LARGE-SCALE PARALLEL GRAPH-BASED SIMULATIONS - MATSim

More documents

Recommendations

Info

Chapter 4 Parallel Queue Model 4.1 Introduction Serial computation has been around for years. With this traditional computing manner, problems run on a single computer/computing node, instructions of a program are executed one after the other by the CPU, only one instruction may be executed at a time. Data transmission through hardware, which is limited by the speed of light [83], determines the speed of a serial computer. In addition to physical constraints, there are also economic limitations since it is increasingly expensive to make a single processor faster. These limitations make it harder to build faster serial computers by saturating the performance of serial computers. Ultimately, the agent-based simulation of large scale transportation scenarios are concerned. A typical scenario would be the 24-hour (about seconds) simulation of a metropolitan area consisting of 10 million travelers. Typical computational speeds of ¡ ¢¡ traffic flow simulations with 1-second update steps are 100 000 vehicles in real time [58, 56, 68]. This results in a computation time of ¡ ¢ ¡£¢£¢ ¡ ¢ ¢ seconds ¤ days. This number is just a rough estimate and subject to the following changes: Increases in CPU speed will reduce the number; ¡ ¢ ¡ ¢£¢ more realistic driving logic will increase the number; smaller time steps [64, 84] will increase the number. This means that such a traffic flow simulation running on a single computing node is too slow for practical or academic treatment of large scale problems. In addition, computer time is needed for activity generation, route generation, learning, etc. In consequence, it makes sense to explore parallel/distributed computing as an option. Parallel/distributed computing has the advantages of using non-local resources, competitive cost/performance ratio and overcoming finite memory constraint of single computers that are subject to. In parallel computing computational problems are solved by using several computing resources which may consist of a single computer with multiple processors or a number of computers connected through a network, which is called a PC cluster, or a combination of both. In order to solve a computational problem through parallel computing, one must think about (i) how to partition the tasks into subtasks, and (ii) how to provide the data exchange between the subtasks. Before explaining these issues, parallel architectures will be discussed in the following paragraphs. The categorization of parallel computers has been done in many different ways, among which Flynn’s Classical Taxonomy [83] is the one most commonly used. This classification 37
depends on the dimensions (single or multiple) of instructions and data. Each combination gives a different category: Single Instruction Single Data (SISD): Same instruction stream executes one data stream which causes deterministic execution. Most PCs, single CPU workstations and mainframes have this feature. Single Instruction Multiple Data (SIMD): Same instruction stream executes different data on different computing nodes. Examples are CM-2, IBM 9000, Cray C90. Multiple Instruction Single Data (MISD): Different instruction streams run on the same data. It is the one less commonly used. Multiple Instruction Multiple Data (MIMD): The most popular type of parallel computers. Each processor runs a different set of instructions on different data. Execution can be synchronous or asynchronous, deterministic or non-deterministic. Most supercomputers and PC clusters are of this type. 4.1.1 Message Exchange This work concentrates on clusters of coupled PCs, i.e., Linux [39] boxes connected through 100 Mbit Ethernet [77] Local Area Network (LAN). Using this type of clusters, which is costeffective, can achieve a performance close to that of a vector computer [57]. This is, in part, due to the fact that multi-agent simulations do not vectorize well, so that vector computers offer no particular advantages. Hence, PC clusters are expected to be the dominant high performance computing technology in the area of multi-agent traffic flow simulations for many years to come. With respect to data exchange between subtasks, there are, in general, two main approaches to inter-processor communication. One of them is called message passing between processors; the alternative is called shared-address space, where variables are kept in a common pool globally available to all processors. Each paradigm has its own advantages and disadvantages. In the shared-address space approach, all variables are globally accessible by all processors. Despite multiple processors operating independently, they share the same memory resources. The shared-address space approach makes it simpler for the user to achieve parallelism but since the memory bandwidth is limited, severe bottlenecks are inevitable with an increasing number of processors, or alternatively such shared memory parallel computers become very expensive. For those reasons, message passing is the focus point. In the message passing approach, there are independent cooperating processors. Each processor has a private local memory in order to keep the variables and data, and thus can access local data very rapidly. If an exchange of the information is needed between the processors, the processors communicate and synchronize by passing messages, which are simple send and receive instructions. Message passing can be imagined to be similar to sending a letter. The following phases happen during a message passing operation. 1. The message needs to be packed i.e. the computer is told which data needs to be sent. 2. The message is sent. 3. The message may then take some time on the network until it finally arrives in the receiver’s inbox. 4. The receiver has to officially receive the message, i.e. to take it out of the inbox. 38
Page 1 and 2: DISS. ETH NO. LARGE-SCALE PARALLEL
Page 3 and 4: Zusammenfassung Für das Modelliere
Page 5 and 6: Contents Abstract Zusammenfassung A
Page 7 and 8: 6.9 Results of the Buffered Events
Page 9 and 10: 4.10 Packing vehicle data with MPI
Page 11 and 12: List of Tables 3.1 Performance resu
Page 13 and 14: The strategical world: Concepts whi
Page 15 and 16: queues, on the other hand, does not
Page 17 and 18: Handling Constraints - Original alg
Page 19 and 20: node spatial queue buffer acc to ca
Page 21 and 22: Algorithm-3 for Vehicle Movement th
Page 23 and 24: - network - nodes - - - node i d =
Page 25 and 26: ) ¥ § ¡ £ ) ¥ ) ¥ ) ¡ £
Page 27 and 28: has the attributes such as type, le
Page 29 and 30: 3.2 The Benchmark The traffic flow
Page 31 and 32: make ‘ ‘ Nodes ’ ’ a c o n
Page 33 and 34: ¡ ¥ ¥ t y p e d e f v e c t o r
Page 35 and 36: 1024 RTR, Diff. Data Str. for Graph
Page 37 and 38: 1024 RTR, Diff. Data Str. for Parki
Page 39 and 40: push_back(O1) push_back(02) head =
Page 41 and 42: The sequence of attributes is not i
Page 43 and 44: ¡ ¡ ¡ c l a s s Plan / / data s
Page 45 and 46: Writing Time Explanation Raw Local
Page 47: RunTime Graph Data Parking-Waiting
Page 51 and 52: 4.2 Parallel Computing in Transport
Page 53 and 54: 350000 300000 250000 200000 150000
Page 55 and 56: ¢¡ ¥ ¡ * 6 §¡ ¥ ¢¡ ¥ 8
Page 57 and 58: ¥¦¨§ ¡¦©§ §¡ ¥ *
Page 59 and 60: 4.5 Experimental Results The parall
Page 61 and 62: ¡ ¥ ¡ If the simulation time s
Page 63 and 64: a long i n t e g e r f o r v e h i
Page 65 and 66: ¡ / / d e f i n e a s t r u c t co
Page 67 and 68: 1024 512 256 RTR for a 3-hour run o
Page 69 and 70: In Section 4.1.2, task parallelizat
Page 71 and 72: Chapter 5 Coupling the Traffic Simu
Page 73 and 74: 100% Initial Plans File Activity Ge
Page 75 and 76: ¡ ¡ ¡ p f map- - ; / / d e f i n
Page 77 and 78: ¡ char myline [MAXSIZE ] ; while (
Page 79 and 80: 5.3 Other Coupling Mechanisms Coupl
Page 81 and 82: pieces of code. For instance, a com
Page 83 and 84: 5.3.5 Module Coupling via Messages
Page 85 and 86: Chapter 6 Events Recorder 6.1 Intro
Page 87 and 88: A2 are collected by the other ER. S
Page 89 and 90: & $ ¡ & $ ¡ 3. TSs
Page 91 and 92: Algorithm A - Traffic Simulator Rep
Page 93 and 94: ¡ 6 ) © £ ¡ © ¤ ¡ 6 ) © (
Page 95 and 96: and ¤¢¤ ¢ ¡ ¢ ¡ ¡ ¢ ¡
Page 97 and 98: 256 64 Packing Only, Raw Events mem
Page 99 and 100:
£ ¥ Time in Secs 256 64 16 4 Send
Page 101 and 102:
Multicast Sending Only, Raw Events,
Page 103 and 104:
Time in Secs 256 64 16 4 1 Effectiv
Page 105 and 106:
¢ Writing Time Explanation Raw XML
Page 107 and 108:
700 600 500 TS-p TS-s ER-er ER-u ER
Page 109 and 110:
¡ ¡ ¦ ¦ © ¡ c r e a t e a C
Page 111 and 112:
i n t i n t e g e r i t e m ; doubl
Page 113 and 114:
When the events are reported to str
Page 115 and 116:
Chapter 7 Plans Server 7.1 Introduc
Page 117 and 118:
Algorithm A - Plans Server while no
Page 119 and 120:
i n t i n t e g e r i t e m ; doubl
Page 121 and 122:
) ) ) £ ¡ © §¦ ©¨ ¡ £ ¡
Page 123 and 124:
) © ¨ ¡ ¡ $ ¢ ¡ ¡ * ¢ $
Page 125 and 126:
For PSs, Send Only (IDs & StartLink
Page 127 and 128:
For TSs, Effective Receive + Unpack
Page 129 and 130:
64 32 16 8 4 2 1 0.5 0.25 For PSs,
Page 131 and 132:
Time nPSs nTSs OP Note 1.87s 1 1 pa
Page 133 and 134:
£ ¡ ¡ § ¤¡2 ) © ¡ ( ¡
Page 135 and 136:
100 Round Trip Travel Time 10 1 0.1
Page 137 and 138:
observe the Internet packet traffic
Page 139 and 140:
A probably better way to reduce the
Page 141 and 142:
[14] Berksekas D. and R. Gallager.
Page 143 and 144:
[51] MPI www page. www-unix.mcs.anl
Page 145 and 146:
[87] W. Willinger W.E. Leland, M. T
Page 147:
Large-scale multi-agent transportat
show all

LARGE-SCALE PARALLEL GRAPH-BASED SIMULATIONS - MATSim

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?