标题： 每秒万亿次浮点运算的引力透镜光线投射技术使用图形处理单元 显示英文标题

标题： Teraflop per second gravitational lensing ray-shooting using graphics processing units

摘要： 利用直接反向光线投射方法计算引力透镜是一项计算开销很大的工作，用于确定放大率图、焦散模式和光变曲线（例如，作为源轮廓和大小的函数）。然而，由于这是一个容易并行化的计算，引力光线投射可以通过可编程图形处理单元（GPU）加速。我们介绍了使用NVIDIA G80代图形处理器以及NVIDIA统一设备计算架构（CUDA）软件开发工具包实现的反向光线投射方法。我们还将代码扩展到多GPU系统，包括一个4-GPU的NVIDIA S1070 Tesla单元。我们在单个GPU上实现了182千兆浮点运算每秒的持续处理性能，在Tesla单元上达到了1.28万亿浮点运算每秒。我们证明，使用Tesla单元可以在不使用分层树代码的情况下，在一天的时间尺度内运行十亿镜头微引力透镜模拟。 显示英文摘要

摘要： Gravitational lensing calculation using a direct inverse ray-shooting approach is a computationally expensive way to determine magnification maps, caustic patterns, and light-curves (e.g. as a function of source profile and size). However, as an easily parallelisable calculation, gravitational ray-shooting can be accelerated using programmable graphics processing units (GPUs). We present our implementation of inverse ray-shooting for the NVIDIA G80 generation of graphics processors using the NVIDIA Compute Unified Device Architecture (CUDA) software development kit. We also extend our code to multiple-GPU systems, including a 4-GPU NVIDIA S1070 Tesla unit. We achieve sustained processing performance of 182 Gflop/s on a single GPU, and 1.28 Tflop/s using the Tesla unit. We demonstrate that billion-lens microlensing simulations can be run on a single computer with a Tesla unit in timescales of order a day without the use of a hierarchical tree code.