adding timing

submit_job_hadoop.py

@@ -5,6 +5,7 @@ This is an example job submit script for hadoop streaming
 import argparse
 import sys
 import os
+import time
 import subprocess
 sys.path.append('./src/')
 import rabit_master as master
@@ -25,5 +26,7 @@ def hadoop_streaming(nslaves, slave_args):
   cmd = '%s jar %s -input %s -output %s -mapper \"%s stdin %d %d stdout %s\" -reducer \"/bin/cat\" -file %s' % (args.hadoop_binary, args.hadoop_streaming_jar, args.input, args.output, args.mapper, args.nclusters, args.iterations, ' '.join(slave_args), args.mapper)
   print cmd
   subprocess.check_call(cmd, shell = True)
-  
+
+start = time.time()
 master.submit(args.nslaves, [], fun_submit= hadoop_streaming)
+print 'All run took %s' % (time.time() - start)

toolkit/kmeans.cpp

@@ -3,6 +3,7 @@
 #include <rabit.h>
 #include <utils.h>
 #include "./toolkit_util.h"
+#include <time.h>
 
 using namespace rabit;
 
@@ -85,6 +86,8 @@ int main(int argc, char *argv[]) {
     printf("Usage: <data_dir> num_cluster max_iter <out_model>\n");
     return 0;
   }
+  clock_t tStart = clock();
+
   srand(0);
   // load the data 
   SparseMat data;
@@ -140,6 +143,7 @@ int main(int argc, char *argv[]) {
   if (rabit::GetRank() == 0) {
     model.centroids.Print(argv[4]);
   }
+  utils::LogPrintf("[%d] Time taken: %f seconds\n", rabit::GetRank(), static_cast<float>(clock() - tStart) / CLOCKS_PER_SEC);
   rabit::Finalize();
   return 0;
 }