portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-16 06:47:06 +01:00
Code Releases Activity

Threading support rework.

Browse Source 
Placed parallel pragmas as macros; implemented deterministic thread reduction in style of
BFM.
This commit is contained in:
Peter Boyle
2015-05-12 07:51:41 +01:00
parent b1d2c60d07
commit 6103c29ee3
26 changed files with 276 additions and 184 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
lib/lattice/Grid_lattice_arith.h
View File

@ -10,7 +10,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
conformable(lhs,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
mult(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@ -21,7 +21,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
conformable(lhs,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
mac(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@ -32,7 +32,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
conformable(lhs,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
sub(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@ -42,7 +42,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const Lattice<obj3> &rhs){
conformable(lhs,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
add(&tmp,&lhs._odata[ss],&rhs._odata[ss]);
@ -56,7 +56,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mult(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
conformable(lhs,ret);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
mult(&tmp,&lhs._odata[ss],&rhs);
@ -67,7 +67,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mac(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
conformable(lhs,ret);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
mac(&tmp,&lhs._odata[ss],&rhs);
@ -78,7 +78,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void sub(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
conformable(lhs,ret);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
sub(&tmp,&lhs._odata[ss],&rhs);
@ -88,7 +88,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void add(Lattice<obj1> &ret,const Lattice<obj2> &lhs,const obj3 &rhs){
conformable(lhs,ret);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
obj1 tmp;
add(&tmp,&lhs._odata[ss],&rhs);
@ -102,7 +102,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mult(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
conformable(ret,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites();ss++){
obj1 tmp;
mult(&tmp,&lhs,&rhs._odata[ss]);
@ -113,7 +113,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void mac(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
conformable(ret,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites();ss++){
obj1 tmp;
mac(&tmp,&lhs,&rhs._odata[ss]);
@ -124,7 +124,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void sub(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
conformable(ret,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites();ss++){
obj1 tmp;
sub(&tmp,&lhs,&rhs._odata[ss]);
@ -134,7 +134,7 @@ namespace Grid {
template<class obj1,class obj2,class obj3>
void add(Lattice<obj1> &ret,const obj2 &lhs,const Lattice<obj3> &rhs){
conformable(ret,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<rhs._grid->oSites();ss++){
obj1 tmp;
add(&tmp,&lhs,&rhs._odata[ss]);
@ -145,7 +145,7 @@ namespace Grid {
template<class sobj,class vobj>
inline void axpy(Lattice<vobj> &ret,sobj a,const Lattice<vobj> &lhs,const Lattice<vobj> &rhs){
conformable(lhs,rhs);
#pragma omp parallel for
PARALLEL_FOR_LOOP
for(int ss=0;ss<lhs._grid->oSites();ss++){
vobj tmp = a*lhs._odata[ss];
vstream(ret._odata[ss],tmp+rhs._odata[ss]);