void
doNameAnnotations(iface::cellml_api::Model* aModel,
iface::cellml_services::CodeGenerator* aCG)
{
// Make an annotation set...
iface::cellml_services::AnnotationToolService* ats
(CreateAnnotationToolService());
iface::cellml_services::AnnotationSet* as(ats->createAnnotationSet());
ats->release_ref();
aCG->useAnnoSet(as);
// Now we go through all variables in the model and set their annotations...
iface::cellml_api::CellMLComponentSet* ccs = aModel->allComponents();
iface::cellml_api::CellMLComponentIterator* cci = ccs->iterateComponents();
ccs->release_ref();
iface::cellml_api::CellMLComponent* comp;
while ((comp = cci->nextComponent()) != NULL)
{
iface::cellml_api::CellMLVariableSet* vs(comp->variables());
std::wstring compname = comp->name();
comp->release_ref();
iface::cellml_api::CellMLVariableIterator* vi(vs->iterateVariables());
vs->release_ref();
iface::cellml_api::CellMLVariable* v;
while ((v = vi->nextVariable()) != NULL)
{
std::wstring name = v->name();
std::wstring varn = compname;
varn += L"_";
varn += name;
std::wstring raten = L"rate_";
raten += varn;
as->setStringAnnotation(v, L"expression", varn.c_str());
as->setStringAnnotation(v, L"expression_d1", raten.c_str());
v->release_ref();
}
vi->release_ref();
}
cci->release_ref();
as->release_ref();
}
bool TraceLog::activateTraceLog( bool isActive ){
if ( isActive ){
try{
std::string traceFileName = "traceLog"+getTime()+".log";
out.open(traceFileName.c_str() );
std::streambuf *countbuf = std::cout.rdbuf();
std::cout.rdbuf ( out.rdbuf() ) ;
AddTimeStamp ats( cout );
std::cout<<"buffer pointed\n";
}
catch ( std::exception &e ){
std::cerr<<"exception caight\n";
return false;
}
return true;
}
else{ //have to do some logic thus std::cout<<"print data\n" won;t print any thing to standard output.
}
}
int main () {
const int n_thre = omp_get_max_threads();
fill_initial_condition();
#pragma omp parallel
{
int tid=2*omp_get_thread_num();
for(int x=0;x<SX;++x) {
for(int y=0;y<SY;++y) {
for(int z=0;z<SZ;++z) {
double u,v; get_solution_at(0,x,y,z, u,v);
ats(sU0,tid,x,y,z)=u;
ats(sV0,tid,x,y,z)=v;
}
}
}
}
std::cerr << "Setting up wall values..." << std::endl;
for(int t = 0;t<T_MAX;++t){
/*
#pragma omp parallel
{
int tid=2*omp_get_thread_num();
for(int x=SX-2;x<SX;++x) {
for(int y=0;y<SY;++y) {
for(int z=0;z<SZ;++z) {
double u,v; get_solution_at(t,x+t,y+t,z+t, u,v);
Uwx[tid][t][x-(SX-2)][y][z] = u;
Vwx[tid][t][x-(SX-2)][y][z] = v;
}
}
}
for(int x=0;x<SX;++x) {
for(int y=SY-2;y<SY;++y) {
for(int z=0;z<SZ;++z) {
double u,v; get_solution_at(t,x+t,y+t,z+t, u,v);
Uwy[tid][t][x][y-(SY-2)][z] = u;
Vwy[tid][t][x][y-(SY-2)][z] = v;
}
}
}
for(int x=0;x<SX;++x) {
for(int y=0;y<SY;++y) {
for(int z=SZ-2;z<SZ;++z) {
double u,v; get_solution_at(t,x+t,y+t,z+t, u,v);
Uwz[tid][t][x][y][z-(SZ-2)] = u;
Vwz[tid][t][x][y][z-(SZ-2)] = v;
}
}
}
*/
}
for(int trial=0;trial<10;++trial) {
std::cerr << "Carrying out simulation..." << std::endl;
// set initial condition
#pragma omp parallel
{
const int tid=2*omp_get_thread_num();
for(int x=0;x<SX;++x) {
for(int y=0;y<SY;++y) {
for(int z=0;z<SZ;++z) {
ats(sU,tid,x,y,z)=ats(sU0,tid,x,y,z);
ats(sV,tid,x,y,z)=ats(sV0,tid,x,y,z);
}
}
}
}
double time_comp=0, time_comm=0, time_begin, time_end;
//for(int heating=0;heating<10;++heating) {
time_begin = wctime();
#pragma omp parallel
for (int tid=0;tid<n_thre;++tid) {
//const int tid=2*omp_get_thread_num();
for(int t = 0; t < T_MAX; ++t){
// destructively update the state
/*
const auto lap = [](Real ar[SX][SY][SZ],int x, int y, int z) {
auto ret = ar[x][y+1][z+1] + ar[x+2][y+1][z+1]
+ ar[x+1][y][z+1] + ar[x+1][y+2][z+1]
+ ar[x+1][y+1][z] + ar[x+1][y+1][z+2]
- 6*ar[x+1][y+1][z+1];
return ret / dx / dx;
};*/
#define lap(ar,b,x,y,z) \
(ats(ar,b,x,y+1,z+1) + ats(ar,b,x+2,y+1,z+1) \
+ ats(ar,b,x+1,y,z+1) + ats(ar,b,x+1,y+2,z+1) \
+ ats(ar,b,x+1,y+1,z) + ats(ar,b,x+1,y+1,z+2) \
- 6*ats(ar,b,x+1,y+1,z+1)) / dx / dx
#pragma omp for collapse(2)
for(int x=0;x<SX-2;++x) {
for(int y=0;y<SY-2;++y) {
#pragma omp simd
for(int z=0;z<SZ-2;++z) {
Real u=ats(sU,tid,x+1,y+1,z+1) ;
Real v=ats(sV,tid,x+1,y+1,z+1) ;
auto du_dt = -Fe * u*v*v + Fu*(1-u) + Du * lap(sU,tid,x,y,z);
auto dv_dt = Fe * u*v*v - Fv*v + Dv * lap(sV,tid,x,y,z);
ats(sU,tid,x,y,z) = u+dt*du_dt;
ats(sV,tid,x,y,z) = v+dt*dv_dt;
}
}
}
}
}
time_end = wctime();
//}
double flop = 29.0 * (SX-2)*(SY-2)*(SZ-2) *T_MAX * n_thre;
double bw_gb= 8.0 * 2 * 7 * (SX-2)*(SY-2)*(SZ-2) *T_MAX * n_thre;
double time_elapse = time_end-time_begin;
{
const int t = T_MAX;
double num[BANK]={0},den[BANK]={0};
#pragma omp parallel
{
int tid=2*omp_get_thread_num();
for(int x=0;x<SX-2;++x) {
for(int y=0;y<SY-2;++y) {
for(int z=0;z<SZ-2;++z) {
double u,v; get_solution_at(t,x+t,y+t,z+t, u,v);
num[tid] += std::abs(u-ats(sU,tid,x,y,z));
den[tid] += 1;
}
}
}
}
double sum_num = 0, sum_den = 0;
for(int i=0;i<BANK;++i){
sum_num += num[i];
sum_den += den[i];
}
std::ostringstream msg;
msg << n_thre << " " << SX << " " << SY << " " << SZ << " " << T_MAX << " "
<< " t: " << time_elapse << " " << time_comm << " " << time_comp << " GFlops: " << flop/time_elapse/1e9<< " GBps: " << bw_gb/time_elapse/1e9<< " error: " << (sum_num/sum_den);
std::ofstream log_file("benchmark-standalone.txt", std::ios::app);
std::cout << msg.str() << std::endl;
log_file << msg.str() << std::endl;
}
}
}
