FIR on Hadoop using hadoop-streaming

标签：style   blog   http   os   io   strong   for   2014   

Prepare Hadoop Streaming

Hadoop streaming allows you to create and run Map/Reduce jobs with any executable or script as the mapper and/or the reducer.

1.Download Hadoop Streaming fit for your hadoop version

For hadoop2.4.0, you can visit the following website and download the jar file:

http://mvnrepository.com/artifact/org.apache.hadoop/hadoop-streaming/2.4.0

Than put the jar file to the $HADOOP_HOME folder.

2.Hadoop-streaming test in my cluster:

bin/hadoop jar hadoop-streaming-2.4.0.jar -input /in -output /out2 -mapper /bin/cat

-reducer /usr/bin/wc

3.Other map reduce demo here:

Python:

http://www.michael-noll.com/tutorials/writing-an-hadoop-mapreduce-program-in-python/

C++:

http://blog.sina.com.cn/s/blog_62a9902f01018rzj.html

FIR hardware:

1.fir.h

#ifndef _FIR_H_
#define _FIR_H_
#include "ap_cint.h"
#define N	58
#define SAMPLES N+10 // just few more samples then number of taps
typedef short	coef_t;
typedef short	data_t;
typedef int38	acc_t;
#endif

2.fir_coef.dat

-378,
-73,
27,
170,
298,
352,
302,
168,
14,
-80,
-64,
53,
186,
216,
40,
-356,
-867,
-1283,
-1366,
-954,
-51,
1132,
2227,
2829,
2647,
1633,
25,
-1712,
-3042,
29229,
-3042,
-1712,
25,
1633,
2647,
2829,
2227,
1132,
-51,
-954,
-1366,
-1283,
-867,
-356,
40,
216,
186,
53,
-64,
-80,
14,
168,
302,
352,
298,
170,
27,
-73,
-378

3.fir.c

#include "fir.h"

void fir (
  data_t *y,
  data_t x
  ) {
  const coef_t c[N+1]={
 #include "fir_coef.dat"
    };
  

  static data_t shift_reg[N];
  acc_t acc;
  int i;
  
  acc=(acc_t)shift_reg[N-1]*(acc_t)c[N];
  loop: for (i=N-1;i!=0;i--) {
    acc+=(acc_t)shift_reg[i-1]*(acc_t)c[i];
    shift_reg[i]=shift_reg[i-1];
  }
  acc+=(acc_t)x*(acc_t)c[0];
  shift_reg[0]=x;
  *y = acc>>15;
}

4.fir_test.c

#include <stdio.h>
#include <math.h>
#include "fir.h"
void fir (
  data_t *y,
  data_t x
  );

int main () {
  FILE   *fp;

  data_t signal, output;

  fp=fopen("fir_impulse.dat","w");
  int i;
  for (i=0;i<SAMPLES;i++) {
	  if(i==0)
		  signal = 0x8000;
	  else
		  signal = 0;
	  fir(&output,signal);
   	  printf("%i %d %d\n",i,(int)signal,(int)output);
//   	  fprintf(fp,"%i %d %d\n",i,signal,output);
  }
  fclose(fp);
  return 0;
}

Here is the project in Vivado HLS:

Here is the Directive View:

In vivado Project Settings:

bd:

done

driver:

FIR on Hadoop using hadoop-streaming,布布扣,bubuko.com

FIR on Hadoop using hadoop-streaming

标签：style   blog   http   os   io   strong   for   2014   

原文地址：http://www.cnblogs.com/shenerguang/p/3891906.html