اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدAcceleration of the tree method with SIMD instruction set
210
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
We have developed a highly-tuned software library that accelerates the calculation of quadrupole terms in the Barnes-Hut tree code by use of a SIMD instruction set on the x86 architecture, Advanced Vector eXtensions 2 (AVX2). Our code is implemented as an extension of Phantom-GRAPE software library (Tanikawa et al. 2012, 2013) that significantly accelerates the calculation of monopole terms. If the same accuracy is required, the calculation of quadrupole terms can accelerate the evaluation of forces than that of only monopole terms because we can approximate gravitational forces from closer particles by quadrupole moments than by only monopole moments. Our implementation can calculate gravitational forces about 1.1 times faster in any system than the combination of the pseudoparticle multipole method and Phantom-GRAPE. Our implementation allows simulating homogeneous systems up to 2.2 times faster than that with only monopole terms, however, speed up for clustered systems is not enough because the increase of approximated interactions is insufficient to negate the increased calculation cost by computing quadrupole terms. We have estimated that improvement in performance can be achieved by the use of a new SIMD instruction set, AVX-512. Our code is expected to be able to accelerate simulations of clustered systems up to 1.08 times faster on AVX-512 environment than that with only monopole terms.
قيم البحث
اقرأ أيضاً
With the advent of increasingly complex hardware in real-time embedded systems (processors with performance enhancing features such as pipelines, cache hierarchy, multiple cores), many processors now have a set-associative L2 cache. Thus, there is a
need for considering cache hierarchies when validating the temporal behavior of real-time systems, in particular when estimating tasks worst-case execution times (WCETs). To the best of our knowledge, there is only one approach for WCET estimation for systems with cache hierarchies [Mueller, 1997], which turns out to be unsafe for set-associative caches. In this paper, we highlight the conditions under which the approach described in [Mueller, 1997] is unsafe. A safe static instruction cache analysis method is then presented. Contrary to [Mueller, 1997] our method supports set-associative and fully associative caches. The proposed method is experimented on medium-size and large programs. We show that the method is most of the time tight. We further show that in all cases WCET estimations are much tighter when considering the cache hierarchy than when considering only the L1 cache. An evaluation of the analysis time is conducted, demonstrating that analysing the cache hierarchy has a reasonable computation time.
In this paper, we study how certain conditions can affect the transformations on the states of the memory of a strict load-store Maurer ISA, when half of the data memory serves as the part of the operating unit.
Simple graph algorithms such as PageRank have recently been the target of numerous hardware accelerators. Yet, there also exist much more complex graph mining algorithms for problems such as clustering or maximal clique listing. These algorithms are
memory-bound and thus could be accelerated by hardware techniques such as Processing-in-Memory (PIM). However, they also come with non-straightforward parallelism and complicated memory access patterns. In this work, we address this with a simple yet surprisingly powerful observation: operations on sets of vertices, such as intersection or union, form a large part of many complex graph mining algorithms, and can offer rich and simple parallelism at multiple levels. This observation drives our cross-layer design, in which we (1) expose set operations using a novel programming paradigm, (2) express and execute these operations efficiently with carefully designed set-centric ISA extensions called SISA, and (3) use PIM to accelerate SISA instructions. The key design idea is to alleviate the bandwidth needs of SISA instructions by mapping set operations to two types of PIM: in-DRAM bulk bitwise computing for bitvectors representing high-degree vertices, and near-memory logic layers for integer arrays representing low-degree vertices. Set-centric SISA-enhanced algorithms are efficient and outperform hand-tuned baselines, offering more than 10x speedup over the established Bron-Kerbosch algorithm for listing maximal cliques. We deliver more than 10 SISA set-centric algorithm formulations, illustrating SISAs wide applicability.
Customization of processor architectures through Instruction Set Extensions (ISEs) is an effective way to meet the growing performance demands of embedded applications. A high-quality ISE generation approach needs to obtain results close to those ach
ieved by experienced designers, particularly for complex applications that exhibit regularity: expert designers are able to exploit manually such regularity in the data flow graphs to generate high-quality ISEs. In this paper, we present ISEGEN, an approach that identifies high-quality ISEs by iterative improvement following the basic principles of the well-known Kernighan-Lin (K-L) min-cut heuristic. Experimental results on a number of MediaBench, EEMBC and cryptographic applications show that our approach matches the quality of the optimal solution obtained by exhaustive search. We also show that our ISEGEN technique is on average 20x faster than a genetic formulation that generates equivalent solutions. Furthermore, the ISEs identified by our technique exhibit 35% more speedup than the genetic solution on a large cryptographic application (AES) by effectively exploiting its regular structure.
We have developed a new parallel tree method which will be called the forest method hereafter. This new method uses the sectional Voronoi tessellation (SVT) for the domain decomposition. The SVT decomposes a whole space into polyhedra and allows thei
r flat borders to move by assigning different weights. The forest method determines these weights based on the load balancing among processors by means of the over-load diffusion (OLD). Moreover, since all the borders are flat, before receiving the data from other processors, each processor can collect enough data to calculate the gravity force with precision. Both the SVT and the OLD are coded in a highly vectorizable manner to accommodate on vector parallel processors. The parallel code based on the forest method with the Message Passing Interface is run on various platforms so that a wide portability is guaranteed. Extensive calculations with 15 processors of Fujitsu VPP300/16R indicate that the code can calculate the gravity force exerted on 10^5 particles in each second for some ideal dark halo. This code is found to enable an N-body simulation with 10^7 or more particles for a wide dynamic range and is therefore a very powerful tool for the study of galaxy formation and large-scale structure in the universe.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
