检索结果-内蒙古大学图书馆

您好，读者！请登录

内蒙古大学图书馆

首页
概况
党建
资源
服务
科研支持
- 论文收录引用证明
- 科技查新
知识产权
档案馆
帮助

咨询与建议

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

您的常用邮箱：*

您的手机号码：*

问题描述：

当前已输入0个字，您还可以输入200个字

全部搜索
期刊论文
图书
学位论文
标准
纸本馆藏
外文资源发现
数据库导航
超星发现

高级检索

时间限定

出版年份：

文献类型

图书期刊文献学位论文多媒体

馆藏选择

电子馆藏纸本馆藏

核心期刊

全部期刊 SCI 收录期刊 SSCI 收录期刊 EI 收录期刊 CSCD 收录期刊 CSSCI 收录期刊

语言

中文英文

文献类型

期刊文献图书学位论文标准纸本馆藏

帮助

文字说明：

T=题名（书名、题名），A=作者（责任者），K=主题词，P=出版物名称，PU=出版社名称，O=机构（作者单位、学位授予单位、专利申请人），L=中图分类号，C=学科分类号，U=全部字段，Y=年（出版发行年、学位年度、标准发布年）

检索规则说明：

AND代表“并且”；OR代表“或者”；NOT代表“不包含”；(注意必须大写,运算符两边需空一格)

检索范例：

范例一：(K=图书馆学 OR K=情报学) AND A=范并思 AND Y=1982-2016
范例二：P=计算机应用与软件 AND (U=C++ OR U=Basic) NOT K=Visual AND Y=2011-2016

分类表

所选分类

>> <<

限定检索结果

文献类型

10,938 篇 会议
403 篇 期刊文献
26 册 图书

馆藏范围

11,367 篇 电子文献
0 种 纸本馆藏

日期分布

学科分类号

5,439 篇 工学
- 5,210 篇 计算机科学与技术...
- 2,511 篇 软件工程
- 861 篇 电气工程
- 688 篇 信息与通信工程
- 350 篇 电子科学与技术（可...
- 328 篇 控制科学与工程
- 186 篇 网络空间安全
- 58 篇 机械工程
- 56 篇 生物医学工程（可授...
- 55 篇 动力工程及工程热...
- 49 篇 生物工程
- 48 篇 建筑学
- 43 篇 仪器科学与技术
- 39 篇 光学工程
- 38 篇 土木工程
- 37 篇 测绘科学与技术
- 33 篇 力学（可授工学、理...
- 32 篇 安全科学与工程
1,324 篇 理学
- 1,122 篇 数学
- 151 篇 统计学（可授理学、...
- 106 篇 物理学
- 104 篇 系统科学
- 63 篇 生物学
- 44 篇 化学
- 38 篇 地球物理学
996 篇 管理学
- 835 篇 管理科学与工程(可...
- 304 篇 工商管理
- 196 篇 图书情报与档案管...
64 篇 经济学
- 63 篇 应用经济学
34 篇 法学
- 33 篇 社会学
24 篇 医学
21 篇 农学
9 篇 教育学
4 篇 文学
1 篇 军事学

主题

1,684 篇 parallel process...
1,192 篇 concurrent compu...
1,133 篇 distributed comp...
1,039 篇 distributed proc...
996 篇 computer science
901 篇 computer archite...
823 篇 computational mo...
751 篇 application soft...
684 篇 hardware
521 篇 parallel algorit...
497 篇 parallel program...
475 篇 runtime
462 篇 algorithm design...
441 篇 processor schedu...
433 篇 graphics process...
427 篇 parallel process...
419 篇 conferences
419 篇 scalability
395 篇 libraries
367 篇 costs

机构

33 篇 ibm thomas j. wa...
31 篇 school of comput...
27 篇 college of compu...
25 篇 oak ridge nation...
23 篇 oak ridge natl l...
22 篇 pacific northwes...
22 篇 oak ridge nation...
20 篇 argonne national...
19 篇 department of co...
18 篇 georgia inst tec...
18 篇 pacific northwes...
17 篇 department of co...
17 篇 virginia tech de...
16 篇 department of co...
16 篇 department of co...
16 篇 department of co...
15 篇 school of comput...
15 篇 sandia national ...
15 篇 georgia institut...
14 篇 department of co...

作者

38 篇 jack dongarra
29 篇 dongarra jack
24 篇 bader david a.
21 篇 nakano koji
20 篇 a. choudhary
19 篇 feng wu-chun
19 篇 p. sadayappan
18 篇 zomaya albert y.
16 篇 hoefler torsten
16 篇 schulz martin
15 篇 ito yasuaki
15 篇 s.k. das
15 篇 azad ariful
15 篇 talbi el-ghazali
14 篇 gagan agrawal
14 篇 h.j. siegel
14 篇 prasad sushil k.
14 篇 david a. bader
13 篇 s. ranka
13 篇 kishore kothapal...

语言

11,342 篇 英文
17 篇 中文
8 篇 其他

检索条件"任意字段=International Parallel and Distributed Processing Symposium"

共 11367 条记录，以下是121-130 订阅

全选清除本页清除全部题录导出标记到"检索档案"

详细简洁

排序：

Cloud Services Enable Efficient AI-Guided Simulation Workflows across Heterogeneous Resources

Cloud Services Enable Efficient AI-Guided Simulation Workflo...

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Ward, Logan Pauloski, J. Gregory Hayot-Sasson, Valerie Chard, Ryan Babuji, Yadu Sivaraman, Ganesh Choudhury, Sutanay Chard, Kyle Thakur, Rajeev Foster, Ian Argonne Natl Lab Lemont IL 60439 USA Univ Chicago Chicago IL 60637 USA Pacific Northwest Natl Lab Richland WA 99354 USA

ISBN: (纸本)9798350311990

Applications that fuse machine learning and simulation can benefit from the use of multiple computing resources, with, for example, simulation codes running on highly parallel supercomputers and AI training and inference tasks on specialized accelerators. Here, we present our experiences deploying two AI-guided simulation workflows across such heterogeneous systems. A unique aspect of our approach is our use of cloud-hosted management services to manage challenging aspects of cross-resource authentication and authorization, function-as-a-service (FaaS) function invocation, and data transfer. We show that these methods can achieve performance parity with systems that rely on direct connection between resources. We achieve parity by integrating the FaaS system and data transfer capabilities with a system that passes data by reference among managers and workers, and a user-configurable steering algorithm to hide data transfer latencies. We anticipate that this ease of use can enable routine use of heterogeneous resources in computational science.

关键词： Heterogeneous Computing Function-as-a-Service Machine Learning distributed Systems Computational Steering

来源：评论

学校读者我要写书评

暂无评论

ByteTransformer: A High-Performance Transformer Boosted for Variable-Length Inputs 37

ByteTransformer: A High-Performance Transformer Boosted for ...

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Zhai, Yujia Jiang, Chengquan Wang, Leyuan Jia, Xiaoying Zhang, Shang Chen, Zizhong Liu, Xin Zhu, Yibo Univ Calif Riverside Riverside CA USA ByteDance Ltd Beijing Peoples R China NVIDIA Corp Santa Clara CA USA

ISBN: (纸本)9798350337662

Transformers have become keystone models in natural language processing over the past decade. They have achieved great popularity in deep learning applications, but the increasing sizes of the parameter spaces required by transformer models generate a commensurate need to accelerate performance. Natural language processing problems are also routinely faced with variable-length sequences, as word counts commonly vary among sentences. Existing deep learning frameworks pad variable-length sequences to a maximal length, which adds significant memory and computational overhead. In this paper, we present ByteTransformer, a high-performance transformer boosted for variable-length inputs. We propose a padding-free algorithm that liberates the entire transformer from redundant computations on zero padded tokens. In addition to algorithmic-level optimization, we provide architecture-aware optimizations for transformer functional modules, especially the performancecritical algorithm Multi-Head Attention (MHA). Experimental results on an NVIDIA A100 GPU with variable-length sequence inputs validate that our fused MHA outperforms PyTorch by 6.13x. The end-to-end performance of ByteTransformer for a forward BERT transformer surpasses state-of-the-art transformer frameworks, such as PyTorch JIT, TensorFlow XLA, Tencent TurboTransformer, Microsoft DeepSpeed-Inference and NVIDIA FasterTransformer, by 87%, 131%, 138%, 74% and 55%, respectively. We also demonstrate the general applicability of our optimization methods to other BERT-like models, including ALBERT, DistilBERT, and DeBERTa.

关键词： Transformer BERT Multi-head Attention Large Language Models Natural Language processing NVIDIA GPU CUTLASS

来源：评论

学校读者我要写书评

暂无评论

parallel Inference of Phylogenetic Stands with Gentrius

Parallel Inference of Phylogenetic Stands with Gentrius

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Togkousidis, Anastasis Chernomor, Olga Stamatakis, Alexandros Heidelberg Inst Theoret Studies Computat Mol Evolut Grp Heidelberg Germany Karlsruhe Inst Technol Inst Theoret Informat Karlsruhe Germany Univ Vienna Ctr Integrat Bioinformat Vienna CIBIV Max Perutz Labs Vienna Austria Med Univ Vienna Vienna Bio Ctr VBC Vienna Austria Fdn Res & Technol Hellas Inst Comp Sci Biodivers Comp Grp Iraklion Greece

ISBN: (纸本)9798350311990

Multi-locus datasets are frequently used to infer phylogenies instead of using single locus. Missing data constitute a common challenge in such datasets as they can lead to stands, that is, sets of trees that are compatible with the incomplete per-locus trees. Under many common criteria the trees from one stand have identical score. Hence, identifying stands and determining their sizes is of crucial importance for a robust phylogenetic analysis. Recently, Chernomor et al. published Gentrius, a branch-and-bound algorithm that enumerates all stand trees given a set of unrooted incomplete locus trees. Despite its efficiency, the pattern and proportion of missing data in multi-locus datasets can still induce extremely long execution times. Here, we introduce the parallel version of the Gentrius algorithm. Our parallelization deploys a thread-pooling mechanism that maintains threads that finish early in busy-wait mode, such that they can contribute to solving long-running tasks. Thereby, we substantially reduce load imbalance and attain high parallel efficiency. Our performance assessment up to 16 cores yields linear parallel speedups on both, simulated, and empirical data. The parallel version of Gentrius is available as open source code under GNU GPL at https://***/togkousa/iqtree2/tree/terragen. All data we used for our analyses, are available for download at https://***/exelixis/material/***.

关键词： phylogenetic stands phylogenetic terraces multi-locus data parallel computing shared memory parallelism

来源：评论

学校读者我要写书评

暂无评论

FaultyRank: A Graph-based parallel File System Checker 37

FaultyRank: A Graph-based Parallel File System Checker

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Kamat, Saisha Islam, Abdullah Al Raqibul Zheng, Mai Dai, Dong Univ N Carolina Dept Comp Sci Charlotte NC 28223 USA Iowa State Univ Dept Elect & Comp Engn Ames IA USA

ISBN: (纸本)9798350337662

Similar to local file system checkers such as e2fsck for Ext4, a parallel file system (PFS) checker ensures the file system's correctness. The basic idea of file system checkers is straightforward: important metadata are stored redundantly in separate places for cross-checking;inconsistent metadata will be repaired or overwritten by its 'more correct' counterpart, which is defined by the developers. Unfortunately, implementing the idea for PFSes is non-trivial due to the system complexity. Although many popular parallel file systems already contain dedicated checkers (e.g., LFSCK for Lustre, BeeGFS-FSCK for BeeGFS, mmfsck for GPFS), the existing checkers often cannot detect or repair inconsistencies accurately due to one fundamental limitation: they rely on a fixed set of consistency rules predefined by developers, which cannot cover the various failure scenarios that may occur in practice. In this study, we propose a new graph-based method to build PFS checkers. Specifically, we model important PFS metadata into graphs, then generalize the logic of cross-checking and repairing into graph analytic tasks. We design a new graph algorithm, FaultyRank, to quantitatively calculate the correctness of each metadata object. By leveraging the calculated correctness, we are able to recommend the most promising repairs to users. Based on the idea, we implement a prototype of FaultyRank on Lustre, one of the most widely used parallel file systems, and compare it with Lustre's default file system checker LFSCK. Our experiments show that FaultyRank can achieve the same checking and repairing logic as LFSCK. Moreover, it is capable of detecting and repairing complicated PFS consistency issues that LFSCK can not handle. We also show the performance advantage of FaultyRank compared with LFSCK. Through this study, we believe FaultyRank opens a new opportunity for building PFS checkers effectively and efficiently.

关键词： checker file system checker graph lustre

来源：评论

学校读者我要写书评

暂无评论

MPipeMoE: Memory Efficient MoE for Pre-trained Models with Adaptive Pipeline parallelism 37

MPipeMoE: Memory Efficient MoE for Pre-trained Models with A...

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Zhang, Zheng Yang, Donglin Xia, Yaqi Ding, Liang Tao, Dacheng Zhou, Xiaobo Cheng, Dazhao Wuhan Univ Sch Comp Sci Wuhan Peoples R China Nvidia Corp Santa Clara CA USA JD com Inc JD Explore Acad Beijing Peoples R China Univ Macau Zhuhai Peoples R China

ISBN: (纸本)9798350337662

Recently, Mixture-of-Experts (MoE) has become one of the most popular techniques to scale pre-trained models to extraordinarily large sizes. Dynamic activation of experts allows for conditional computation, increasing the number of parameters of neural networks, which is critical for absorbing the vast amounts of knowledge available in many deep learning areas. However, despite the existing system and algorithm optimizations, there are significant challenges to be tackled when it comes to the inefficiencies of communication and memory consumption. In this paper, we present the design and implementation of MPipeMoE, a high-performance library that accelerates MoE training with adaptive and memory-efficient pipeline parallelism. Inspired by that the MoE training procedure can be divided into multiple independent sub-stages, we design adaptive pipeline parallelism with an online algorithm to configure the granularity of the pipelining. Further, we analyze the memory footprint breakdown of MoE training and identify that activations and temporary buffers are the primary contributors to the overall memory footprint. Toward memory efficiency, we propose memory reusing strategies to reduce memory requirements by eliminating memory redundancies, and develop an adaptive selection component to determine the optimal strategy that considers both hardware capacities and model characteristics at runtime. We implement MPipeMoE upon PyTorch and evaluate it with common MoE models in a physical cluster consisting of 8 NVIDIA DGX A100 servers. Compared with the state-of-art approach, MPipeMoE achieves up to 2.8x speedup and reduces memory footprint by up to 47% in training large models.

关键词： Mixture of Experts Pipeline parallelism distributed Training Memory Efficiency

来源：评论

学校读者我要写书评

暂无评论

Enabling Multi-threading in Heterogeneous Quantum-Classical Programming Models

Enabling Multi-threading in Heterogeneous Quantum-Classical ...

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Hayashi, Akihiro Adams, Austin Young, Jeffrey McCaskey, Alexander Dumitrescu, Eugene Sarkar, Vivek Conte, Thomas M. Georgia Inst Technol Atlanta GA 30332 USA NVIDIA Corp Santa Clara CA USA Oak Ridge Natl Lab Oak Ridge TN USA

ISBN: (纸本)9798350311990

While quantum computers enable significant performance improvements for certain classes of applications, building a well-defined programming model has been a pressing issue. In this paper, we address some of the key limitations to realizing a generic heterogeneous parallel programming model for quantumclassical heterogeneous platforms. We discuss our experience in enabling user-level multi-threading in QCOR [1] as well as challenges that need to be addressed for programming future quantum-classical systems. Specifically, we discuss our design and implementation of introducing C++-based parallel constructs to enable 1) parallel execution of a quantum kernel with std::thread and 2) asynchronous execution with std::async. To do so, we provide a detailed overview of the current implementation of the QCOR programming model and runtime, and discuss how we add 1) thread-safety to some of its user-facing API routines, and 2) increase parallelism in QCOR by removing data races that inhibit multi-threading so as to better utilize available computing resources. We also present preliminary performance results with the Quantum++ [2] back end on a single-node Ryzen9 3900X machine that has 12 physical cores (24 hardware threads) with 128GB of RAM. The results show that running two Bell kernels with 12 threads per kernel in parallel outperforms running the kernels one after the other each with 24 threads (1.63x improvement). In addition, we observe the same trend when running two Shor's algorthm kernels in parallel (1.22x faster than executing the kernels one after the other). Furthermore, the parallel version is better in terms of strong scalability. We believe that our design, implementation, and results will open up an opportunity not only for 1) enabling quicker prototyping of parallel-aware quantum-classical algorithms on quantum circuit simulators in the short-term, but also for 2) realizing a generic parallel programming model for quantumclassical heterogeneous platforms in

关键词： Quantum-Classical Programming Models parallel Programming Models QCOR Heterogeneous Computing

来源：评论

学校读者我要写书评

暂无评论

EDDIS: Accelerating distributed Data -parallel DNN Training for Heterogeneous GPU Cluster

EDDIS: Accelerating Distributed Data -Parallel DNN Training ...

引用

1st international Conference on Smart Energy Systems and Artificial Intelligence (SESAI)

作者： Ahn, Shinyoung Ahn, Hooyoung Choi, Hyeonseong Lee, Jaehyun ETRI Supercomp Syst Res Sect Daejeon South Korea MangoBoost Software Team Seoul South Korea Puzzle AI Inc Res Inst Daejeon South Korea

ISBN: (纸本)9798350364613;9798350364606

EDDIS is a novel distributed deep learning library designed to efficiently utilize heterogeneous GPU resources for training deep neural networks (DNNs), addressing scalability and conuminication challenges in distributed training environments. It offers three training modes (synchronous, asynchronous, and hybrid) and supports TensorFlow and PyTorch frameworks. EDDIS significantly accelerates DNN training in heterogeneous GPU settings, achieving up to 17.5x faster training with 16 nodes compared to a single node. Remarkably, the Hybrid training mode surpasses Horovod, achieving training speeds 2.53 times faster for the ResNet50 model.

关键词： Terms Deep Neural Network DNN Heterogeneous GPU distributed Training Data -parallel Training EDDIS SoftMemoryBox SSGD EASGD Hybrid SGD

来源：评论

学校读者我要写书评

暂无评论

Constrained Approximate Query processing with Error and Response Time-Bound Guarantees for Efficient Big Data Analytics 24

Constrained Approximate Query Processing with Error and Resp...

引用

33rd international symposium on High-Performance parallel and distributed Computing (HPDC)

作者： Kim, Sungsoo Park, Choon Seo Lee, Taewhi Nam, Kihyuk ETRI Daejeon South Korea

ISBN: (纸本)9798400704130

Approximate query processing (AQP) is a technique for obtaining approximate answers to queries over large datasets. AQP techniques trade off accuracy for speed, making them ideal for scenarios where exact answers are not required or the cost of obtaining exact answers is prohibitive. This paper proposes a novel machine learning (ML)-based AQP framework that leverages both generative and inferential ML models to improve accuracy and efficiency. The framework first constructs a generative ML model that learns the underlying data distribution and then generates synthetic data that follows the same distribution. The proposed framework also includes a mechanism for constrained approximate query processing (CAQP) with bounded errors and bounded response times. This allows users to specify the desired error bound for the results of an aggregation query. The framework then selects a subset of the synthetic data that is guaranteed to satisfy the error bound. An evaluation of the proposed framework using the Instacart benchmark dataset and queries demonstrates significant efficiency improvements in AQP compared to existing techniques.

关键词： approximate query processing

来源：评论

学校读者我要写书评

暂无评论

An Efficient 2D Method for Training Super-Large Deep Learning Models 37

An Efficient 2D Method for Training Super-Large Deep Learnin...

引用

37th IEEE international parallel and distributed processing symposium (IPDPS)

作者： Xu, Qifan You, Yang Univ Calif Los Angeles Los Angeles CA 90095 USA Natl Univ Singapore Singapore Singapore

ISBN: (纸本)9798350337662

Since the rise of Transformer [22] and BERT [6], large language models [7, 12] have been proposed and shown unprecedented performance in tasks like translation, classification, and text generation. However, due to the memory constraint, model parallelism must be used to split the model across multiple processors. Interlayer partition, intra-layer partition, and sparse activation are the major approaches to achieve model parallelism. Among them, interlayer partition [10, 11] often requires the model to be explicitly expressed as a stack of sub-modules, the number of which equals to the number of processors, and would introduce either gradient staleness or bubble overhead;while the sparse activation [12] is primarily designed for Google TPU cluster and hard to deploy on GPU servers, intra-layer partition [17], especially Megatron-LM [18], can be easily deployed on GPU servers and has been adopted in subsequent works like Turing-NLG and M6. Though as pioneers of intra-layer parallelism, they still show memory redundancy and sub-optimal communication efficiency, which reveals the space for further improvements. In this work, we leverage SUMMA [21] and propose Optimus, a highly efficient and scalable paradigm for training super-large language models. In Optimus, activations and gradients are partitioned and distributed along processors all the way through forward and backward propagations, with hardly any memory redundancy. The isoefficiency of communication in pure model parallelism improves from W similar to p(3) for Megatron-LM, to W similar to (root p logp)(3) for our Optimus. This framework is implemented with open-source deep learning framework, PyTorch, and consolidates existing techniques such as mixed precision training [13], activation checkpointing [5], and data parallelism. In experiments on TACC Frontera supercomputers, Optimus shows 1.48x the speed for training, 1.78x speed for inference, and 8x the maximum batch size over Megatron-LM on 64 GPUs in pure mod

关键词： matrix-matrix multiplication distributed training neural networks natural language processing

来源：评论

学校读者我要写书评

暂无评论

IDIOMS: Index-powered distributed Object-centric Metadata Search for Scientific Data Management 24

IDIOMS: Index-powered Distributed Object-centric Metadata Se...

引用

24th IEEE/ACM international symposium on Cluster, Cloud, and Internet Computing (CCGrid)

作者： Zhang, Wei Tang, Houjun Byna, Suren Lawrence Berkeley Natl Lab Berkeley CA 94720 USA Ohio State Univ Columbus OH USA

ISBN: (纸本)9798350395679;9798350395662

Affix-oriented metadata search is one of the essential fuzzy search capabilities that allow users to find data of interest in their voluminous data set with incomplete query conditions. With the recent transition towards object-centric data management systems in the science community, there is a paramount need for the support of such features in distributed settings. However, existing metadata search solutions either do not support efficient affix-oriented metadata search or do not suit well in a distributed setting of object-centric data management systems. To bridge this gap, we introduce IDIOMS, a metadata search solution underpinned by a distributed metadata index, meticulously designed to enable high-performance affix-oriented metadata search for parallel object-centric storage. One of the standout features of IDIOMS is its efficiency in supporting four distinct types of highly demanded metadata queries. Furthermore, IDIOMS is flexibly catering to both independent and collective metadata search operations. Our experimental comparisons with SoMeta, a state-of-the-art metadata query method, demonstrate more than 400x performance boost for independent queries and up to 300x performance improvements for collective queries, while keeping a small index management overhead.

关键词： Metadata Search Metadata Management parallel Data processing distributed Data Management Scientific Data Management

来源：评论

学校读者我要写书评

暂无评论

没有更多数据了...

全选清除本页清除全部题录导出标记到“检索档案”

共500页 << < 9 10 11 12 13 14 15 16 17 18 > >>

检索报告对象比较合并检索0

隐藏清空

合并搜索

回到顶部

执行限定条件

内容：

评分：

请选择保存的检索档案：

请选择收藏分类：

订阅名称：

通借通还

温馨提示：

图书名称：

借书校区：

取书校区：

手机号码：

邮箱地址：

一卡通帐号：

电话和邮箱必须正确填写，我们会与您联系确认。

联系人：

所在院系：

联系邮箱：

联系电话：

内蒙古自治区呼和浩特市赛罕区大学西街235号邮编: 010021

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：