void ArrayVersion_misaligned(BenchmarkExt<int>& bench)
{
bench.beginImplementation("Array<T,1> (misal.)");
while (!bench.doneImplementationBenchmark())
{
int N = bench.getParameter();
long iters = bench.getIterations();
cout << bench.currentImplementation() << ": N = " << N << endl;
Array<double,1> xfill(N+6);
Array<double,1> x(xfill(Range(0,N+0-1)));
initializeRandomDouble(x.dataFirst(), N);
Array<double,1> afill(N+6);
Array<double,1> a(afill(Range(1,N+1-1)));
initializeRandomDouble(a.dataFirst(), N);
Array<double,1> bfill(N+6);
Array<double,1> b(bfill(Range(2,N+2-1)));
initializeRandomDouble(b.dataFirst(), N);
Array<double,1> cfill(N+6);
Array<double,1> c(cfill(Range(3,N+3-1)));
initializeRandomDouble(c.dataFirst(), N);
Array<double,1> dfill(N+6);
Array<double,1> d(dfill(Range(4,N+4-1)));
initializeRandomDouble(d.dataFirst(), N);
Array<double,1> yfill(N+6);
Array<double,1> y(yfill(Range(5,N+5-1)));
initializeRandomDouble(y.dataFirst(), N);
bench.start();
for (long i=0; i < iters; ++i)
{
x = a*b + c*d; y = b+d;
sink();
}
bench.stop();
bench.startOverhead();
for (long i=0; i < iters; ++i) {
sink();
}
bench.stopOverhead();
}
bench.endImplementation();
}
void ArrayVersion_unaligned(BenchmarkExt<int>& bench)
{
bench.beginImplementation("Array<T,1> (unal.)");
while (!bench.doneImplementationBenchmark())
{
int N = bench.getParameter();
long iters = bench.getIterations();
cout << bench.currentImplementation() << ": N = " << N << endl;
Array<double,1> xfill(N+1);
Array<double,1> x(xfill(Range(1,N)));
initializeRandomDouble(x.dataFirst(), N);
Array<double,1> afill(N+1);
Array<double,1> a(afill(Range(1,N)));
initializeRandomDouble(a.dataFirst(), N);
Array<double,1> bfill(N+1);
Array<double,1> b(bfill(Range(1,N)));
initializeRandomDouble(b.dataFirst(), N);
Array<double,1> cfill(N+1);
Array<double,1> c(cfill(Range(1,N)));
initializeRandomDouble(c.dataFirst(), N);
Array<double,1> dfill(N+1);
Array<double,1> d(dfill(Range(1,N)));
initializeRandomDouble(d.dataFirst(), N);
bench.start();
for (long i=0; i < iters; ++i)
{
x = (a+b)*(c+d);
sink();
}
bench.stop();
bench.startOverhead();
for (long i=0; i < iters; ++i) {
sink();
}
bench.stopOverhead();
}
bench.endImplementation();
}
void ArrayVersion_misaligned(BenchmarkExt<int>& bench, double u, double v, double w, double z)
{
bench.beginImplementation("Array<T,1> (misal.)");
while (!bench.doneImplementationBenchmark())
{
int N = bench.getParameter();
long iters = bench.getIterations();
cout << bench.currentImplementation() << ": N = " << N << endl;
Array<double,1> afill(N+5);
Array<double,1> a(afill(Range(0,N+0-1)));
initializeRandomDouble(a.dataFirst(), N);
Array<double,1> bfill(N+5);
Array<double,1> b(bfill(Range(1,N+1-1)));
initializeRandomDouble(b.dataFirst(), N);
Array<double,1> cfill(N+5);
Array<double,1> c(cfill(Range(2,N+2-1)));
initializeRandomDouble(c.dataFirst(), N);
Array<double,1> dfill(N+5);
Array<double,1> d(dfill(Range(3,N+3-1)));
initializeRandomDouble(d.dataFirst(), N);
Array<double,1> xfill(N+5);
Array<double,1> x(xfill(Range(4,N+4-1)));
initializeRandomDouble(x.dataFirst(), N);
bench.start();
for (long i=0; i < iters; ++i)
{
x=(1.0-c*c)/((4*w)*sin(1.0+c*c-2*v*c))*a*b*u*exp(-z*d);
sink();
}
bench.stop();
bench.startOverhead();
for (long i=0; i < iters; ++i) {
sink();
}
bench.stopOverhead();
}
bench.endImplementation();
}
//////////////////
// parsing function
//////////////////
void parse(char* prev_link, char* page){
char *tmp, *link, *copy;
char *saveptr = NULL;
copy = strdup(page);
tmp = strtok_r(page, " \n", &saveptr);
while(tmp != NULL){
if((link = checkif_link(tmp)) != NULL){
pthread_mutex_lock(&d_lock);
while(downldItems == dq_size)
pthread_cond_wait(&dfull, &d_lock);
if(!check_visited(link)){
dfill(link);//means that the link has not been downloaded before
pthread_cond_signal(&dempty);
}
pthread_mutex_unlock(&d_lock);
_edge_function(prev_link, link);
}
tmp = strtok_r(NULL, " \n", &saveptr);
}
/////////
pthread_mutex_lock(&w_lock);
work--;
pthread_cond_signal(&done);
pthread_mutex_unlock(&w_lock);
}
///Main function! Kind of.
int crawl(char *start_url,
int download_workers,
int parse_workers,
int queue_size,
char * (*_fetch_fn)(char *url),
void (*_edge_fn)(char *from, char *to)) {
//start code here:
pthread_initialize();
//assign queue_size to global var
dq_size = queue_size;
_fetch_function = _fetch_fn;
_edge_function = _edge_fn;
p_init();
d_init();
dfill(start_url);
check_visited(start_url);
pthread_t pid[download_workers], cid[parse_workers];
int i;
for(i = 0; i < download_workers; i++){
pthread_create(&pid[i], NULL, download_thread, NULL);
}
for(i = 0; i < parse_workers; i++){
pthread_create(&cid[i], NULL, parse_thread, NULL);
}
pthread_mutex_lock(&w_lock);
while(work > 0){
pthread_cond_wait(&done, &w_lock);
}
pthread_mutex_unlock(&w_lock);
return 0;
}
/*
* Main program : Molecular Dynamics simulation.
*/
int main(){
int move;
double x[npart*3], vh[npart*3], f[npart*3];
double ekin;
double vel;
double sc;
double start, time;
/*
* Parameter definitions
*/
double den = 0.83134;
double side = pow((double)npart/den,0.3333333);
double tref = 0.722;
double rcoff = (double)mm/4.0;
double h = 0.064;
int irep = 10;
int istop = 20;
int iprint = 5;
int movemx = 20;
double a = side/(double)mm;
double hsq = h*h;
double hsq2 = hsq*0.5;
double tscale = 16.0/((double)npart-1.0);
double vaver = 1.13*sqrt(tref/24.0);
/*
* Initial output
*/
printf(" Molecular Dynamics Simulation example program\n");
printf(" ---------------------------------------------\n");
printf(" number of particles is ............ %6d\n",npart);
printf(" side length of the box is ......... %13.6f\n",side);
printf(" cut off is ........................ %13.6f\n",rcoff);
printf(" reduced temperature is ............ %13.6f\n",tref);
printf(" basic timestep is ................. %13.6f\n",h);
printf(" temperature scale interval ........ %6d\n",irep);
printf(" stop scaling at move .............. %6d\n",istop);
printf(" print interval .................... %6d\n",iprint);
printf(" total no. of steps ................ %6d\n",movemx);
/*
* Generate fcc lattice for atoms inside box
*/
fcc(x, npart, mm, a);
/*
* Initialise velocities and forces (which are zero in fcc positions)
*/
mxwell(vh, 3*npart, h, tref);
dfill(3*npart, 0.0, f, 1);
/*
* Start of md
*/
printf("\n i ke pe e temp "
" pres vel rp\n ----- ---------- ----------"
" ---------- -------- -------- -------- ----\n");
start = secnds();
for (move=1; move<=movemx; move++) {
/*
* Move the particles and partially update velocities
*/
domove(3*npart, x, vh, f, side);
/*
* Compute forces in the new positions and accumulate the virial
* and potential energy.
*/
forces(npart, x, f, side, rcoff);
//kernelWrapper(npart,x,f,side,rcoff);
/*
* Scale forces, complete update of velocities and compute k.e.
*/
ekin=mkekin(npart, f, vh, hsq2, hsq);
/*
* Average the velocity and temperature scale if desired
*/
vel=velavg(npart, vh, vaver, h);
if (move<istop && fmod(move, irep)==0) {
sc=sqrt(tref/(tscale*ekin));
dscal(3*npart, sc, vh, 1);
ekin=tref/tscale;
}
/*
* Sum to get full potential energy and virial
*/
if (fmod(move, iprint)==0)
prnout(move, ekin, epot, tscale, vir, vel, count, npart, den);
}
time = secnds() - start;
printf("Time = %f\n",(float) time);
}