Merge commit 'd87691ec603db325d5b1c5db1186295a748df7cc' as 'subtree/rabit'

subtree/rabit/test/.gitignore

@@ -0,0 +1,4 @@
+*.mpi
+test_*
+*_test
+*_recover

subtree/rabit/test/Makefile

@@ -0,0 +1,41 @@
+export CC  = gcc
+export CXX = g++
+export MPICXX = mpicxx
+export LDFLAGS= -pthread -lm  -lrt -L../lib
+export CFLAGS = -Wall -O3 -msse2  -Wno-unknown-pragmas -fPIC -I../include  -std=c++11
+
+# specify tensor path
+BIN = speed_test model_recover local_recover lazy_recover
+OBJ = $(RABIT_OBJ) speed_test.o model_recover.o local_recover.o lazy_recover.o
+MPIBIN = speed_test.mpi
+.PHONY: clean all lib mpi
+
+all: $(BIN) $(MPIBIN)
+lib:
+	cd ..;make;cd -
+mpi:
+	cd ..;make mpi;cd -
+# programs 
+speed_test.o: speed_test.cc ../include/*.h lib mpi
+model_recover.o: model_recover.cc ../include/*.h lib
+local_recover.o: local_recover.cc ../include/*.h lib
+lazy_recover.o: lazy_recover.cc ../include/*.h lib
+
+# we can link against MPI version to get use MPI
+speed_test: speed_test.o  $(RABIT_OBJ)
+speed_test.mpi: speed_test.o $(MPIOBJ)
+model_recover: model_recover.o  $(RABIT_OBJ)
+local_recover: local_recover.o  $(RABIT_OBJ)
+lazy_recover: lazy_recover.o  $(RABIT_OBJ)
+
+$(BIN) : 
+	$(CXX) $(CFLAGS) -o $@ $(filter %.cpp %.o %.c %.cc, $^) $(LDFLAGS) -lrabit_mock
+
+$(OBJ) : 
+	$(CXX) -c $(CFLAGS) -o $@ $(firstword $(filter %.cpp %.c %.cc, $^) )
+
+$(MPIBIN) : 
+	$(MPICXX) $(CFLAGS) -o $@ $(filter %.cpp %.o %.c %.cc, $^) $(LDFLAGS) -lrabit_mpi
+
+clean:
+	$(RM) $(OBJ) $(BIN) $(MPIBIN) $(MPIOBJ) *~ ../src/*~

subtree/rabit/test/README.md

@@ -0,0 +1,18 @@
+Testcases of Rabit
+====
+This folder contains internal testcases to test correctness and efficiency of rabit API
+
+The example running scripts for testcases are given by test.mk
+* type ```make -f test.mk testcasename``` to run certain testcase
+
+
+Helper Scripts
+====
+* test.mk contains Makefile documentation of all testcases
+* keepalive.sh helper bash to restart a program when it dies abnormally
+
+List of Programs
+====
+* speed_test: test the running speed of rabit API
+* test_local_recover: test recovery of local state when error happens
+* test_model_recover: test recovery of global state when error happens

subtree/rabit/test/lazy_recover.cc

@@ -0,0 +1,126 @@
+// this is a test case to test whether rabit can recover model when 
+// facing an exception
+#include <rabit.h>
+#include <rabit/utils.h>
+#include <cstdio>
+#include <cstdlib>
+#include <cmath>
+using namespace rabit;
+
+// dummy model
+class Model : public rabit::ISerializable {
+ public:
+  // iterations
+  std::vector<float> data;
+  // load from stream
+  virtual void Load(rabit::IStream &fi) {
+    fi.Read(&data);
+  }
+  /*! \brief save the model to the stream */
+  virtual void Save(rabit::IStream &fo) const {
+    fo.Write(data);
+  }
+  virtual void InitModel(size_t n) {
+    data.clear();
+    data.resize(n, 1.0f);
+  }
+};
+
+inline void TestMax(Model *model, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = iter + 111;
+
+  std::vector<float> ndata(model->data.size());
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z  + model->data[i];
+  }
+  rabit::Allreduce<op::Max>(&ndata[0], ndata.size());  
+
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rmax = (i * 1) % z + model->data[i];
+    for (int r = 0; r < nproc; ++r) {
+      rmax = std::max(rmax, (float)((i * (r+1)) % z) + model->data[i]);
+    }
+    utils::Check(rmax == ndata[i], "[%d] TestMax check failurem i=%lu, rmax=%f, ndata=%f", rank, i, rmax, ndata[i]);
+  }
+}
+
+inline void TestSum(Model *model, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = 131 + iter;
+
+  std::vector<float> ndata(model->data.size());
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z + model->data[i];
+  }
+  Allreduce<op::Sum>(&ndata[0], ndata.size());
+
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rsum = model->data[i] * nproc;
+    for (int r = 0; r < nproc; ++r) {
+      rsum += (float)((i * (r+1)) % z);
+    }
+    utils::Check(fabsf(rsum - ndata[i]) < 1e-5 ,
+                 "[%d] TestSum check failure, local=%g, allreduce=%g", rank, rsum, ndata[i]);
+  }
+  model->data = ndata;
+}
+
+inline void TestBcast(size_t n, int root, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  std::string s; s.resize(n);      
+  for (size_t i = 0; i < n; ++i) {
+    s[i] = char(i % 126 + 1);
+  }
+  std::string res;
+  if (root == rank) {
+    res = s;
+    rabit::Broadcast(&res, root);
+  } else {
+    rabit::Broadcast(&res, root);
+  }
+  utils::Check(res == s, "[%d] TestBcast fail", rank);
+}
+
+int main(int argc, char *argv[]) {
+  if (argc < 3) {
+    printf("Usage: <ndata> <config>\n");
+    return 0;
+  }
+  int n = atoi(argv[1]);
+  rabit::Init(argc, argv);
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  std::string name = rabit::GetProcessorName();
+  Model model;  
+  srand(0);
+  int ntrial = 0;
+  for (int i = 1; i < argc; ++i) {
+    int n;
+    if (sscanf(argv[i], "rabit_num_trial=%d", &n) == 1) ntrial = n; 
+  } 
+  int iter = rabit::LoadCheckPoint(&model);
+  if (iter == 0) {
+    model.InitModel(n);
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  } else {
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  }
+  for (int r = iter; r < 3; ++r) { 
+    TestMax(&model, ntrial, r);
+    printf("[%d] !!!TestMax pass, iter=%d\n",  rank, r);  
+    int step = std::max(nproc / 3, 1);
+    for (int i = 0; i < nproc; i += step) {
+      TestBcast(n, i, ntrial, r);
+    }
+    printf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
+    TestSum(&model, ntrial, r);
+    printf("[%d] !!!TestSum pass, iter=%d\n", rank, r);
+    rabit::LazyCheckPoint(&model);
+    printf("[%d] !!!CheckPont pass, iter=%d\n", rank, r);
+  }
+  rabit::Finalize();
+  return 0;
+}

subtree/rabit/test/local_recover.cc

@@ -0,0 +1,138 @@
+// this is a test case to test whether rabit can recover model when 
+// facing an exception
+#include <rabit.h>
+#include <rabit/utils.h>
+#include <cstdio>
+#include <cstdlib>
+#include <cmath>
+
+using namespace rabit;
+
+// dummy model
+class Model : public rabit::ISerializable {
+ public:
+  // iterations
+  std::vector<float> data;
+  // load from stream
+  virtual void Load(rabit::IStream &fi) {
+    fi.Read(&data);
+  }
+  /*! \brief save the model to the stream */
+  virtual void Save(rabit::IStream &fo) const {
+    fo.Write(data);
+  }
+  virtual void InitModel(size_t n, float v) {
+    data.clear();
+    data.resize(n, v);
+  }
+};
+
+inline void TestMax(Model *model, Model *local, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = iter + 111;  
+  std::vector<float> ndata(model->data.size());
+  rabit::Allreduce<op::Max>(&ndata[0], ndata.size(), 
+                            [&]() {
+                              // use lambda expression to prepare the data
+                              for (size_t i = 0; i < ndata.size(); ++i) {
+                                ndata[i] = (i * (rank+1)) % z  + local->data[i];
+                              }
+                            });  
+
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rmax = (i * 1) % z + model->data[i];
+    for (int r = 0; r < nproc; ++r) {
+      rmax = std::max(rmax, (float)((i * (r+1)) % z) + model->data[i] + r);
+    }
+    utils::Check(rmax == ndata[i], "[%d] TestMax check failure", rank);
+  }
+  model->data = ndata;
+  local->data = ndata;
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    local->data[i] = ndata[i] + rank;
+  }
+}
+
+inline void TestSum(Model *model, Model *local, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = 131 + iter;
+
+  std::vector<float> ndata(model->data.size());
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z + local->data[i];
+  }
+  Allreduce<op::Sum>(&ndata[0], ndata.size());
+  
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rsum = 0.0f;
+    for (int r = 0; r < nproc; ++r) {
+      rsum += (float)((i * (r+1)) % z) + model->data[i] + r;
+    }
+    utils::Check(fabsf(rsum - ndata[i]) < 1e-5 ,
+                 "[%d] TestSum check failure, local=%g, allreduce=%g", rank, rsum, ndata[i]);
+  }
+  model->data = ndata;
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    local->data[i] = ndata[i] + rank;
+  }
+}
+
+inline void TestBcast(size_t n, int root, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  std::string s; s.resize(n);      
+  for (size_t i = 0; i < n; ++i) {
+    s[i] = char(i % 126 + 1);
+  }
+  std::string res;
+  if (root == rank) {
+    res = s;
+    rabit::Broadcast(&res, root);
+  } else {
+    rabit::Broadcast(&res, root);
+  }
+  utils::Check(res == s, "[%d] TestBcast fail", rank);
+}
+
+int main(int argc, char *argv[]) {
+  if (argc < 3) {
+    printf("Usage: <ndata>\n");
+    return 0;
+  }
+  int n = atoi(argv[1]);
+  rabit::Init(argc, argv);
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  std::string name = rabit::GetProcessorName();
+  Model model, local;  
+  srand(0);
+  int ntrial = 0;
+  for (int i = 1; i < argc; ++i) {
+    int n;
+    if (sscanf(argv[i], "repeat=%d", &n) == 1) ntrial = n; 
+  } 
+  int iter = rabit::LoadCheckPoint(&model, &local);
+  if (iter == 0) {
+    model.InitModel(n, 1.0f);
+    local.InitModel(n, 1.0f + rank);
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  } else {
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  }
+  for (int r = iter; r < 3; ++r) { 
+    TestMax(&model, &local, ntrial, r);
+    printf("[%d] !!!TestMax pass, iter=%d\n",  rank, r);  
+    int step = std::max(nproc / 3, 1);
+    for (int i = 0; i < nproc; i += step) {
+      TestBcast(n, i, ntrial, r);
+    }
+    printf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
+    TestSum(&model, &local, ntrial, r);
+    printf("[%d] !!!TestSum pass, iter=%d\n", rank, r);
+    rabit::CheckPoint(&model, &local);
+    printf("[%d] !!!CheckPont pass, iter=%d\n", rank, r);
+  }
+  rabit::Finalize();
+  return 0;
+}

subtree/rabit/test/local_recover.py

@@ -0,0 +1,25 @@
+#!/usr/bin/python
+import rabit
+import numpy as np
+
+rabit.init(lib='mock')
+rank = rabit.get_rank()
+n = 10
+nround = 3
+data = np.ones(n) * rank
+
+version, model, local = rabit.load_checkpoint(True)
+if version == 0:
+    model = np.zeros(n)
+    local = np.ones(n)
+else:
+    print '[%d] restart from version %d' % (rank, version)
+
+for i in xrange(version, nround):    
+    res = rabit.allreduce(data + model+local, rabit.SUM)
+    print '[%d] iter=%d: %s' % (rank, i, str(res))
+    model = res
+    local[:] = i
+    rabit.checkpoint(model, local)
+
+rabit.finalize()

subtree/rabit/test/model_recover.cc

@@ -0,0 +1,127 @@
+// this is a test case to test whether rabit can recover model when 
+// facing an exception
+#include <rabit.h>
+#include <rabit/utils.h>
+#include <cstdio>
+#include <cstdlib>
+#include <cmath>
+using namespace rabit;
+
+// dummy model
+class Model : public rabit::ISerializable {
+ public:
+  // iterations
+  std::vector<float> data;
+  // load from stream
+  virtual void Load(rabit::IStream &fi) {
+    fi.Read(&data);
+  }
+  /*! \brief save the model to the stream */
+  virtual void Save(rabit::IStream &fo) const {
+    fo.Write(data);
+  }
+  virtual void InitModel(size_t n) {
+    data.clear();
+    data.resize(n, 1.0f);
+  }
+};
+
+inline void TestMax(Model *model, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = iter + 111;
+
+  std::vector<float> ndata(model->data.size());
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z  + model->data[i];
+  }
+  rabit::Allreduce<op::Max>(&ndata[0], ndata.size());  
+
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rmax = (i * 1) % z + model->data[i];
+    for (int r = 0; r < nproc; ++r) {
+      rmax = std::max(rmax, (float)((i * (r+1)) % z) + model->data[i]);
+    }
+    utils::Check(rmax == ndata[i], "[%d] TestMax check failurem i=%lu, rmax=%f, ndata=%f", rank, i, rmax, ndata[i]);
+  }
+  model->data = ndata;
+}
+
+inline void TestSum(Model *model, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  const int z = 131 + iter;
+
+  std::vector<float> ndata(model->data.size());
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z + model->data[i];
+  }
+  Allreduce<op::Sum>(&ndata[0], ndata.size());
+
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    float rsum = model->data[i] * nproc;
+    for (int r = 0; r < nproc; ++r) {
+      rsum += (float)((i * (r+1)) % z);
+    }
+    utils::Check(fabsf(rsum - ndata[i]) < 1e-5 ,
+                 "[%d] TestSum check failure, local=%g, allreduce=%g", rank, rsum, ndata[i]);
+  }
+  model->data = ndata;
+}
+
+inline void TestBcast(size_t n, int root, int ntrial, int iter) {
+  int rank = rabit::GetRank();
+  std::string s; s.resize(n);      
+  for (size_t i = 0; i < n; ++i) {
+    s[i] = char(i % 126 + 1);
+  }
+  std::string res;
+  if (root == rank) {
+    res = s;
+    rabit::Broadcast(&res, root);
+  } else {
+    rabit::Broadcast(&res, root);
+  }
+  utils::Check(res == s, "[%d] TestBcast fail", rank);
+}
+
+int main(int argc, char *argv[]) {
+  if (argc < 3) {
+    printf("Usage: <ndata> <config>\n");
+    return 0;
+  }
+  int n = atoi(argv[1]);
+  rabit::Init(argc, argv);
+  int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  std::string name = rabit::GetProcessorName();
+  Model model;  
+  srand(0);
+  int ntrial = 0;
+  for (int i = 1; i < argc; ++i) {
+    int n;
+    if (sscanf(argv[i], "rabit_num_trial=%d", &n) == 1) ntrial = n; 
+  } 
+  int iter = rabit::LoadCheckPoint(&model);
+  if (iter == 0) {
+    model.InitModel(n);
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  } else {
+    printf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
+  }
+  for (int r = iter; r < 3; ++r) { 
+    TestMax(&model, ntrial, r);
+    printf("[%d] !!!TestMax pass, iter=%d\n",  rank, r);  
+    int step = std::max(nproc / 3, 1);
+    for (int i = 0; i < nproc; i += step) {
+      TestBcast(n, i, ntrial, r);
+    }
+    printf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
+    TestSum(&model, ntrial, r);
+    printf("[%d] !!!TestSum pass, iter=%d\n", rank, r);
+    rabit::CheckPoint(&model);
+    printf("[%d] !!!CheckPont pass, iter=%d\n", rank, r);
+  }
+  rabit::Finalize();
+  return 0;
+}

subtree/rabit/test/speed_runner.py

@@ -0,0 +1,34 @@
+import os
+import argparse
+import sys
+
+def main():
+  parser = argparse.ArgumentParser(description='TODO')
+  parser.add_argument('-ho', '--host_dir', required=True)
+  parser.add_argument('-s', '--submit_script', required=True)
+  parser.add_argument('-rex', '--rabit_exec', required=True)
+  parser.add_argument('-mpi', '--mpi_exec', required=True)
+  args = parser.parse_args()
+
+  ndata = [10**4, 10**5, 10**6, 10**7]
+  nrepeat = [10**4, 10**3, 10**2, 10]
+
+  machines = [2,4,8,16,31]
+
+  executables = [args.rabit_exec, args.mpi_exec]
+
+  for executable in executables:
+    sys.stderr.write('Executable %s' % executable)
+    sys.stderr.flush()
+    for i, data in enumerate(ndata):
+      for machine in machines:
+        host_file = os.path.join(args.host_dir, 'hosts%d' % machine)
+        cmd = 'python %s %d %s %s %d %d' % (args.submit_script, machine, host_file, executable, data, nrepeat[i])
+        sys.stderr.write('data=%d, repeat=%d, machine=%d\n' % (data, nrepeat[i], machine))
+        sys.stderr.flush()
+        os.system(cmd)
+    sys.stderr.write('\n')
+    sys.stderr.flush()
+
+if __name__ == "__main__":
+  main()

subtree/rabit/test/speed_test.cc

@@ -0,0 +1,100 @@
+// This program is used to test the speed of rabit API
+#include <rabit.h>
+#include <rabit/utils.h>
+#include <rabit/timer.h>
+#include <cstdio>
+#include <cstdlib>
+#include <cmath>
+#include <time.h>
+
+using namespace rabit;
+
+double max_tdiff, sum_tdiff, bcast_tdiff, tot_tdiff;
+
+inline void TestMax(size_t n) {
+  int rank = rabit::GetRank();
+  std::vector<float> ndata(n);
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % 111;
+  }
+  double tstart = utils::GetTime();
+  rabit::Allreduce<op::Max>(&ndata[0], ndata.size());
+  max_tdiff += utils::GetTime() - tstart;
+}
+
+inline void TestSum(size_t n) {
+  int rank = rabit::GetRank();
+  const int z = 131;
+  std::vector<float> ndata(n);
+  for (size_t i = 0; i < ndata.size(); ++i) {
+    ndata[i] = (i * (rank+1)) % z;
+  }
+  double tstart = utils::GetTime();
+  rabit::Allreduce<op::Sum>(&ndata[0], ndata.size());  
+  sum_tdiff += utils::GetTime() - tstart;
+}
+
+inline void TestBcast(size_t n, int root) {
+  int rank = rabit::GetRank();
+  std::string s; s.resize(n);
+  for (size_t i = 0; i < n; ++i) {
+    s[i] = char(i % 126 + 1);
+  }
+  std::string res;
+  res.resize(n);
+  if (root == rank) {
+    res = s;
+  }
+  double tstart = utils::GetTime();  
+  rabit::Broadcast(&res[0], res.length(), root);
+  bcast_tdiff += utils::GetTime() - tstart;  
+}
+
+inline void PrintStats(const char *name, double tdiff, int n, int nrep, size_t size) {
+  int nproc = rabit::GetWorldSize();
+  double tsum = tdiff;
+  rabit::Allreduce<op::Sum>(&tsum, 1);
+  double tavg = tsum / nproc;
+  double tsqr = tdiff - tavg;
+  tsqr *= tsqr;
+  rabit::Allreduce<op::Sum>(&tsqr, 1);
+  double tstd = sqrt(tsqr / nproc);
+  if (rabit::GetRank() == 0) {
+    rabit::TrackerPrintf("%s: mean=%g, std=%g sec\n", name, tavg, tstd);
+    double ndata = n;
+    ndata *= nrep * size;
+    if (n != 0) {
+      rabit::TrackerPrintf("%s-speed: %g MB/sec\n", name, (ndata / tavg) / 1024 / 1024 );
+    }
+  }
+}
+
+int main(int argc, char *argv[]) {
+  if (argc < 3) {
+    printf("Usage: <ndata> <nrepeat>\n");
+    return 0;
+  }
+  srand(0);
+  int n = atoi(argv[1]);
+  int nrep = atoi(argv[2]);
+  utils::Check(nrep >= 1, "need to at least repeat running once");
+  rabit::Init(argc, argv);
+  //int rank = rabit::GetRank();
+  int nproc = rabit::GetWorldSize();
+  std::string name = rabit::GetProcessorName();
+  max_tdiff = sum_tdiff = bcast_tdiff = 0;
+  double tstart = utils::GetTime();
+  for (int i = 0; i < nrep; ++i) {
+    TestMax(n);
+    TestSum(n);
+    TestBcast(n, rand() % nproc);
+  }
+  tot_tdiff = utils::GetTime() - tstart;
+  // use allreduce to get the sum and std of time
+  PrintStats("max_tdiff", max_tdiff, n, nrep, sizeof(float));
+  PrintStats("sum_tdiff", sum_tdiff, n, nrep, sizeof(float));
+  PrintStats("bcast_tdiff", bcast_tdiff, n, nrep, sizeof(char));
+  PrintStats("tot_tdiff", tot_tdiff, 0, nrep, sizeof(float));
+  rabit::Finalize();
+  return 0;
+}

subtree/rabit/test/test.mk

@@ -0,0 +1,26 @@
+# this is a makefile used to show testcases of rabit
+.PHONY: all
+
+all:
+
+# this experiment test recovery with actually process exit, use keepalive to keep program alive
+model_recover_10_10k:
+	../tracker/rabit_demo.py -n 10 model_recover 10000 mock=0,0,1,0 mock=1,1,1,0
+
+model_recover_10_10k_die_same:
+	../tracker/rabit_demo.py -n 10 model_recover 10000 mock=0,0,1,0 mock=1,1,1,0 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0
+
+model_recover_10_10k_die_hard:
+	../tracker/rabit_demo.py -n 10 model_recover 10000 mock=0,0,1,0 mock=1,1,1,0 mock=1,1,1,1 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0 mock=8,1,2,0 mock=4,1,3,0
+
+local_recover_10_10k:
+	../tracker/rabit_demo.py -n 10 local_recover 10000 mock=0,0,1,0 mock=1,1,1,0 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0 mock=1,1,1,1
+
+pylocal_recover_10_10k:
+	../tracker/rabit_demo.py -n 10 ./local_recover.py 10000 mock=0,0,1,0 mock=1,1,1,0 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0 mock=1,1,1,1
+
+lazy_recover_10_10k_die_hard:
+	../tracker/rabit_demo.py -n 10 lazy_recover 10000 mock=0,0,1,0 mock=1,1,1,0 mock=1,1,1,1 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0 mock=8,1,2,0 mock=4,1,3,0
+
+lazy_recover_10_10k_die_same:
+	../tracker/rabit_demo.py -n 10 lazy_recover 10000 mock=0,0,1,0 mock=1,1,1,0 mock=0,1,1,0 mock=4,1,1,0 mock=9,1,1,0