Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

cleanup testcases

Browse Source
This commit is contained in:
tqchen
2014-12-18 23:50:59 -08:00
parent aa2cb38543
commit 58331067f8
18 changed files with 152 additions and 306 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
test/Makefile
View File

@@ -5,33 +5,30 @@ export LDFLAGS= -pthread -lm -lrt
export CFLAGS = -Wall -O3 -msse2 -Wno-unknown-pragmas -fPIC -I../src
# specify tensor path
BIN = test_allreduce test_recover test_model_recover speed_test test_local_recover
BIN = speed_test test_model_recover test_local_recover
# objectives that makes up rabit library
RABIT_OBJ = allreduce_base.o allreduce_robust.o engine.o
MPIOBJ = engine_mpi.o
OBJ = $(RABIT_OBJ) test_allreduce.o test_recover.o test_model_recover.o speed_test.o test_local_recover.o
MPIBIN = test_allreduce.mpi speed_test.mpi
OBJ = $(RABIT_OBJ) speed_test.o test_model_recover.o test_local_recover.o
MPIBIN = speed_test.mpi
.PHONY: clean all
all: $(BIN) $(MPIBIN)
# the rabit library
allreduce_base.o: ../src/allreduce_base.cc ../src/*.h
engine.o: ../src/engine.cc ../src/*.h
allreduce_robust.o: ../src/allreduce_robust.cc ../src/*.h
engine_mpi.o: ../src/engine_mpi.cc
test_allreduce.o: test_allreduce.cpp ../src/*.h
allreduce_robust.o: ../src/allreduce_robust.cc ../src/*.h
# programs
speed_test.o: speed_test.cpp ../src/*.h
test_recover.o: test_recover.cpp ../src/*.h
test_model_recover.o: test_model_recover.cpp ../src/*.h
test_local_recover.o: test_local_recover.cpp ../src/*.h
# we can link against MPI version to get use MPI
test_allreduce: test_allreduce.o $(RABIT_OBJ)
test_allreduce.mpi: test_allreduce.o $(MPIOBJ)
speed_test: speed_test.o $(RABIT_OBJ)
speed_test.mpi: speed_test.o $(MPIOBJ)
test_recover: test_recover.o $(RABIT_OBJ)
test_model_recover: test_model_recover.o $(RABIT_OBJ)
test_local_recover: test_local_recover.o $(RABIT_OBJ)
@@ -48,4 +45,4 @@ $(MPIOBJ) :
$(MPICXX) -c $(CFLAGS) -o $@ $(firstword $(filter %.cpp %.c %.cc, $^) )
clean:
$(RM) $(OBJ) $(BIN) $(MPIBIN) *~ ../src/*~
$(RM) $(OBJ) $(BIN) $(MPIBIN) $(MPIOBJ) *~ ../src/*~

18
test/README.md Normal file
View File

@@ -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

196
test/config.h Normal file
View File

@@ -0,0 +1,196 @@
#ifndef RABIT_UTILS_CONFIG_H_
#define RABIT_UTILS_CONFIG_H_
/*!
* \file config.h
* \brief helper class to load in configures from file
* \author Tianqi Chen
*/
#include <cstdio>
#include <cstring>
#include <string>
#include <istream>
#include <fstream>
#include "./utils.h"
namespace rabit {
namespace utils {
/*!
* \brief base implementation of config reader
*/
class ConfigReaderBase {
public:
/*!
* \brief get current name, called after Next returns true
* \return current parameter name
*/
inline const char *name(void) const {
return s_name;
}
/*!
* \brief get current value, called after Next returns true
* \return current parameter value
*/
inline const char *val(void) const {
return s_val;
}
/*!
* \brief move iterator to next position
* \return true if there is value in next position
*/
inline bool Next(void) {
while (!this->IsEnd()) {
GetNextToken(s_name);
if (s_name[0] == '=') return false;
if (GetNextToken( s_buf ) || s_buf[0] != '=') return false;
if (GetNextToken( s_val ) || s_val[0] == '=') return false;
return true;
}
return false;
}
// called before usage
inline void Init(void) {
ch_buf = this->GetChar();
}
protected:
/*!
* \brief to be implemented by subclass,
* get next token, return EOF if end of file
*/
virtual char GetChar(void) = 0;
/*! \brief to be implemented by child, check if end of stream */
virtual bool IsEnd(void) = 0;
private:
char ch_buf;
char s_name[100000], s_val[100000], s_buf[100000];
inline void SkipLine(void) {
do {
ch_buf = this->GetChar();
} while (ch_buf != EOF && ch_buf != '\n' && ch_buf != '\r');
}
inline void ParseStr(char tok[]) {
int i = 0;
while ((ch_buf = this->GetChar()) != EOF) {
switch (ch_buf) {
case '\\': tok[i++] = this->GetChar(); break;
case '\"': tok[i++] = '\0'; return;
case '\r':
case '\n': Error("ConfigReader: unterminated string");
default: tok[i++] = ch_buf;
}
}
Error("ConfigReader: unterminated string");
}
inline void ParseStrML(char tok[]) {
int i = 0;
while ((ch_buf = this->GetChar()) != EOF) {
switch (ch_buf) {
case '\\': tok[i++] = this->GetChar(); break;
case '\'': tok[i++] = '\0'; return;
default: tok[i++] = ch_buf;
}
}
Error("unterminated string");
}
// return newline
inline bool GetNextToken(char tok[]) {
int i = 0;
bool new_line = false;
while (ch_buf != EOF) {
switch (ch_buf) {
case '#' : SkipLine(); new_line = true; break;
case '\"':
if (i == 0) {
ParseStr(tok); ch_buf = this->GetChar(); return new_line;
} else {
Error("ConfigReader: token followed directly by string");
}
case '\'':
if (i == 0) {
ParseStrML( tok ); ch_buf = this->GetChar(); return new_line;
} else {
Error("ConfigReader: token followed directly by string");
}
case '=':
if (i == 0) {
ch_buf = this->GetChar();
tok[0] = '=';
tok[1] = '\0';
} else {
tok[i] = '\0';
}
return new_line;
case '\r':
case '\n':
if (i == 0) new_line = true;
case '\t':
case ' ' :
ch_buf = this->GetChar();
if (i > 0) {
tok[i] = '\0';
return new_line;
}
break;
default:
tok[i++] = ch_buf;
ch_buf = this->GetChar();
break;
}
}
return true;
}
};
/*!
* \brief an iterator use stream base, allows use all types of istream
*/
class ConfigStreamReader: public ConfigReaderBase {
public:
/*!
* \brief constructor
* \param istream input stream
*/
explicit ConfigStreamReader(std::istream &fin) : fin(fin) {}
protected:
virtual char GetChar(void) {
return fin.get();
}
/*! \brief to be implemented by child, check if end of stream */
virtual bool IsEnd(void) {
return fin.eof();
}
private:
std::istream &fin;
};
/*!
* \brief an iterator that iterates over a configure file and gets the configures
*/
class ConfigIterator: public ConfigStreamReader {
public:
/*!
* \brief constructor
* \param fname name of configure file
*/
explicit ConfigIterator(const char *fname) : ConfigStreamReader(fi) {
fi.open(fname);
if (fi.fail()) {
utils::Error("cannot open file %s", fname);
}
ConfigReaderBase::Init();
}
/*! \brief destructor */
~ConfigIterator(void) {
fi.close();
}
private:
std::ifstream fi;
};
} // namespace utils
} // namespace rabit
#endif // RABIT_UTILS_CONFIG_H_

6
test/keepalive.sh
View File

@@ -6,9 +6,9 @@ then
exit -1
fi
nrep=0
echo ./$@ task_id=$OMPI_COMM_WORLD_RANK
until ./$@ task_id=$OMPI_COMM_WORLD_RANK repeat=$nrep; do
echo ./$@ rabit_task_id=$OMPI_COMM_WORLD_RANK
until ./$@ rabit_task_id=$OMPI_COMM_WORLD_RANK repeat=$nrep; do
sleep 1
nrep=$((nrep+1))
echo ./$@ job_id=$OMPI_COMM_WORLD_RANK repeat=$nrep
echo ./$@ rabit_task_id=$OMPI_COMM_WORLD_RANK repeat=$nrep
done

121
test/mock.h Normal file
View File

@@ -0,0 +1,121 @@
#ifndef RABIT_MOCK_H
#define RABIT_MOCK_H
/*!
* \file mock.h
* \brief This file defines a mock object to test the system
* \author Ignacio Cano
*/
#include "./rabit.h"
#include "./config.h"
#include <map>
#include <sstream>
#include <fstream>
struct MockException {
};
namespace rabit {
/*! \brief namespace of mock */
namespace test {
class Mock {
public:
explicit Mock(const int& rank, char *config, char* round_dir) : rank(rank) {
Init(config, round_dir);
}
template<typename OP>
inline void Allreduce(float *sendrecvbuf, size_t count) {
utils::Assert(verify(allReduce), "[%d] error when calling allReduce", rank);
rabit::Allreduce<OP>(sendrecvbuf, count);
}
inline int LoadCheckPoint(utils::ISerializable *global_model,
utils::ISerializable *local_model) {
utils::Assert(verify(loadCheckpoint), "[%d] error when loading checkpoint", rank);
return rabit::LoadCheckPoint(global_model, local_model);
}
inline void CheckPoint(const utils::ISerializable *global_model,
const utils::ISerializable *local_model) {
utils::Assert(verify(checkpoint), "[%d] error when checkpointing", rank);
rabit::CheckPoint(global_model, local_model);
}
inline void Broadcast(std::string *sendrecv_data, int root) {
utils::Assert(verify(broadcast), "[%d] error when broadcasting", rank);
rabit::Broadcast(sendrecv_data, root);
}
private:
inline void Init(char* config, char* round_dir) {
std::stringstream ss;
ss << round_dir << "node" << rank << ".round";
const char* round_file = ss.str().c_str();
std::ifstream ifs(round_file);
int current_round = 1;
if (!ifs.good()) {
// file does not exists, it's the first time, so save the current round to 1
std::ofstream ofs(round_file);
ofs << current_round;
ofs.close();
} else {
// file does exists, read the previous round, increment by one, and save it back
ifs >> current_round;
current_round++;
ifs.close();
std::ofstream ofs(round_file);
ofs << current_round;
ofs.close();
}
printf("[%d] in round %d\n", rank, current_round);
utils::ConfigIterator itr(config);
while (itr.Next()) {
char round[4], node_rank[4];
sscanf(itr.name(), "%[^_]_%s", round, node_rank);
int i_node_rank = atoi(node_rank);
// if it's something for me
if (i_node_rank == rank) {
int i_round = atoi(round);
// in my current round
if (i_round == current_round) {
printf("[%d] round %d, value %s\n", rank, i_round, itr.val());
if (strcmp("allreduce", itr.val())) record(allReduce);
else if (strcmp("broadcast", itr.val())) record(broadcast);
else if (strcmp("loadcheckpoint", itr.val())) record(loadCheckpoint);
else if (strcmp("checkpoint", itr.val())) record(checkpoint);
}
}
}
}
inline void record(std::map<int,bool>& m) {
m[rank] = false;
}
inline bool verify(std::map<int,bool>& m) {
bool result = true;
if (m.find(rank) != m.end()) {
result = m[rank];
}
return result;
}
int rank;
std::map<int,bool> allReduce;
std::map<int,bool> broadcast;
std::map<int,bool> loadCheckpoint;
std::map<int,bool> checkpoint;
};
} // namespace test
} // namespace rabit
#endif // RABIT_MOCK_H

2
test/speed_runner.py
View File

@@ -31,4 +31,4 @@ def main():
sys.stderr.flush()
if __name__ == "__main__":
main()
main()

4
test/speed_test.cpp
View File

@@ -1,3 +1,4 @@
// This program is used to test the speed of rabit API
#include <rabit.h>
#include <utils.h>
#include <timer.h>
@@ -12,8 +13,6 @@ double max_tdiff, sum_tdiff, bcast_tdiff, tot_tdiff;
inline void TestMax(size_t n) {
int rank = rabit::GetRank();
//int nproc = rabit::GetWorldSize();
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % 111;
@@ -25,7 +24,6 @@ inline void TestMax(size_t n) {
inline void TestSum(size_t n) {
int rank = rabit::GetRank();
//int nproc = rabit::GetWorldSize();
const int z = 131;
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {

20
test/test.mk Normal file
View File

@@ -0,0 +1,20 @@
ifndef $(nslave)
nslave=2
endif
ifndef $(ndata)
ndata=10
endif
# this is a makefile used to show testcases of rabit
.PHONY: model_recover local_recover speed
local_recover:
../submit_mpi.py $(nslave) local test_local_recover $(ndata) rabit_local_replica=1
local_recover_10_10k:
../submit_mpi.py 10 local test_local_recover 10000 rabit_local_replica=1
# this experiment test recovery with actually process exit, use keepalive to keep program alive
model_recover_10_10k:
../submit_mpi.py 10 local keepalive.sh test_model_recover 10000

90
test/test_allreduce.cpp
View File

@@ -1,90 +0,0 @@
#include <rabit.h>
#include <utils.h>
#include <cstdio>
#include <cstdlib>
#include <cmath>
#include <mock.h>
using namespace rabit;
inline void TestMax(test::Mock &mock, size_t n) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % 111;
}
mock.Allreduce<op::Max>(&ndata[0], ndata.size());
for (size_t i = 0; i < ndata.size(); ++i) {
float rmax = (i * 1) % 111;
for (int r = 0; r < nproc; ++r) {
rmax = std::max(rmax, (float)((i * (r+1)) % 111));
}
utils::Check(rmax == ndata[i], "[%d] TestMax check failure", rank);
}
}
inline void TestSum(test::Mock &mock, size_t n) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = 131;
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % z;
}
mock.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);
}
utils::Check(fabsf(rsum - ndata[i]) < 1e-5 ,
"[%d] TestSum check failure, local=%g, allreduce=%g", rank, rsum, ndata[i]);
}
}
inline void TestBcast(test::Mock &mock, 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;
if (root == rank) {
res = s;
mock.Broadcast(&res, root);
} else {
mock.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();
test::Mock mock(rank, argv[2], argv[3]);
utils::LogPrintf("[%d] start at %s\n", rank, name.c_str());
TestMax(mock, n);
utils::LogPrintf("[%d] !!!TestMax pass\n", rank);
TestSum(mock, n);
utils::LogPrintf("[%d] !!!TestSum pass\n", rank);
int step = std::max(nproc / 3, 1);
for (int i = 0; i < nproc; i += step) {
TestBcast(mock, n, i);
}
utils::LogPrintf("[%d] !!!TestBcast pass\n", rank);
rabit::Finalize();
printf("[%d] all check pass\n", rank);
return 0;
}

68
test/test_local_recover.cpp
View File

@@ -5,12 +5,28 @@
#include <cstdio>
#include <cstdlib>
#include <cmath>
#include <mock.h>
#include "./mock.h"
using namespace rabit;
struct MockException {
};
namespace rabit {
namespace test {
inline void CallBegin(const char *fun, int ntrial, int iter) {
int rank = rabit::GetRank();
if (!strcmp(fun, "Allreduce::Sum")) {
if (ntrial == iter && rank == 0) throw MockException();
}
if (!strcmp(fun, "Allreduce::Max")) {
if (ntrial == iter && rank == 3) throw MockException();
}
}
inline void CallEnd(const char *fun, int ntrial, int iter) {
int rank = rabit::GetRank();
if (!strcmp(fun, "Allreduce::Bcast")) {
if (ntrial == iter && rand() % 10 == rank) throw MockException();
}
}
}
}
// dummy model
class Model : public rabit::utils::ISerializable {
@@ -31,7 +47,7 @@ class Model : public rabit::utils::ISerializable {
}
};
inline void TestMax(test::Mock &mock, Model *model, Model *local, int ntrial, int iter) {
inline void TestMax(Model *model, Model *local, int ntrial, int iter) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = iter + 111;
@@ -39,11 +55,11 @@ inline void TestMax(test::Mock &mock, Model *model, Model *local, int ntrial, in
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];
}
mock.Allreduce<op::Max>(&ndata[0], ndata.size());
if (ntrial == iter && rank == 1) {
throw MockException();
}
}
test::CallBegin("Allreduce::Max", ntrial, iter);
rabit::Allreduce<op::Max>(&ndata[0], ndata.size());
test::CallEnd("Allreduce::Max", ntrial, iter);
for (size_t i = 0; i < ndata.size(); ++i) {
float rmax = (i * 1) % z + model->data[i];
for (int r = 0; r < nproc; ++r) {
@@ -58,7 +74,7 @@ inline void TestMax(test::Mock &mock, Model *model, Model *local, int ntrial, in
}
}
inline void TestSum(test::Mock &mock, Model *model, Model *local, int ntrial, int iter) {
inline void TestSum(Model *model, Model *local, int ntrial, int iter) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = 131 + iter;
@@ -67,11 +83,10 @@ inline void TestSum(test::Mock &mock, Model *model, Model *local, int ntrial, in
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % z + local->data[i];
}
if (ntrial == iter && rank == 0) {
throw MockException();
}
mock.Allreduce<op::Sum>(&ndata[0], ndata.size());
test::CallBegin("Allreduce::Sum", ntrial, iter);
Allreduce<op::Sum>(&ndata[0], ndata.size());
test::CallEnd("Allreduce::Sum", ntrial, iter);
for (size_t i = 0; i < ndata.size(); ++i) {
float rsum = 0.0f;
for (int r = 0; r < nproc; ++r) {
@@ -86,7 +101,7 @@ inline void TestSum(test::Mock &mock, Model *model, Model *local, int ntrial, in
}
}
inline void TestBcast(test::Mock &mock, size_t n, int root, int ntrial) {
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) {
@@ -95,16 +110,20 @@ inline void TestBcast(test::Mock &mock, size_t n, int root, int ntrial) {
std::string res;
if (root == rank) {
res = s;
mock.Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
rabit::Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
} else {
mock.Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
rabit::Broadcast(&res, root);
test::CallEnd("Broadcast", ntrial, iter);
}
utils::Check(res == s, "[%d] TestBcast fail", rank);
}
int main(int argc, char *argv[]) {
if (argc < 3) {
printf("Usage: <ndata> <config>\n");
printf("Usage: <ndata>\n");
return 0;
}
int n = atoi(argv[1]);
@@ -112,7 +131,6 @@ int main(int argc, char *argv[]) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
std::string name = rabit::GetProcessorName();
test::Mock mock(rank, argv[2], argv[3]);
Model model, local;
srand(0);
int ntrial = 0;
@@ -131,14 +149,14 @@ int main(int argc, char *argv[]) {
utils::LogPrintf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
}
for (int r = iter; r < 3; ++r) {
TestMax(mock, &model, &local, ntrial, r);
TestMax(&model, &local, ntrial, r);
utils::LogPrintf("[%d] !!!TestMax pass, iter=%d\n", rank, r);
int step = std::max(nproc / 3, 1);
for (int i = 0; i < nproc; i += step) {
//TestBcast(mock, n, i, ntrial);
TestBcast(n, i, ntrial, r);
}
//utils::LogPrintf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
TestSum(mock, &model, &local, ntrial, r);
utils::LogPrintf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
TestSum(&model, &local, ntrial, r);
utils::LogPrintf("[%d] !!!TestSum pass, iter=%d\n", rank, r);
rabit::CheckPoint(&model, &local);
utils::LogPrintf("[%d] !!!CheckPont pass, iter=%d\n", rank, r);

69
test/test_model_recover.cpp
View File

@@ -5,12 +5,28 @@
#include <cstdio>
#include <cstdlib>
#include <cmath>
#include <mock.h>
#include "./mock.h"
using namespace rabit;
struct MockException {
};
namespace rabit {
namespace test {
inline void CallBegin(const char *fun, int ntrial, int iter) {
int rank = rabit::GetRank();
if (!strcmp(fun, "Allreduce::Sum")) {
if (ntrial == iter && rank == 0) exit(-1);
}
if (!strcmp(fun, "Allreduce::Max")) {
if (ntrial == iter && rank == 3) exit(-1);
}
}
inline void CallEnd(const char *fun, int ntrial, int iter) {
int rank = rabit::GetRank();
if (!strcmp(fun, "Allreduce::Bcast")) {
if (ntrial == iter && rand() % 10 == rank) exit(-1);
}
}
}
}
// dummy model
class Model : public rabit::utils::ISerializable {
@@ -31,7 +47,7 @@ class Model : public rabit::utils::ISerializable {
}
};
inline void TestMax(test::Mock &mock, Model *model, int ntrial, int iter) {
inline void TestMax(Model *model, int ntrial, int iter) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = iter + 111;
@@ -40,10 +56,10 @@ inline void TestMax(test::Mock &mock, Model *model, int ntrial, int iter) {
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % z + model->data[i];
}
mock.Allreduce<op::Max>(&ndata[0], ndata.size());
if (ntrial == 0 && rank == 3) {
exit(-1);
}
test::CallBegin("Allreduce::Max", ntrial, iter);
rabit::Allreduce<op::Max>(&ndata[0], ndata.size());
test::CallEnd("Allreduce::Max", ntrial, iter);
for (size_t i = 0; i < ndata.size(); ++i) {
float rmax = (i * 1) % z + model->data[i];
for (int r = 0; r < nproc; ++r) {
@@ -54,7 +70,7 @@ inline void TestMax(test::Mock &mock, Model *model, int ntrial, int iter) {
model->data = ndata;
}
inline void TestSum(test::Mock &mock, Model *model, int ntrial, int iter) {
inline void TestSum(Model *model, int ntrial, int iter) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = 131 + iter;
@@ -63,11 +79,9 @@ inline void TestSum(test::Mock &mock, Model *model, int ntrial, int iter) {
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % z + model->data[i];
}
if (iter == 0 && ntrial==0 && rank == 0) {
throw MockException();
}
mock.Allreduce<op::Sum>(&ndata[0], ndata.size());
test::CallBegin("Allreduce::Sum", ntrial, iter);
Allreduce<op::Sum>(&ndata[0], ndata.size());
test::CallEnd("Allreduce::Sum", ntrial, iter);
for (size_t i = 0; i < ndata.size(); ++i) {
float rsum = model->data[i] * nproc;
@@ -80,7 +94,7 @@ inline void TestSum(test::Mock &mock, Model *model, int ntrial, int iter) {
model->data = ndata;
}
inline void TestBcast(test::Mock &mock, size_t n, int root, int ntrial) {
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) {
@@ -89,9 +103,13 @@ inline void TestBcast(test::Mock &mock, size_t n, int root, int ntrial) {
std::string res;
if (root == rank) {
res = s;
mock.Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
rabit::Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
} else {
mock.Broadcast(&res, root);
test::CallBegin("Broadcast", ntrial, iter);
rabit::Broadcast(&res, root);
test::CallEnd("Broadcast", ntrial, iter);
}
utils::Check(res == s, "[%d] TestBcast fail", rank);
}
@@ -106,7 +124,6 @@ int main(int argc, char *argv[]) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
std::string name = rabit::GetProcessorName();
test::Mock mock(rank, argv[2], argv[3]);
Model model;
srand(0);
int ntrial = 0;
@@ -124,14 +141,14 @@ int main(int argc, char *argv[]) {
utils::LogPrintf("[%d] reload-trail=%d, init iter=%d\n", rank, ntrial, iter);
}
for (int r = iter; r < 3; ++r) {
TestMax(mock, &model, ntrial, r);
TestMax(&model, ntrial, r);
utils::LogPrintf("[%d] !!!TestMax pass, iter=%d\n", rank, r);
//int step = std::max(nproc / 3, 1);
//for (int i = 0; i < nproc; i += step) {
//TestBcast(mock, n, i, ntrial);
//}
//utils::LogPrintf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
TestSum(mock, &model, ntrial, r);
int step = std::max(nproc / 3, 1);
for (int i = 0; i < nproc; i += step) {
TestBcast(n, i, ntrial, r);
}
utils::LogPrintf("[%d] !!!TestBcast pass, iter=%d\n", rank, r);
TestSum(&model, ntrial, r);
utils::LogPrintf("[%d] !!!TestSum pass, iter=%d\n", rank, r);
rabit::CheckPoint(&model);
utils::LogPrintf("[%d] !!!CheckPont pass, iter=%d\n", rank, r);

125
test/test_recover.cpp
View File

@@ -1,125 +0,0 @@
#include <rabit.h>
#include <utils.h>
#include <cstdio>
#include <cstdlib>
#include <cmath>
#include <mock.h>
using namespace rabit;
struct MockException {
};
inline void TestMax(test::Mock &mock, size_t n, int ntrial) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % 111;
}
mock.Allreduce<op::Max>(&ndata[0], ndata.size());
if (ntrial == 0 && rank == 15) throw MockException();
for (size_t i = 0; i < ndata.size(); ++i) {
float rmax = (i * 1) % 111;
for (int r = 0; r < nproc; ++r) {
rmax = std::max(rmax, (float)((i * (r+1)) % 111));
}
utils::Check(rmax == ndata[i], "[%d] TestMax check failure", rank);
}
}
inline void TestSum(test::Mock &mock, size_t n, int ntrial) {
int rank = rabit::GetRank();
int nproc = rabit::GetWorldSize();
const int z = 131;
std::vector<float> ndata(n);
for (size_t i = 0; i < ndata.size(); ++i) {
ndata[i] = (i * (rank+1)) % z;
}
mock.Allreduce<op::Sum>(&ndata[0], ndata.size());
if (ntrial == 0 && rank == 0) throw MockException();
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);
}
utils::Check(fabsf(rsum - ndata[i]) < 1e-5 ,
"[%d] TestSum check failure, local=%g, allreduce=%g", rank, rsum, ndata[i]);
}
}
inline void TestBcast(test::Mock &mock, size_t n, int root, int ntrial) {
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;
mock.Broadcast(&res, root);
} else {
mock.Broadcast(&res, root);
}
utils::Check(res == s, "[%d] TestBcast fail", rank);
}
// dummy model
class Model : public rabit::utils::ISerializable {
public:
// load from stream
virtual void Load(rabit::utils::IStream &fi) {
// do nothing
}
/*! \brief save the model to the stream */
virtual void Save(rabit::utils::IStream &fo) const {
// do nothing
}
virtual void InitModel(void) {
// do nothing
}
};
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();
test::Mock mock(rank, argv[2], argv[3]);
Model model;
srand(0);
int ntrial = 0;
while (true) {
try {
if (rabit::LoadCheckPoint(&model) == 0) {
model.InitModel();
}
utils::LogPrintf("[%d/%d] start at %s\n", rank, ntrial, name.c_str());
TestMax(mock, n, ntrial);
utils::LogPrintf("[%d/%d] !!!TestMax pass\n", rank, ntrial);
TestSum(mock, n, ntrial);
utils::LogPrintf("[%d/%d] !!!TestSum pass\n", rank, ntrial);
int step = std::max(nproc / 3, 1);
for (int i = 0; i < nproc; i += step) {
TestBcast(mock, n, i, ntrial);
}
utils::LogPrintf("[%d] !!!TestBcast pass\n", rank);
// reach here
break;
} catch (MockException &e) {
rabit::engine::GetEngine()->InitAfterException();
++ntrial;
}
}
rabit::Finalize();
printf("[%d] all check pass\n", rank);
return 0;
}

1
test/testcase0.conf
View File

@@ -1 +0,0 @@
# Test Case 0 -> nothing fails

9
test/testcase1.conf
View File

@@ -1,9 +0,0 @@
# Test Case example config
# You configure which methods should fail
# Format <round>_<rank> = <operation>
# <operation> can be one of the following = allreduce, broadcast, loadcheckpoint, checkpoint
1_0 = allreduce
1_1 = broadcast
2_2 = allreduce