葡萄酒:一个数值软件库,用于在x86体系结构上使用SIMD指令集加速无碰撞$N$ -体模拟
Phantom-GRAPE: numerical software library to accelerate collisionless $N$-body simulation with SIMD instruction set on x86 architecture
摘要:
我们开发了一个用于无碰撞N体模拟的数值软件库,名为“Ammetom-GRAPE”,它通过使用x86架构的新SIMD指令集扩展AVX(SSE的增强版本)来高度加速粒子之间的力计算。在我们的库中,不仅可以快速计算牛顿力,还可以快速计算任意形状的有心力f(r),其具有有限的截止半径r_cut(即f(r)=0,r>r_cut)。使用英特尔酷睿i7- 2600处理器,我们测量了我们的库的性能,这两种力量。在牛顿力的情况下,我们用1个处理器核心实现了每秒2 x 10^9次交互,这比没有显式使用SIMD指令的实现的性能高20倍,比SSE指令高2倍。使用4个处理器内核,我们获得了每秒8 x 10^9次交互的性能。在任意形状的力的情况下,我们可以计算1 x 10^9和4 x 10^9每秒的相互作用,分别与1和4个处理器内核。单处理器核的性能分别是不使用SIMD指令和使用SSE指令的性能的6倍和2倍。这些性能弱依赖于粒子的数量。这与GPU加速的力计算的性能强烈依赖于粒子数量的事实形成了良好的对比。性能对粒子数量的依赖性很弱,适合于无碰撞N体模拟,因为这些模拟通常使用复杂的N体求解器(如Tree和TreePM方法)与单独的时间步长方案相结合进行。使用我们的库加速的无碰撞N体模拟比使用GPU加速的模拟具有显著的优势,特别是在大规模并行环境中。
英文摘要:
(Abridged) We have developed a numerical software library for collisionless N-body simulations named "Phantom-GRAPE" which highly accelerates force calculations among particles by use of a new SIMD instruction set extension to the x86 architecture, AVX, an enhanced version of SSE. In our library, not only the Newton's forces, but also central forces with an arbitrary shape f(r), which has a finite cutoff radius r_cut (i.e. f(r)=0 at r>r_cut), can be quickly computed. Using an Intel Core i7--2600 processor, we measure the performance of our library for both the forces. In the case of Newton's forces, we achieve 2 x 10^9 interactions per second with 1 processor core, which is 20 times higher than the performance of an implementation without any explicit use of SIMD instructions, and 2 times than that with the SSE instructions. With 4 processor cores, we obtain the performance of 8 x 10^9 interactions per second. In the case of the arbitrarily shaped forces, we can calculate 1 x 10^9 and 4 x 10^9 interactions per second with 1 and 4 processor cores, respectively. The performance with 1 processor core is 6 times and 2 times higher than those of the implementations without any use of SIMD instructions and with the SSE instructions. These performances depend weakly on the number of particles. It is good contrast with the fact that the performance of force calculations accelerated by GPUs depends strongly on the number of particles. Substantially weak dependence of the performance on the number of particles is suitable to collisionless N-body simulations, since these simulations are usually performed with sophisticated N-body solvers such as Tree- and TreePM-methods combined with an individual timestep scheme. Collisionless N-body simulations accelerated with our library have significant advantage over those accelerated by GPUs, especially on massively parallel environments.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。