检索结果-南通市图书馆

Towards optimized tensor code generation for deep learning on sunway many-core processor

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2024年第2期18卷 1-15页

作者： Mingzhen LI Changxi LIU Jianjin LIAO Xuegui ZHENG Hailong YANG Rujun SUN Jun XU Lin GAN Guangwen YANG zhongzhi luan Depei QIAN State Key Laboratory of Software Development Environment Beijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China National University of Singapore Singapore 119077Singapore State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi 214000China Science and Technology on Special System Simulation Laboratory Beijing Simulation Center Beijing 100854China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

The flourish of deep learning frameworks and hardware platforms has been demanding an efficient compiler that can shield the diversity in both software and hardware in order to provide application *** the existing deep learning compilers,TVM is well known for its efficiency in code generation and optimization across diverse hardware *** the meanwhile,the Sunway many-core processor renders itself as a competitive candidate for its attractive computational power in both scientific computing and deep learning *** paper combines the trends in these two ***,we propose swTVM that extends the original TVM to support ahead-of-time compilation for architecture requiring cross-compilation such as *** addition,we leverage the architecture features during the compilation such as core group for massive parallelism,DMA for high bandwidth memory transfer and local device memory for data locality,in order to generate efficient codes for deep learning workloads on *** experiment results show that the codes generated by swTVM achieve 1.79x improvement of inference latency on average compared to the state-of-the-art deep learning framework on Sunway,across eight representative *** work is the first attempt from the compiler perspective to bridge the gap of deep learning and Sunway processor particularly with productivity and efficiency in *** believe this work will encourage more people to embrace the power of deep learning and Sunwaymany-coreprocessor.

关键词： sunway processor deep learning compiler code generation performance optimization

Differentiating Data Collection for Cloud Environment Monitoring

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

China Communications 2014年第4期11卷 13-24页

作者： MENG You luan zhongzhi QIAN Depei Sino-German Joint Software Institute (JSI) Beihang University (BUAA) Beijing 100083 China

In a growing number of information processing applications,data takes the form of continuous data streams rather than traditional stored *** systems that seek to provide monitoring services in cloud environment must be prepared to deal gracefully with huge data collections without compromising system *** this paper,we show that by using a concept of urgent data,our system can shorten the response time for most 'urgent' queries while guarantee lower bandwidth *** argue that monitoring data can be treated *** data capture critical system events;the arrival of these data will significantly influence the monitoring reaction speed which is called urgent *** speed urgent data collections can help system to react in real time when facing fatal errors.A cloud environment in production,MagicCube,is used as a test *** experiments over both real world and synthetic traces show that when using urgent data,monitoring system can lower the response latency compared with existing monitoring approaches.

关键词： cloud computing cloud monitoring urgent data rule engine constraint

Coordinating workload balancing and power switching in renewable energy powered data center

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2016年第3期10卷 574-587页

作者： Xian LI Rui WANG zhongzhi luan Yi LIU Depei QIAN State Key Laboratory of Software Development Environment Beihang University Beijing 100191 China School of Computer Science and Engineering Beihang University Beijing 100191 China

There has been growing concern about energy consumption and environmental impact of datacenters. Some pioneers begin to power datacenters with renewable energy to offset carbon footprint. However, it is challenging to integrate intermittent renewable energy into datacenter power system. Grid-tied system is widely deployed in renewable energy powered datacenters. But the drawbacks （e.g. Harmonic dis- turbance and costliness） of grid tie inverter harass this design. Besides, the mixture of green load and brown load makes power management heavily depend on software measurement and monitoring, which often suffers inaccuracy. We propose DualPower, a novel power provisioning architecture that en- ables green datacenters to integrate renewable power supply without grid tie inverters. To optimize DualPower operation, we propose a specially designed power management frame- work to coordinate workload balancing with power supply switching. We evaluate three optimization schemes （LM, PS and JO） under different datacenter operation scenarios on our trace-driven simulation platform. The experimental results show that DualPower can be as efficient as grid-tied system and has good scalability. In contrast to previous works, Du- alPower integrates renewable power at lower cost and main- tains full availability of datacenter servers.

关键词： renewable energy green computing power pro-visioning power management

Improving multiprocessor performance with fine-grain coherence bypass

在线全文

学校读者我要写书评

暂无评论

Science China(Information Sciences) 2015年第1期58卷 84-98页

作者： WANG Hui WANG Rui luan zhongzhi QIAN XueHai QIAN DePei Sino-German Joint Software Institute School of Computer Science and EngineeringBeihang University University of Illinois Urbana-Champaign

Efficient and scalable cache coherence protocol is crucial to high-performance servers with sharedmemory. The directory-based cache coherence protocol is more desirable than the snooping-based protocol with respect to the scalability. However, even for the former protocol, scaling to a large number of cores is also challenging due to the additional area requirements of the directories. We observed that a significant percentage of the referenced memory blocks were only accessed by a single core(even in parallel applications) which could be considered as private memory blocks. An intuitive motivation from this observation is that memory blocks accessed by a single core do not require coherence maintenance. The issue is to identify the private block and track the change of its access pattern. We propose a novel hardware approach to(1) dynamically identify the shared memory blocks at the cache block level, and(2) bypass the coherence procedure for the private memory blocks. This approach increases the effectiveness of the directory-based approach and therefore improves the system performance. Experimental results showed that, our approach can on an average(1) avoid the coherence tracking of about 54% referenced memory blocks,(2) reduce the coherence overhead by 77%,(3) avoid 8% L2 cache misses, and(4) shorten the execution time of parallel applications by 13%.

关键词： many-core cache coherence private memory block fine-grain coherence high performance

A novel index system describing program runtime characteristics for workload consolidation

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2019年第3期13卷 489-499页

作者： Lin WANG Depei QIAN Rui WANG zhongzhi luan Hailong YANG Huaxiang ZHANG School of Information Science and Engineering Shandong Normal University Jinan 250014 China School of Computer Science and Engineering Beihang University Beijing 100191 China School of Data and Computer Science Sun Yat-sen University Guangzhou 510275 China Institute of Data Science and Technology Shandong Normal University Jinan 250014 China

Workload consolidation is a common method to improve the resource utilization in clusters or data centers. In order to achieve efficient workload consolidation, the runtime characteristics of a program should be taken into con-sideration in scheduling. In this paper, we propose a novel index system for efficiently describing the program runtime characteristics. With the help of this index system, programs can be classified by the following runtime characteristics: 1) dependence to multi-dimensional resources including CPU, disk I/O, memory and network I/O;and 2) impact and vulnerability to resource sharing embodied by resource usage and resource sensitivity. In order to verify the effectiveness of this novel index system in workload consolidation, a scheduling strategy, Sche-index, using the new index system for workload consolidation is proposed. Experiment results show that compared with traditional least-loaded scheduling strategy, Sche-index can improve both program performance and system resource utilization significantly.

关键词： index system runtime characteristics workload consolidation cluster scheduling

Field-induced Néel vector bi-reorientation of a ferrimagnetic insulator in the vicinity of compensation temperature

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chinese Physics B 2021年第2期30卷 481-486页

作者： Peng Wang Hui Zhao zhongzhi luan Siyu Xia Tao Feng Lifan Zhou College of Mathematics and Physics Qingdao University of Science and TechnologyQingdao 266061China National Laboratory of Solid State Microstructures Jiangsu Provincial Key Laboratory for Nanotechnology and Department of PhysicsNanjing UniversityNanjing 210093China

The spin Hall magnetoresistance(SMR)effect in Pt/Gd_(3)Fe_(5)O_(12)(Gd IG)bilayers was systematically *** sign of SMR changes twice with increasing magnetic field in the vicinity of the magnetization compensation point(TM)of Gd ***,conventional SMR theory predicts the invariant SMR sign in the heterostructure composed of a heavy metal film in contact with a ferromagnetic or antiferromagnetic *** conclude that this is because of the significant enhancement of the magnetic moment of the Gd sub-lattice and the unchanged moment of the Fe sub-lattice with a relatively large field,meaning that a small net magnetic moment is induced at *** a result,the Néel vector aligns with the field after the spin-flop transition,meaning that a bi-reorientation of the Néel vector is *** calculations based on the Néel’s theory and SMR theory also support our *** findings indicate that the Néel-vector direction of a ferrimagnet can be tuned across a wide range by a relatively low external field around TM.

关键词： spin Hall magnetoresistance ferrimagnets magnetic insulators magnetization switching

Accelerating the cryo-EM structure determination in RELION on GPU cluster

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2022年第3期16卷 21-39页

作者： Xin YOU Hailong YANG zhongzhi luan Depei QIAN Sino-German Joint Software Institute School of Computer Science and EngineeringBeihang UniversityBeijing 100191China State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi 214125

The cryo-electron microscopy(cryo-EM)is one of the most powerful technologies available today for structural *** RELION(Regularized Likelihood Optimization)implements a Bayesian algorithm for cryo-EM structure determination,which is one of the most widely used software in this *** researchers have devoted effort to improve the performance of RELION to satisfy the analysis for the ever-increasing volume of *** this paper,we focus on performance analysis of the most time-consuming computation steps in RELION and identify their performance bottlenecks for specific *** propose several performance optimization strategies to improve the overall performance of RELION,including optimization of expectation step,parallelization of maximization step,accelerating the computation of symmetries,and memory affinity *** experiment results show that our proposed optimizations achieve significant speedups of RELION across representative *** addition,we perform roofline model analysis to understand the effectiveness of our optimizations.

关键词： cryo-EM structure determination performance optimization GPU acceleration RELION

swSpAMM:optimizing large-scale sparse approximate matrix multiplication on Sunway Taihulight

在线全文

学校读者我要写书评

暂无评论

Frontiers of Computer Science 2023年第4期17卷 29-41页

作者： Xiaoyan LIU Yi LIU Bohong YIN Hailong YANG zhongzhi luan Depei QIAN State Key Laboratory of Software Development Environment Beijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China

Although matrix multiplication plays an essential role in a wide range of applications,previous works only focus on optimizing dense or sparse matrix *** Sparse Approximate Matrix Multiply(SpAMM)is an algorithm to accelerate the multiplication of decay matrices,the sparsity of which is between dense and sparse *** addition,large-scale decay matrix multiplication is performed in scientific applications to solve cutting-edge *** optimize large-scale decay matrix multiplication using SpAMM on supercomputers such as Sunway Taihulight,we present swSpAMM,an optimized SpAMM algorithm by adapting the computation characteristics to the architecture features of Sunway ***,we propose both intra-node and inter-node optimizations to accelerate swSpAMM for large-scale *** intra-node optimizations,we explore algorithm parallelization and block-major data layout that are tailored to better utilize the architecture advantage of Sunway *** inter-node optimizations,we propose a matrix organization strategy for better distributing sub-matrices across nodes and a dynamic scheduling strategy for improving load balance across *** compare swSpAMM with the existing GEMM library on a single node as well as large-scale matrix multiplication methods on multiple *** experiment results show that swSpAMM achieves a speedup up to 14.5×and 2.2×when compared to xMath library on a single node and 2D GEMM method on multiple nodes,respectively.

关键词： approximate calculation sunway processor performance optimization

从网格到“东数西算”:构建国家算力基础设施

在线全文

学校读者我要写书评

暂无评论

北京航空航天大学学报 2022年第9期48卷 1561-1574页

作者：钱德沛栾钟治刘轶北京航空航天大学计算机学院北京100083

简要回顾了几十年来计算机使用方式的变迁,介绍了基于网络计算技术的国家高性能计算基础设施CNGrid的设计与实现。讨论了在“东数西算”战略工程背景下中国算力发展的新趋势,以及国家算力基础设施发展面临的新的技术挑战,并对中国未来... 详细信息

简要回顾了几十年来计算机使用方式的变迁,介绍了基于网络计算技术的国家高性能计算基础设施CNGrid的设计与实现。讨论了在“东数西算”战略工程背景下中国算力发展的新趋势,以及国家算力基础设施发展面临的新的技术挑战,并对中国未来超算应用生态和算力基础设施建设提出了展望。

关键词：高性能计算基础设施网格计算 CNGrid 东数西算