that the speedup approaches to the theoretical upper bound, that is, the number of computing units in the platform.The hardware prototype of the platform is integrated by TSMC’s 65 nm process and the operating frequency reaches 833 MHz, corresponding to Throughput was 38.091 Gbps.