/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
manager_t* managerPtr;
client_t** clients;
TIMER_T start;
TIMER_T stop;
/* Initialization */
parseArgs(argc, (char** const)argv);
SIM_GET_NUM_CPU(global_params[PARAM_CLIENTS]);
managerPtr = initializeManager();
assert(managerPtr != NULL);
clients = initializeClients(managerPtr);
assert(clients != NULL);
long numThread = global_params[PARAM_CLIENTS];
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
/* Run transactions */
printf("Running clients... ");
fflush(stdout);
GOTO_SIM();
TIMER_READ(start);
#ifdef OTM
#pragma omp parallel
{
client_run(clients);
}
#else
thread_start(client_run, (void*)clients);
#endif
TIMER_READ(stop);
GOTO_REAL();
puts("done.");
printf("Time = %0.6lf\n",
TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
checkTables(managerPtr);
/* Clean up */
printf("Deallocating memory... ");
fflush(stdout);
freeClients(clients);
/*
* TODO: The contents of the manager's table need to be deallocated.
*/
manager_free(managerPtr);
puts("done.");
fflush(stdout);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
GOTO_REAL();
SETUP_NUMBER_TASKS(6);
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
SIM_GET_NUM_CPU(global_numThread);
TM_STARTUP(global_numThread, 0);
P_MEMORY_STARTUP(global_numThread);
SETUP_NUMBER_THREADS(global_numThread);
thread_startup(global_numThread);
int repeat = global_repeats;
double time_total = 0.0;
double energy_total = 0.0;
for (; repeat > 0; --repeat) {
global_meshPtr = mesh_alloc();
assert(global_meshPtr);
long initNumElement = mesh_read(global_meshPtr, global_inputPrefix);
global_workHeapPtr = heap_alloc(1, &element_heapCompare);
assert(global_workHeapPtr);
long initNumBadElement = initializeWork(global_workHeapPtr, global_meshPtr);
TIMER_T start;
TIMER_READ(start);
GOTO_SIM();
thread_start(process, NULL);
GOTO_REAL();
TIMER_T stop;
TIMER_READ(stop);
double time_tmp = TIMER_DIFF_SECONDS(start, stop);
PRINT_STATS();
time_total += time_tmp;
}
printf("Time = %0.3lf\n",
time_total);
fflush(stdout);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
MAIN(argc, argv) {
TIMER_T start;
TIMER_T stop;
GOTO_REAL();
parseArgs(argc, (char** const)argv);
long sz = (long)global_params[PARAM_SIZE];
global_array = (aligned_type_t*) malloc(sz * sizeof(aligned_type_t));
long k = 0;
for (; k < sz; k++) {
global_array[k].value = 0;
}
long numThread = global_params[PARAM_THREADS];
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
printf("Running clients... ");
fflush(stdout);
TIMER_READ(start);
GOTO_SIM();
thread_start(client_run, (void*)&numThread);
GOTO_REAL();
TIMER_READ(stop);
puts("done.");
printf("Time = %0.6lf\n", TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
long i = 0;
long sum = 0;
for (;i < sz; i++) {
sum += global_array[i].value;
// printf("%ld\n", global_array[i].value);
}
if (sum != 0) {
printf("Problem, sum was not zero!: %ld\n", sum);
}
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
GOTO_REAL();
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
long numThread = global_params[PARAM_THREAD];
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
maze_t* mazePtr = maze_alloc();
assert(mazePtr);
long numPathToRoute = maze_read(mazePtr, global_inputFile);
router_t* routerPtr = router_alloc(global_params[PARAM_XCOST],
global_params[PARAM_YCOST],
global_params[PARAM_ZCOST],
global_params[PARAM_BENDCOST]);
assert(routerPtr);
list_t* pathVectorListPtr = list_alloc(NULL);
assert(pathVectorListPtr);
/*
* Run transactions
*/
router_solve_arg_t routerArg = {routerPtr, mazePtr, pathVectorListPtr};
TIMER_T startTime;
TIMER_READ(startTime);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
router_solve((void *)&routerArg);
}
#else
thread_start(router_solve, (void*)&routerArg);
#endif
GOTO_REAL();
TIMER_T stopTime;
TIMER_READ(stopTime);
long numPathRouted = 0;
list_iter_t it;
list_iter_reset(&it, pathVectorListPtr);
while (list_iter_hasNext(&it, pathVectorListPtr)) {
vector_t* pathVectorPtr = (vector_t*)list_iter_next(&it, pathVectorListPtr);
numPathRouted += vector_getSize(pathVectorPtr);
}
printf("Paths routed = %li\n", numPathRouted);
printf("Elapsed time = %f seconds\n", TIMER_DIFF_SECONDS(startTime, stopTime));
/*
* Check solution and clean up
*/
assert(numPathRouted <= numPathToRoute);
bool_t status = maze_checkPaths(mazePtr, pathVectorListPtr, global_doPrint);
assert(status == TRUE);
puts("Verification passed.");
maze_free(mazePtr);
router_free(routerPtr);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
long numThread = global_params[PARAM_THREAD];
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
maze_t* mazePtr = maze_alloc();
assert(mazePtr);
long numPathToRoute = maze_read(mazePtr, global_inputFile);
router_t* routerPtr = router_alloc(global_params[PARAM_XCOST],
global_params[PARAM_YCOST],
global_params[PARAM_ZCOST],
global_params[PARAM_BENDCOST]);
assert(routerPtr);
list_t* pathVectorListPtr = list_alloc(NULL);
assert(pathVectorListPtr);
/*
* Run transactions
*/
router_solve_arg_t routerArg = {routerPtr, mazePtr, pathVectorListPtr};
// NB: Since ASF/PTLSim "REAL" is native execution, and since we are using
// wallclock time, we want to be sure we read time inside the
// simulator, or else we report native cycles spent on the benchmark
// instead of simulator cycles.
GOTO_SIM();
TIMER_T startTime;
TIMER_READ(startTime);
#ifdef OTM
#pragma omp parallel
{
router_solve((void *)&routerArg);
}
#else
thread_start(router_solve, (void*)&routerArg);
#endif
TIMER_T stopTime;
TIMER_READ(stopTime);
// NB: As above, timer reads must be done inside of the simulated region
// for PTLSim/ASF
GOTO_REAL();
long numPathRouted = 0;
list_iter_t it;
list_iter_reset(&it, pathVectorListPtr);
while (list_iter_hasNext(&it, pathVectorListPtr)) {
vector_t* pathVectorPtr = (vector_t*)list_iter_next(&it, pathVectorListPtr);
numPathRouted += vector_getSize(pathVectorPtr);
}
printf("Paths routed = %li\n", numPathRouted);
printf("Elapsed time = %f seconds\n", TIMER_DIFF_SECONDS(startTime, stopTime));
/*
* Check solution and clean up
*/
assert(numPathRouted <= numPathToRoute);
bool status = maze_checkPaths(mazePtr, pathVectorListPtr, global_doPrint);
assert(status);
puts("Verification passed.");
maze_free(mazePtr);
router_free(routerPtr);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
char exitmsg[1024];
GOTO_REAL();
load_syncchar_map("sync_char.map.yada");
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
sprintf(exitmsg, "END BENCHMARK %s-parallel-phase\n", argv[0]);
SIM_GET_NUM_CPU(global_numThread);
TM_STARTUP(global_numThread);
P_MEMORY_STARTUP(global_numThread);
thread_startup(global_numThread);
global_meshPtr = mesh_alloc();
assert(global_meshPtr);
printf("Angle constraint = %lf\n", global_angleConstraint);
printf("Reading input... ");
long initNumElement = mesh_read(global_meshPtr, global_inputPrefix);
puts("done.");
global_workHeapPtr = heap_alloc(1, &element_heapCompare);
assert(global_workHeapPtr);
long initNumBadElement = initializeWork(global_workHeapPtr, global_meshPtr);
printf("Initial number of mesh elements = %li\n", initNumElement);
printf("Initial number of bad elements = %li\n", initNumBadElement);
printf("Starting triangulation...");
fflush(stdout);
/*
* Run benchmark
*/
TIMER_T start;
TIMER_READ(start);
OSA_PRINT("entering parallel phase\n",0);
START_INSTRUMENTATION();
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
process();
}
#else
thread_start(process, NULL);
#endif
GOTO_REAL();
OSA_PRINT("exiting parallel phase\n",0);
OSA_PRINT(exitmsg,0);
STOP_INSTRUMENTATION();
TIMER_T stop;
TIMER_READ(stop)
puts(" done.");
printf("Elapsed time = %0.3lf\n",
TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
/*
* Check solution
*/
long finalNumElement = initNumElement + global_totalNumAdded;
printf("Final mesh size = %li\n", finalNumElement);
printf("Number of elements processed = %li\n", global_numProcess);
fflush(stdout);
#if 0
bool_t isSuccess = mesh_check(global_meshPtr, finalNumElement);
#else
bool_t isSuccess = TRUE;
#endif
printf("Final mesh is %s\n", (isSuccess ? "valid." : "INVALID!"));
fflush(stdout);
assert(isSuccess);
/*
* TODO: deallocate mesh and work heap
*/
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN (argc,argv)
{
TIMER_T start;
TIMER_T stop;
GOTO_REAL();
/* Initialization */
parseArgs(argc, (char** const)argv);
SIM_GET_NUM_CPU(global_params[PARAM_THREAD]);
printf("Creating gene and segments... ");
fflush(stdout);
long geneLength = global_params[PARAM_GENE];
long segmentLength = global_params[PARAM_SEGMENT];
long minNumSegment = global_params[PARAM_NUMBER];
long numThread = global_params[PARAM_THREAD];
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
random_t* randomPtr = random_alloc();
assert(randomPtr != NULL);
random_seed(randomPtr, 0);
gene_t* genePtr = gene_alloc(geneLength);
assert( genePtr != NULL);
gene_create(genePtr, randomPtr);
char* gene = genePtr->contents;
segments_t* segmentsPtr = segments_alloc(segmentLength, minNumSegment);
assert(segmentsPtr != NULL);
segments_create(segmentsPtr, genePtr, randomPtr);
sequencer_t* sequencerPtr = sequencer_alloc(geneLength, segmentLength, segmentsPtr);
assert(sequencerPtr != NULL);
puts("done.");
printf("Gene length = %li\n", genePtr->length);
printf("Segment length = %li\n", segmentsPtr->length);
printf("Number segments = %li\n", vector_getSize(segmentsPtr->contentsPtr));
fflush(stdout);
/* Benchmark */
printf("Sequencing gene... ");
fflush(stdout);
TIMER_READ(start);
GOTO_SIM();
thread_startup(numThread, sequencer_run, (void*)sequencerPtr);
thread_start();
GOTO_REAL();
TIMER_READ(stop);
puts("done.");
printf("Time = %lf\n", TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
/* Check result */
{
char* sequence = sequencerPtr->sequence;
int result = strcmp(gene, sequence);
printf("Sequence matches gene: %s\n", (result ? "no" : "yes"));
if (result) {
printf("gene = %s\n", gene);
printf("sequence = %s\n", sequence);
}
fflush(stdout);
assert(strlen(sequence) >= strlen(gene));
}
/* Clean up */
printf("Deallocating memory... ");
fflush(stdout);
sequencer_free(sequencerPtr);
segments_free(segmentsPtr);
gene_free(genePtr);
random_free(randomPtr);
puts("done.");
fflush(stdout);
al_dump(&sequencerLock);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
GOTO_REAL();
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
long numThread = global_params[PARAM_THREAD];
long numVar = global_params[PARAM_VAR];
long numRecord = global_params[PARAM_RECORD];
long randomSeed = global_params[PARAM_SEED];
long maxNumParent = global_params[PARAM_NUMBER];
long percentParent = global_params[PARAM_PERCENT];
global_insertPenalty = global_params[PARAM_INSERT];
global_maxNumEdgeLearned = global_params[PARAM_EDGE];
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
printf("Random seed = %li\n", randomSeed);
printf("Number of vars = %li\n", numVar);
printf("Number of records = %li\n", numRecord);
printf("Max num parents = %li\n", maxNumParent);
printf("%% chance of parent = %li\n", percentParent);
printf("Insert penalty = %li\n", global_insertPenalty);
printf("Max num edge learned / var = %li\n", global_maxNumEdgeLearned);
printf("Operation quality factor = %f\n", global_operationQualityFactor);
fflush(stdout);
/*
* Generate data
*/
printf("Generating data... ");
fflush(stdout);
random_t* randomPtr = random_alloc();
assert(randomPtr);
random_seed(randomPtr, randomSeed);
data_t* dataPtr = data_alloc(numVar, numRecord, randomPtr);
assert(dataPtr);
net_t* netPtr = data_generate(dataPtr, -1, maxNumParent, percentParent);
puts("done.");
fflush(stdout);
/*
* Generate adtree
*/
adtree_t* adtreePtr = adtree_alloc();
assert(adtreePtr);
printf("Generating adtree... ");
fflush(stdout);
TIMER_T adtreeStartTime;
TIMER_READ(adtreeStartTime);
adtree_make(adtreePtr, dataPtr);
TIMER_T adtreeStopTime;
TIMER_READ(adtreeStopTime);
puts("done.");
fflush(stdout);
printf("Adtree time = %f\n",
TIMER_DIFF_SECONDS(adtreeStartTime, adtreeStopTime));
fflush(stdout);
/*
* Score original network
*/
float actualScore = score(netPtr, adtreePtr);
net_free(netPtr);
/*
* Learn structure of Bayesian network
*/
learner_t* learnerPtr = learner_alloc(dataPtr, adtreePtr, numThread);
assert(learnerPtr);
data_free(dataPtr); /* save memory */
printf("Learning structure...");
fflush(stdout);
TIMER_T learnStartTime;
TIMER_READ(learnStartTime);
GOTO_SIM();
learner_run(learnerPtr);
GOTO_REAL();
TIMER_T learnStopTime;
TIMER_READ(learnStopTime);
puts("done.");
fflush(stdout);
printf("Time = %f\n",
TIMER_DIFF_SECONDS(learnStartTime, learnStopTime));
fflush(stdout);
/*
* Check solution
*/
bool_t status = net_isCycle(learnerPtr->netPtr);
assert(!status);
#ifndef SIMULATOR
float learnScore = learner_score(learnerPtr);
printf("Learn score = %f\n", learnScore);
#endif
printf("Actual score = %f\n", actualScore);
/*
* Clean up
*/
fflush(stdout);
random_free(randomPtr);
#ifndef SIMULATOR
adtree_free(adtreePtr);
# if 0
learner_free(learnerPtr);
# endif
#endif
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
MAIN(argc, argv)
{
GOTO_REAL();
SETUP_NUMBER_TASKS(10);
/*
* Tuple for Scalable Data Generation
* stores startVertex, endVertex, long weight and other info
*/
graphSDG* SDGdata;
/*
* The graph data structure for this benchmark - see defs.h
*/
graph* G;
#ifdef ENABLE_KERNEL2
/*
* Kernel 2
*/
edge* maxIntWtList;
edge* soughtStrWtList;
long maxIntWtListSize;
long soughtStrWtListSize;
#endif /* ENABLE_KERNEL2 */
#ifdef ENABLE_KERNEL3
# ifndef ENABLE_KERNEL2
# error KERNEL3 requires KERNEL2
# endif
/*
* Kernel 3
*/
V* intWtVList = NULL;
V* strWtVList = NULL;
Vl** intWtVLList = NULL;
Vl** strWtVLList = NULL;
Vd* intWtVDList = NULL;
Vd* strWtVDList = NULL;
#endif /* ENABLE_KERNEL3 */
double totalTime = 0.0;
/* -------------------------------------------------------------------------
* Preamble
* -------------------------------------------------------------------------
*/
/*
* User Interface: Configurable parameters, and global program control
*/
getUserParameters(argc, (char** const) argv);
SIM_GET_NUM_CPU(THREADS);
TM_STARTUP(THREADS, 0);
P_MEMORY_STARTUP(THREADS);
SETUP_NUMBER_THREADS(THREADS);
thread_startup(THREADS);
double time_total = 0.0;
int repeat = REPEATS;
for (; repeat > 0; --repeat) {
SDGdata = (graphSDG*)malloc(sizeof(graphSDG));
assert(SDGdata);
genScalData_seq(SDGdata);
G = (graph*)malloc(sizeof(graph));
assert(G);
computeGraph_arg_t computeGraphArgs;
computeGraphArgs.GPtr = G;
computeGraphArgs.SDGdataPtr = SDGdata;
TIMER_T start;
TIMER_READ(start);
GOTO_SIM();
thread_start(computeGraph, (void*)&computeGraphArgs);
GOTO_REAL();
TIMER_T stop;
TIMER_READ(stop);
double time_tmp = TIMER_DIFF_SECONDS(start, stop);
PRINT_STATS();
time_total += time_tmp;
}
totalTime += time_total;
#ifdef ENABLE_KERNEL2
/* -------------------------------------------------------------------------
* Kernel 2 - Find Max weight and sought string
* -------------------------------------------------------------------------
*/
printf("\nKernel 2 - getStartLists() beginning execution...\n");
maxIntWtListSize = 0;
soughtStrWtListSize = 0;
maxIntWtList = (edge*)malloc(sizeof(edge));
assert(maxIntWtList);
soughtStrWtList = (edge*)malloc(sizeof(edge));
assert(soughtStrWtList);
getStartLists_arg_t getStartListsArg;
getStartListsArg.GPtr = G;
getStartListsArg.maxIntWtListPtr = &maxIntWtList;
getStartListsArg.maxIntWtListSize = &maxIntWtListSize;
getStartListsArg.soughtStrWtListPtr = &soughtStrWtList;
getStartListsArg.soughtStrWtListSize = &soughtStrWtListSize;
TIMER_READ(start);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
getStartLists((void*)&getStartListsArg);
}
#else
thread_start(getStartLists, (void*)&getStartListsArg);
#endif
GOTO_REAL();
TIMER_T stop;
TIMER_READ(stop);
time = TIMER_DIFF_SECONDS(start, stop);
totalTime += time;
printf("\n\tgetStartLists() completed execution.\n");
printf("\nTime taken for kernel 2 is %9.6f sec.\n\n", time);
#endif /* ENABLE_KERNEL2 */
#ifdef ENABLE_KERNEL3
/* -------------------------------------------------------------------------
* Kernel 3 - Graph Extraction
* -------------------------------------------------------------------------
*/
printf("\nKernel 3 - findSubGraphs() beginning execution...\n");
if (K3_DS == 0) {
intWtVList = (V*)malloc(G->numVertices * maxIntWtListSize * sizeof(V));
assert(intWtVList);
strWtVList = (V*)malloc(G->numVertices * soughtStrWtListSize * sizeof(V));
assert(strWtVList);
findSubGraphs0_arg_t findSubGraphs0Arg;
findSubGraphs0Arg.GPtr = G;
findSubGraphs0Arg.intWtVList = intWtVList;
findSubGraphs0Arg.strWtVList = strWtVList;
findSubGraphs0Arg.maxIntWtList = maxIntWtList;
findSubGraphs0Arg.maxIntWtListSize = maxIntWtListSize;
findSubGraphs0Arg.soughtStrWtList = soughtStrWtList;
findSubGraphs0Arg.soughtStrWtListSize = soughtStrWtListSize;
TIMER_READ(start);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
findSubGraphs0((void*)&findSubGraphs0Arg);
}
#else
thread_start(findSubGraphs0, (void*)&findSubGraphs0Arg);
#endif
GOTO_REAL();
TIMER_READ(stop);
} else if (K3_DS == 1) {
intWtVLList = (Vl**)malloc(maxIntWtListSize * sizeof(Vl*));
assert(intWtVLList);
strWtVLList = (Vl**)malloc(soughtStrWtListSize * sizeof(Vl*));
assert(strWtVLList);
findSubGraphs1_arg_t findSubGraphs1Arg;
findSubGraphs1Arg.GPtr = G;
findSubGraphs1Arg.intWtVLList = intWtVLList;
findSubGraphs1Arg.strWtVLList = strWtVLList;
findSubGraphs1Arg.maxIntWtList = maxIntWtList;
findSubGraphs1Arg.maxIntWtListSize = maxIntWtListSize;
findSubGraphs1Arg.soughtStrWtList = soughtStrWtList;
findSubGraphs1Arg.soughtStrWtListSize = soughtStrWtListSize;
TIMER_READ(start);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
findSubGraphs1((void*)&findSubGraphs1Arg);
}
#else
thread_start(findSubGraphs1, (void*)&findSubGraphs1Arg);
#endif
GOTO_REAL();
TIMER_READ(stop);
/* Verification
on_one_thread {
for (i=0; i<maxIntWtListSize; i++) {
printf("%ld -- ", i);
currV = intWtVLList[i];
while (currV != NULL) {
printf("[%ld %ld] ", currV->num, currV->depth);
currV = currV->next;
}
printf("\n");
}
for (i=0; i<soughtStrWtListSize; i++) {
printf("%ld -- ", i);
currV = strWtVLList[i];
while (currV != NULL) {
printf("[%ld %ld] ", currV->num, currV->depth);
currV = currV->next;
}
printf("\n");
}
}
*/
} else if (K3_DS == 2) {
intWtVDList = (Vd *) malloc(maxIntWtListSize * sizeof(Vd));
assert(intWtVDList);
strWtVDList = (Vd *) malloc(soughtStrWtListSize * sizeof(Vd));
assert(strWtVDList);
findSubGraphs2_arg_t findSubGraphs2Arg;
findSubGraphs2Arg.GPtr = G;
findSubGraphs2Arg.intWtVDList = intWtVDList;
findSubGraphs2Arg.strWtVDList = strWtVDList;
findSubGraphs2Arg.maxIntWtList = maxIntWtList;
findSubGraphs2Arg.maxIntWtListSize = maxIntWtListSize;
findSubGraphs2Arg.soughtStrWtList = soughtStrWtList;
findSubGraphs2Arg.soughtStrWtListSize = soughtStrWtListSize;
TIMER_READ(start);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
findSubGraphs2((void*)&findSubGraphs2Arg);
}
#else
thread_start(findSubGraphs2, (void*)&findSubGraphs2Arg);
#endif
GOTO_REAL();
TIMER_READ(stop);
/* Verification */
/*
on_one_thread {
printf("\nInt weight sub-graphs \n");
for (i=0; i<maxIntWtListSize; i++) {
printf("%ld -- ", i);
for (j=0; j<intWtVDList[i].numArrays; j++) {
printf("\n [Array %ld] - \n", j);
for (k=0; k<intWtVDList[i].arraySize[j]; k++) {
printf("[%ld %ld] ", intWtVDList[i].vList[j][k].num, intWtVDList[i].vList[j][k].depth);
}
}
printf("\n");
}
printf("\nStr weight sub-graphs \n");
for (i=0; i<soughtStrWtListSize; i++) {
printf("%ld -- ", i);
for (j=0; j<strWtVDList[i].numArrays; j++) {
printf("\n [Array %ld] - \n", j);
for (k=0; k<strWtVDList[i].arraySize[j]; k++) {
printf("[%ld %ld] ", strWtVDList[i].vList[j][k].num, strWtVDList[i].vList[j][k].depth);
}
}
printf("\n");
}
}
*/
} else {
assert(0);
}
time = TIMER_DIFF_SECONDS(start, stop);
totalTime += time;
printf("\n\tfindSubGraphs() completed execution.\n");
printf("\nTime taken for kernel 3 is %9.6f sec.\n\n", time);
#endif /* ENABLE_KERNEL3 */
#ifdef ENABLE_KERNEL4
/* -------------------------------------------------------------------------
* Kernel 4 - Graph Clustering
* -------------------------------------------------------------------------
*/
printf("\nKernel 4 - cutClusters() beginning execution...\n");
TIMER_READ(start);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
cutClusters((void*)G);
}
#else
thread_start(cutClusters, (void*)G);
#endif
GOTO_REAL();
TIMER_READ(stop);
time = TIMER_DIFF_SECONDS(start, stop);
totalTime += time;
printf("\n\tcutClusters() completed execution.\n");
printf("\nTime taken for Kernel 4 is %9.6f sec.\n\n", time);
#endif /* ENABLE_KERNEL4 */
printf("Time = %9.6f \n", totalTime);
/* -------------------------------------------------------------------------
* Cleanup
* -------------------------------------------------------------------------
*/
P_FREE(G->outDegree);
P_FREE(G->outVertexIndex);
P_FREE(G->outVertexList);
P_FREE(G->paralEdgeIndex);
P_FREE(G->inDegree);
P_FREE(G->inVertexIndex);
P_FREE(G->inVertexList);
P_FREE(G->intWeight);
P_FREE(G->strWeight);
#ifdef ENABLE_KERNEL3
LONGINT_T i;
LONGINT_T j;
Vl* currV;
Vl* tempV;
if (K3_DS == 0) {
P_FREE(strWtVList);
P_FREE(intWtVList);
}
if (K3_DS == 1) {
for (i = 0; i < maxIntWtListSize; i++) {
currV = intWtVLList[i];
while (currV != NULL) {
tempV = currV->next;
P_FREE(currV);
currV = tempV;
}
}
for (i = 0; i < soughtStrWtListSize; i++) {
currV = strWtVLList[i];
while (currV != NULL) {
tempV = currV->next;
P_FREE(currV);
currV = tempV;
}
}
P_FREE(strWtVLList);
P_FREE(intWtVLList);
}
if (K3_DS == 2) {
for (i = 0; i < maxIntWtListSize; i++) {
for (j = 0; j < intWtVDList[i].numArrays; j++) {
P_FREE(intWtVDList[i].vList[j]);
}
P_FREE(intWtVDList[i].vList);
P_FREE(intWtVDList[i].arraySize);
}
for (i = 0; i < soughtStrWtListSize; i++) {
for (j = 0; j < strWtVDList[i].numArrays; j++) {
P_FREE(strWtVDList[i].vList[j]);
}
P_FREE(strWtVDList[i].vList);
P_FREE(strWtVDList[i].arraySize);
}
P_FREE(strWtVDList);
P_FREE(intWtVDList);
}
P_FREE(soughtStrWtList);
P_FREE(maxIntWtList);
#endif /* ENABLE_KERNEL2 */
P_FREE(SOUGHT_STRING);
P_FREE(G);
P_FREE(SDGdata);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN (argc,argv)
{
TIMER_T start;
TIMER_T stop;
/* Initialization */
parseArgs(argc, (char** const)argv);
SIM_GET_NUM_CPU(global_params[PARAM_THREAD]);
printf("Creating gene and segments... ");
fflush(stdout);
long geneLength = global_params[PARAM_GENE];
long segmentLength = global_params[PARAM_SEGMENT];
long minNumSegment = global_params[PARAM_NUMBER];
long numThread = global_params[PARAM_THREAD];
random_t* randomPtr;
gene_t* genePtr;
char* gene;
segments_t* segmentsPtr;
sequencer_t* sequencerPtr;
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
TM_THREAD_ENTER();
// TM_BEGIN();
randomPtr= random_alloc();
assert(randomPtr != NULL);
random_seed(randomPtr, 0);
genePtr = gene_alloc(geneLength);
assert( genePtr != NULL);
gene_create(genePtr, randomPtr);
gene = genePtr->contents;
segmentsPtr = segments_alloc(segmentLength, minNumSegment);
assert(segmentsPtr != NULL);
segments_create(segmentsPtr, genePtr, randomPtr);
sequencerPtr = sequencer_alloc(geneLength, segmentLength, segmentsPtr);
assert(sequencerPtr != NULL);
//TM_END();
thread_startup(numThread);
puts("done.");
printf("Gene length = %li\n", genePtr->length);
printf("Segment length = %li\n", segmentsPtr->length);
printf("Number segments = %li\n", vector_getSize(segmentsPtr->contentsPtr));
fflush(stdout);
/* Benchmark */
printf("Sequencing gene... ");
fflush(stdout);
// NB: Since ASF/PTLSim "REAL" is native execution, and since we are using
// wallclock time, we want to be sure we read time inside the
// simulator, or else we report native cycles spent on the benchmark
// instead of simulator cycles.
GOTO_SIM();
TIMER_READ(start);
#ifdef OTM
#pragma omp parallel
{
sequencer_run(sequencerPtr);
}
#else
thread_start(sequencer_run, (void*)sequencerPtr);
#endif
TIMER_READ(stop);
// NB: As above, timer reads must be done inside of the simulated region
// for PTLSim/ASF
GOTO_REAL();
puts("done.");
printf("Time = %lf\n", TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
/* Check result */
{
char* sequence;
int result;
//TM_BEGIN();
sequence= sequencerPtr->sequence;
result = strcmp(gene, sequence);
//TM_END();
printf("Sequence matches gene: %s\n", (result ? "no" : "yes"));
if (result) {
printf("gene = %s\n", gene);
printf("sequence = %s\n", sequence);
}
fflush(stdout);
assert(strlen(sequence) >= strlen(gene));
}
/* Clean up */
printf("Deallocating memory... ");
fflush(stdout);
sequencer_free(sequencerPtr);
segments_free(segmentsPtr);
gene_free(genePtr);
random_free(randomPtr);
puts("done.");
fflush(stdout);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
SIM_GET_NUM_CPU(global_numThread);
TM_STARTUP(global_numThread);
P_MEMORY_STARTUP(global_numThread);
thread_startup(global_numThread);
global_meshPtr = mesh_alloc();
assert(global_meshPtr);
printf("Angle constraint = %lf\n", global_angleConstraint);
printf("Reading input... ");
long initNumElement = mesh_read(global_meshPtr, (char*)global_inputPrefix);
puts("done.");
global_workHeapPtr = heap_alloc(1, &yada_heapcompare);
assert(global_workHeapPtr);
long initNumBadElement = initializeWork(global_workHeapPtr, global_meshPtr);
printf("Initial number of mesh elements = %li\n", initNumElement);
printf("Initial number of bad elements = %li\n", initNumBadElement);
printf("Starting triangulation...");
fflush(stdout);
/*
* Run benchmark
*/
// NB: Since ASF/PTLSim "REAL" is native execution, and since we are using
// wallclock time, we want to be sure we read time inside the
// simulator, or else we report native cycles spent on the benchmark
// instead of simulator cycles.
GOTO_SIM();
TIMER_T start;
TIMER_READ(start);
#ifdef OTM
#pragma omp parallel
{
process();
}
#else
thread_start((void(*)(void*))process, NULL);
#endif
TIMER_T stop;
TIMER_READ(stop);
// NB: As above, timer reads must be done inside of the simulated region
// for PTLSim/ASF
GOTO_REAL();
puts(" done.");
printf("Elapsed time = %0.3lf\n",
TIMER_DIFF_SECONDS(start, stop));
fflush(stdout);
/*
* Check solution
*/
long finalNumElement = initNumElement + global_totalNumAdded;
printf("Final mesh size = %li\n", finalNumElement);
printf("Number of elements processed = %li\n", global_numProcess);
fflush(stdout);
#if 1
bool isSuccess = mesh_check(global_meshPtr, finalNumElement);
#else
bool isSuccess = true;
#endif
printf("Final mesh is %s\n", (isSuccess ? "valid." : "INVALID!"));
fflush(stdout);
assert(isSuccess);
/*
* TODO: deallocate mesh and work heap
*/
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
GOTO_REAL();
/*
* Initialization
*/
parseArgs(argc, (char** const)argv);
long numThread = global_params[PARAM_THREAD];
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
long percentAttack = global_params[PARAM_ATTACK];
long maxDataLength = global_params[PARAM_LENGTH];
long numFlow = global_params[PARAM_NUM];
long randomSeed = global_params[PARAM_SEED];
printf("Percent attack = %li\n", percentAttack);
printf("Max data length = %li\n", maxDataLength);
printf("Num flow = %li\n", numFlow);
printf("Random seed = %li\n", randomSeed);
dictionary_t* dictionaryPtr = dictionary_alloc();
assert(dictionaryPtr);
stream_t* streamPtr = stream_alloc(percentAttack);
assert(streamPtr);
long numAttack = stream_generate(streamPtr,
dictionaryPtr,
numFlow,
randomSeed,
maxDataLength);
printf("Num attack = %li\n", numAttack);
decoder_t* decoderPtr = decoder_alloc();
assert(decoderPtr);
vector_t** errorVectors = (vector_t**)malloc(numThread * sizeof(vector_t*));
assert(errorVectors);
long i;
for (i = 0; i < numThread; i++) {
vector_t* errorVectorPtr = vector_alloc(numFlow);
assert(errorVectorPtr);
errorVectors[i] = errorVectorPtr;
}
arg_t arg;
arg.streamPtr = streamPtr;
arg.decoderPtr = decoderPtr;
arg.errorVectors = errorVectors;
/*
* Run transactions
*/
TIMER_T startTime;
TIMER_READ(startTime);
GOTO_SIM();
#ifdef OTM
#pragma omp parallel
{
processPackets((void*)&arg);
}
#else
thread_start(processPackets, (void*)&arg);
#endif
GOTO_REAL();
TIMER_T stopTime;
TIMER_READ(stopTime);
printf("\nTime = %lf\n", TIMER_DIFF_SECONDS(startTime, stopTime));
/*
* Check solution
*/
/*long numFound = 0;
for (i = 0; i < numThread; i++) {
vector_t* errorVectorPtr = errorVectors[i];
long e;
long numError = vector_getSize(errorVectorPtr);
numFound += numError;
for (e = 0; e < numError; e++) {
long flowId = (long)vector_at(errorVectorPtr, e);
bool_t status = stream_isAttack(streamPtr, flowId);
assert(status);
}
}*/
/*printf("Num found = %li\n", numFound);
assert(numFound == numAttack);*/
/*
* Clean up
*/
for (i = 0; i < numThread; i++) {
vector_free(errorVectors[i]);
}
free(errorVectors);
decoder_free(decoderPtr);
stream_free(streamPtr);
dictionary_free(dictionaryPtr);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
GOTO_REAL();
/*
* Initialization
*/
SETUP_NUMBER_TASKS(3);
parseArgs(argc, (char** const)argv);
long numThread = global_params[PARAM_THREAD];
SETUP_NUMBER_THREADS(numThread);
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread, 0);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
long percentAttack = global_params[PARAM_ATTACK];
long maxDataLength = global_params[PARAM_LENGTH];
long numFlow = global_params[PARAM_NUM];
long randomSeed = global_params[PARAM_SEED];
/* printf("Percent attack = %li\n", percentAttack);
printf("Max data length = %li\n", maxDataLength);
printf("Num flow = %li\n", numFlow);
printf("Random seed = %li\n", randomSeed); */
double time_total = 0.0;
int repeats = global_params[PARAM_REPEAT];
for (; repeats > 0; --repeats) {
dictionary_t* dictionaryPtr = dictionary_alloc();
assert(dictionaryPtr);
stream_t* streamPtr = stream_alloc(percentAttack);
assert(streamPtr);
long numAttack = stream_generate(streamPtr,
dictionaryPtr,
numFlow,
randomSeed,
maxDataLength);
// printf("Num attack = %li\n", numAttack);
decoder_t* decoderPtr = decoder_alloc();
assert(decoderPtr);
vector_t** errorVectors = (vector_t**)malloc(numThread * sizeof(vector_t*));
assert(errorVectors);
long i;
for (i = 0; i < numThread; i++) {
vector_t* errorVectorPtr = vector_alloc(numFlow);
assert(errorVectorPtr);
errorVectors[i] = errorVectorPtr;
}
arg_t arg;
arg.streamPtr = streamPtr;
arg.decoderPtr = decoderPtr;
arg.errorVectors = errorVectors;
/*
* Run transactions
*/
TIMER_T startTime;
TIMER_READ(startTime);
tm_time_t start_clock=TM_TIMER_READ();
GOTO_SIM();
thread_start(processPackets, (void*)&arg);
GOTO_REAL();
TIMER_T stopTime;
tm_time_t end_clock=TM_TIMER_READ();
TIMER_READ(stopTime);
double time_tmp = TIMER_DIFF_SECONDS(startTime, stopTime);
time_total += time_tmp;
PRINT_STATS();
PRINT_CLOCK_THROUGHPUT(end_clock-start_clock);
/*
* Check solution
*/
long numFound = 0;
for (i = 0; i < numThread; i++) {
vector_t* errorVectorPtr = errorVectors[i];
long e;
long numError = vector_getSize(errorVectorPtr);
numFound += numError;
for (e = 0; e < numError; e++) {
long flowId = (long)vector_at(errorVectorPtr, e);
bool_t status = stream_isAttack(streamPtr, flowId);
assert(status);
}
}
// printf("Num found = %li\n", numFound);
assert(numFound == numAttack);
/*
* Clean up
*/
for (i = 0; i < numThread; i++) {
vector_free(errorVectors[i]);
}
free(errorVectors);
decoder_free(decoderPtr);
stream_free(streamPtr);
dictionary_free(dictionaryPtr);
}
printf("Time = %f\n", time_total);
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
MAIN_RETURN(0);
}
/* =============================================================================
* main
* =============================================================================
*/
MAIN(argc, argv)
{
int i;
//FILE * wfile;
# ifdef SIMULATOR
printf("SIMULATOR is defined\n");
# else
printf("SIMULATOR is not defined\n");
# endif /* !SIMULATOR */
#if defined(STM)
printf("STM is defined\n");
# else
printf("STM is not defined\n");
# endif /* !SIMULATOR */
GOTO_REAL();
/*
* Initialization
*/
if(argc !=3)
printf("needs two input arguments, \"max_count, num_threads\",argc=%d\n\n", argc);
max_count = atoi(argv[1]);
long numThread = atoi(argv[2]);
if(numThread > MAX_NUM_OF_THREADS)
printf("num_threads can be at most %d\n\n", MAX_NUM_OF_THREADS);
//else
//printf("numThread = %d\n\n", (int)numThread);
for(i=0; i<MAX_NUM_OF_THREADS; i++)
threads_arg[i]=i;
SIM_GET_NUM_CPU(numThread);
TM_STARTUP(numThread);
P_MEMORY_STARTUP(numThread);
thread_startup(numThread);
/*
* Run transactions
*/
//router_solve_arg_t routerArg = {routerPtr, mazePtr, pathVectorListPtr};
TIMER_T startTime;
TIMER_READ(startTime);
GOTO_SIM();
thread_start(func_count, (void*)threads_arg);
GOTO_REAL();
TIMER_T stopTime;
TIMER_READ(stopTime);
//printf("Elapsed time = %f seconds\n", TIMER_DIFF_SECONDS(startTime, stopTime));
/*
* Check solution and clean up
*/
TM_SHUTDOWN();
P_MEMORY_SHUTDOWN();
GOTO_SIM();
thread_shutdown();
//wfile = fopen ("exe_time.txt","a");
//fprintf(wfile, "Elapsed time = %f, Aborts/starts=%f, Aborts=%f, starts=%f\n", TIMER_DIFF_SECONDS(startTime, stopTime), (float)((float)(AbortTally*100)/((float)StartTally)), (float)AbortTally, (float)StartTally);
//fclose(wfile);
//printf("final value of counter=%d\n\n", my_counter);
MAIN_RETURN(0);
}