【第21期】观点:人工智能到底用 GPU?还是用 FPGA?

poj 1502 MPI Maelstrom

原创 2013年12月03日 20:12:30

MPI Maelstrom

Time Limit: 1000MS

 

Memory Limit: 10000K

Total Submissions: 2367

 

Accepted: 1390

Description

BIT has recently takendelivery of their new supercomputer, a 32 processor Apollo Odyssey distributedshared memory machine with a hierarchical communication subsystem. ValentineMcKee's research advisor, Jack Swigert, has asked her to benchmark the newsystem. 
``Since the Apollo is a distributed shared memory machine, memory access andcommunication times are not uniform,'' Valentine told Swigert. ``Communicationis fast between processors that share the same memory subsystem, but it isslower between processors that are not on the same subsystem. Communicationbetween the Apollo and machines in our lab is slower yet.'' 

``How is Apollo's port of the Message Passing Interface (MPI) working out?''Swigert asked. 

``Not so well,'' Valentine replied. ``To do a broadcast of a message from oneprocessor to all the other n-1 processors, they just do a sequence of n-1sends. That really serializes things and kills the performance.'' 

``Is there anything you can do to fix that?'' 

``Yes,'' smiled Valentine. ``There is. Once the first processor has sent themessage to another, those two can then send messages to two other hosts at thesame time. Then there will be four hosts that can send, and so on.'' 

``Ah, so you can do the broadcast as a binary tree!'' 

``Not really a binary tree -- there are some particular features of our networkthat we should exploit. The interface cards we have allow each processor tosimultaneously send messages to any number of the other processors connected toit. However, the messages don't necessarily arrive at the destinations at thesame time -- there is a communication cost involved. In general, we need totake into account the communication costs for each link in our networktopologies and plan accordingly to minimize the total time required to do abroadcast.''

Input

The input will describethe topology of a network connecting n processors. The first line of the inputwill be n, the number of processors, such that 1 <= n <= 100. 

The rest of the input defines an adjacency matrix, A. The adjacency matrix issquare and of size n x n. Each of its entries will be either an integer or thecharacter x. The value of A(i,j) indicates the expense of sending a messagedirectly from node i to node j. A value of x for A(i,j) indicates that amessage cannot be sent directly from node i to node j. 

Note that for a node to send a message to itself does not require networkcommunication, so A(i,i) = 0 for 1 <= i <= n. Also, you may assume thatthe network is undirected (messages can go in either direction with equal overhead),so that A(i,j) = A(j,i). Thus only the entries on the (strictly) lowertriangular portion of A will be supplied. 

The input to your program will be the lower triangular section of A. That is,the second line of input will contain one entry, A(2,1). The next line willcontain two entries, A(3,1) and A(3,2), and so on.

Output

Your program shouldoutput the minimum communication time required to broadcast a message from thefirst processor to all the other processors.

Sample Input

5

50

30 5

100 20 50

10 x x 10

Sample Output

35

题目就是Dijkstra算法题, 题意就是求第一个点到各个点最短路径中的最大值。

  学到直接把字符串改为整型,长整型:

atof()             将字符串转换成浮点数值 
atoi()                 将字符串转换成整数值 
atol()                 将字符串转换成长整数值 
strtod()             将字符串转换成双精度型数值 
strtol()             将字符串转换成长型数值 

输入的时候需要用到的小技巧,直接调用void dijkstra()就可以 了。。。

#include<cstdio>

#include<iostream>

#include<string>

#include<cstring>

#include<algorithm>

#include<iostream>

#include<stdio.h>

#include<algorithm>

#include<string.h>

#include<stdlib.h>

using namespace std;

# define INF  999999999;

const int maxn=1000;

intdis[maxn],g[maxn][maxn],n;

bool v[maxn];

char t[maxn];

int min(int a,intb){

    return a<b?a:b;

}

 

int main(){

    while(cin>>n){

         for(int i=1;i<=n;i++){

            for(int j=1;j<=n;j++){

                if(i==j){

                    g[i][j]=0;

                    //g[j][i]=0;

                }

                else{

                    g[i][j]=INF;

                    //g[j][i]=INF;

                }

            }

         }

         for(int i=2;i<=n;i++){

            for(int j=1;j<=i-1;j++){

                scanf("%s",&t);

                if(t[0]=='x'){

                    g[j][i]=INF;

                    g[i][j]=INF;

                }

                else{

                   g[j][i]=atoi(t);  // atoi();将字符串转为整型。

                   g[i][j]=atoi(t);

                }

            }

 

         }

         for(int i=1;i<=n;i++){

         dis[i]=INF;

         }

         dis[1]=0;

         memset(v,0,sizeof v);

         for(int i=1;i<=n;i++){

            int mark=-1,mindis=INF;

            for(int j=1;j<=n;j++){

               if(!v[j]&&dis[j]<mindis){

                    mindis=dis[j];

                    mark=j;

                }

            }

            v[mark]=1;

            for(int j=1;j<=n;j++){

                if(!v[j]){

                   dis[j]=min(dis[j],dis[mark]+g[mark][j]);

                }

            }

         }

 

         /*for(int j=1;j<=n;j++){

            cout<<dis[j]<<endl;

         }*/

         int themax=0;

         for(int j=2;j<=n;j++){

            if(themax<dis[j])

                themax=dis[j];

         }

         printf("%d\n",themax);

    }

 

}


版权声明:本文为博主原创文章,未经博主允许不得转载。 举报

相关文章推荐

poj 1502 MPI Maelstrom(最短路)

poj 1502 MPI MaelstromDescription BIT has recently taken delivery of their new supercomputer, a 32 ...

POJ 1502 MPI Maelstrom(Dijkstra)

POJ 1502 MPI Maelstrom(Dijkstra) http://poj.org/problem?id=1502 题意: (整个第一段题意描述都是无关的

POJ--1502:MPI Maelstrom (最短路径:Dijkstra算法 & Floyd算法)

题目源地址:http://poj.org/problem?id=1502 程序源代码:

MPI集群搭建指南

由于有时需要使用到MPI集群来进行迭代运算,就尝试着自己去搭建着MPI集群。不过这也是到去年工作的事情了。我现在凭我的记忆去把它回忆出来。   从零开始,搭建MPI集群是一件很费周折的事情。MPI是一种编程模型,业界有很多种实现,其中开源版本就有好几个版本。对比过openfabrics公司提供的<a href="https://www.openfabrics.org/resources/ofed-for-linux-ofed

poj 1502 MPI Maelstrom

MPI Maelstrom (Message Passing Interface ) n台电脑,要实现广播信息。每一台电脑能同时传播信息给所有与它相邻的电脑。 给出连接矩阵的下三角,'x'表示无法...
收藏助手
不良信息举报
您举报文章:深度学习:神经网络中的前向传播和反向传播算法推导
举报原因:
原因补充:

(最多只允许输入30个字)