void loadLinkToGPU(cudaGaugeField* cudaGauge, cpuGaugeField* cpuGauge, QudaGaugeParam* param)
{
if (cudaGauge->Precision() != cpuGauge->Precision()){
errorQuda("Mismatch between CPU precision and CUDA precision");
}
QudaPrecision prec = cudaGauge->Precision();
#ifdef MULTI_GPU
const int* Z = cudaGauge->X();
#endif
int pad = cudaGauge->Pad();
int Vsh_x = param->X[1]*param->X[2]*param->X[3]/2;
int Vsh_y = param->X[0]*param->X[2]*param->X[3]/2;
int Vsh_z = param->X[0]*param->X[1]*param->X[3]/2;
int Vsh_t = param->X[0]*param->X[1]*param->X[2]/2;
static void* ghost_cpuGauge[4];
static void* ghost_cpuGauge_diag[16];
#ifdef MULTI_GPU
static int allocated = 0;
int Vs[4] = {2*Vsh_x, 2*Vsh_y, 2*Vsh_z, 2*Vsh_t};
if (!allocated) {
for(int i=0;i < 4; i++) {
size_t ghost_bytes = 8*Vs[i]*gaugeSiteSize*prec;
#ifdef GPU_DIRECT
ghost_cpuGauge[i] = pinned_malloc(ghost_bytes);
#else
ghost_cpuGauge[i] = safe_malloc(ghost_bytes);
#endif
}
/*
* nu | |
* |_____|
* mu
*/
for(int nu=0;nu < 4;nu++){
for(int mu=0; mu < 4;mu++){
if(nu == mu){
ghost_cpuGauge_diag[nu*4+mu] = NULL;
}else{
//the other directions
int dir1, dir2;
for(dir1= 0; dir1 < 4; dir1++){
if(dir1 !=nu && dir1 != mu){
break;
}
}
for(dir2=0; dir2 < 4; dir2++){
if(dir2 != nu && dir2 != mu && dir2 != dir1){
break;
}
}
//int rc = posix_memalign((void**)&ghost_cpuGauge_diag[nu*4+mu], ALIGNMENT, Z[dir1]*Z[dir2]*gaugeSiteSize*prec);
size_t nbytes = Z[dir1]*Z[dir2]*gaugeSiteSize*prec;
#ifdef GPU_DIRECT
ghost_cpuGauge_diag[nu*4+mu] = pinned_malloc(nbytes);
#else
ghost_cpuGauge_diag[nu*4+mu] = safe_malloc(nbytes);
#endif
memset(ghost_cpuGauge_diag[nu*4+mu], 0, nbytes);
}
}
}
allocated = 1;
}
int optflag=1;
// driver for for packalllink
exchange_cpu_sitelink(param->X, (void**)cpuGauge->Gauge_p(), ghost_cpuGauge, ghost_cpuGauge_diag, prec, param, optflag);
#endif
do_loadLinkToGPU(param->X, cudaGauge->Even_p(), cudaGauge->Odd_p(), (void**)cpuGauge->Gauge_p(),
ghost_cpuGauge, ghost_cpuGauge_diag,
cudaGauge->Reconstruct(), cudaGauge->Bytes(), cudaGauge->VolumeCB(), pad,
Vsh_x, Vsh_y, Vsh_z, Vsh_t,
prec, cpuGauge->Order());
#ifdef MULTI_GPU
if(!(param->preserve_gauge & QUDA_FAT_PRESERVE_COMM_MEM)) {
for(int i=0;i < 4;i++){
host_free(ghost_cpuGauge[i]);
}
for(int i=0;i <4; i++){
for(int j=0;j <4; j++){
if (i != j) host_free(ghost_cpuGauge_diag[i*4+j]);
}
}
allocated = 0;
}
#endif
}
static void
llfat_test(int test)
{
QudaGaugeParam qudaGaugeParam;
#ifdef MULTI_GPU
void* ghost_sitelink[4];
void* ghost_sitelink_diag[16];
#endif
initQuda(device);
cpu_prec = prec;
gSize = cpu_prec;
qudaGaugeParam = newQudaGaugeParam();
qudaGaugeParam.anisotropy = 1.0;
qudaGaugeParam.X[0] = xdim;
qudaGaugeParam.X[1] = ydim;
qudaGaugeParam.X[2] = zdim;
qudaGaugeParam.X[3] = tdim;
setDims(qudaGaugeParam.X);
qudaGaugeParam.cpu_prec = cpu_prec;
qudaGaugeParam.cuda_prec = prec;
qudaGaugeParam.gauge_order = gauge_order;
qudaGaugeParam.type=QUDA_WILSON_LINKS;
qudaGaugeParam.reconstruct = link_recon;
/*
qudaGaugeParam.flag = QUDA_FAT_PRESERVE_CPU_GAUGE
| QUDA_FAT_PRESERVE_GPU_GAUGE
| QUDA_FAT_PRESERVE_COMM_MEM;
*/
qudaGaugeParam.preserve_gauge =0;
void* fatlink;
if (cudaMallocHost((void**)&fatlink, 4*V*gaugeSiteSize*gSize) == cudaErrorMemoryAllocation) {
errorQuda("ERROR: cudaMallocHost failed for fatlink\n");
}
void* longlink;
if (cudaMallocHost((void**)&longlink, 4*V*gaugeSiteSize*gSize) == cudaErrorMemoryAllocation) {
errorQuda("ERROR: cudaMallocHost failed for longlink\n");
} // page-locked memory
void* sitelink[4];
for(int i=0;i < 4;i++){
if (cudaMallocHost((void**)&sitelink[i], V*gaugeSiteSize*gSize) == cudaErrorMemoryAllocation) {
errorQuda("ERROR: cudaMallocHost failed for sitelink\n");
}
}
void* sitelink_ex[4];
for(int i=0;i < 4;i++){
if (cudaMallocHost((void**)&sitelink_ex[i], V_ex*gaugeSiteSize*gSize) == cudaErrorMemoryAllocation) {
errorQuda("ERROR: cudaMallocHost failed for sitelink_ex\n");
}
}
void* milc_sitelink;
milc_sitelink = (void*)malloc(4*V*gaugeSiteSize*gSize);
if(milc_sitelink == NULL){
errorQuda("ERROR: allocating milc_sitelink failed\n");
}
void* milc_sitelink_ex;
milc_sitelink_ex = (void*)malloc(4*V_ex*gaugeSiteSize*gSize);
if(milc_sitelink_ex == NULL){
errorQuda("Error: allocating milc_sitelink failed\n");
}
createSiteLinkCPU(sitelink, qudaGaugeParam.cpu_prec, 1);
if(gauge_order == QUDA_MILC_GAUGE_ORDER){
for(int i=0; i<V; ++i){
for(int dir=0; dir<4; ++dir){
char* src = (char*)sitelink[dir];
memcpy((char*)milc_sitelink + (i*4 + dir)*gaugeSiteSize*gSize, src+i*gaugeSiteSize*gSize, gaugeSiteSize*gSize);
}
}
}
int X1=Z[0];
int X2=Z[1];
int X3=Z[2];
int X4=Z[3];
for(int i=0; i < V_ex; i++){
int sid = i;
int oddBit=0;
if(i >= Vh_ex){
sid = i - Vh_ex;
oddBit = 1;
}
int za = sid/E1h;
int x1h = sid - za*E1h;
int zb = za/E2;
int x2 = za - zb*E2;
int x4 = zb/E3;
int x3 = zb - x4*E3;
int x1odd = (x2 + x3 + x4 + oddBit) & 1;
int x1 = 2*x1h + x1odd;
if( x1< 2 || x1 >= X1 +2
|| x2< 2 || x2 >= X2 +2
|| x3< 2 || x3 >= X3 +2
|| x4< 2 || x4 >= X4 +2){
#ifdef MULTI_GPU
continue;
#endif
}
x1 = (x1 - 2 + X1) % X1;
x2 = (x2 - 2 + X2) % X2;
x3 = (x3 - 2 + X3) % X3;
x4 = (x4 - 2 + X4) % X4;
int idx = (x4*X3*X2*X1+x3*X2*X1+x2*X1+x1)>>1;
if(oddBit){
idx += Vh;
}
for(int dir= 0; dir < 4; dir++){
char* src = (char*)sitelink[dir];
char* dst = (char*)sitelink_ex[dir];
memcpy(dst+i*gaugeSiteSize*gSize, src+idx*gaugeSiteSize*gSize, gaugeSiteSize*gSize);
// milc ordering
memcpy((char*)milc_sitelink_ex + (i*4 + dir)*gaugeSiteSize*gSize, src+idx*gaugeSiteSize*gSize, gaugeSiteSize*gSize);
}//dir
}//i
double act_path_coeff[6];
for(int i=0;i < 6;i++){
act_path_coeff[i]= 0.1*i;
}
//only record the last call's performance
//the first one is for creating the cpu/cuda data structures
struct timeval t0, t1;
void** sitelink_ptr;
QudaComputeFatMethod method = (test) ? QUDA_COMPUTE_FAT_EXTENDED_VOLUME : QUDA_COMPUTE_FAT_STANDARD;
if(gauge_order == QUDA_QDP_GAUGE_ORDER){
sitelink_ptr = (test) ? (void**)sitelink_ex : (void**)sitelink;
}else{
sitelink_ptr = (test) ? (void**)milc_sitelink_ex : (void**)milc_sitelink;
}
void* longlink_ptr = longlink;
#ifdef MULTI_GPU
if(!test) longlink_ptr = NULL; // Have to have an extended volume for the long-link calculation
#endif
gettimeofday(&t0, NULL);
computeKSLinkQuda(fatlink, longlink_ptr, NULL, milc_sitelink, act_path_coeff, &qudaGaugeParam, method);
gettimeofday(&t1, NULL);
double secs = TDIFF(t0,t1);
void* fat_reflink[4];
void* long_reflink[4];
for(int i=0;i < 4;i++){
fat_reflink[i] = malloc(V*gaugeSiteSize*gSize);
if(fat_reflink[i] == NULL){
errorQuda("ERROR; allocate fat_reflink[%d] failed\n", i);
}
long_reflink[i] = malloc(V*gaugeSiteSize*gSize);
if(long_reflink[i] == NULL) errorQuda("ERROR; allocate long_reflink[%d] failed\n", i);
}
if (verify_results){
//FIXME: we have this compplication because references takes coeff as float/double
// depending on the precision while the GPU code aways take coeff as double
void* coeff;
double coeff_dp[6];
float coeff_sp[6];
for(int i=0;i < 6;i++){
coeff_sp[i] = coeff_dp[i] = act_path_coeff[i];
}
if(prec == QUDA_DOUBLE_PRECISION){
coeff = coeff_dp;
}else{
coeff = coeff_sp;
}
#ifdef MULTI_GPU
int optflag = 0;
//we need x,y,z site links in the back and forward T slice
// so it is 3*2*Vs_t
int Vs[4] = {Vs_x, Vs_y, Vs_z, Vs_t};
for(int i=0;i < 4; i++){
ghost_sitelink[i] = malloc(8*Vs[i]*gaugeSiteSize*gSize);
if (ghost_sitelink[i] == NULL){
printf("ERROR: malloc failed for ghost_sitelink[%d] \n",i);
exit(1);
}
}
/*
nu | |
|_____|
mu
*/
for(int nu=0;nu < 4;nu++){
for(int mu=0; mu < 4;mu++){
if(nu == mu){
ghost_sitelink_diag[nu*4+mu] = NULL;
}else{
//the other directions
int dir1, dir2;
for(dir1= 0; dir1 < 4; dir1++){
if(dir1 !=nu && dir1 != mu){
break;
}
}
for(dir2=0; dir2 < 4; dir2++){
if(dir2 != nu && dir2 != mu && dir2 != dir1){
break;
}
}
ghost_sitelink_diag[nu*4+mu] = malloc(Z[dir1]*Z[dir2]*gaugeSiteSize*gSize);
if(ghost_sitelink_diag[nu*4+mu] == NULL){
errorQuda("malloc failed for ghost_sitelink_diag\n");
}
memset(ghost_sitelink_diag[nu*4+mu], 0, Z[dir1]*Z[dir2]*gaugeSiteSize*gSize);
}
}
}
exchange_cpu_sitelink(qudaGaugeParam.X, sitelink, ghost_sitelink, ghost_sitelink_diag, qudaGaugeParam.cpu_prec, &qudaGaugeParam, optflag);
llfat_reference_mg(fat_reflink, sitelink, ghost_sitelink, ghost_sitelink_diag, qudaGaugeParam.cpu_prec, coeff);
{
int R[4] = {2,2,2,2};
exchange_cpu_sitelink_ex(qudaGaugeParam.X, R, sitelink_ex, QUDA_QDP_GAUGE_ORDER, qudaGaugeParam.cpu_prec, 0, 4);
computeLongLinkCPU(long_reflink, sitelink_ex, qudaGaugeParam.cpu_prec, coeff);
}
#else
llfat_reference(fat_reflink, sitelink, qudaGaugeParam.cpu_prec, coeff);
computeLongLinkCPU(long_reflink, sitelink, qudaGaugeParam.cpu_prec, coeff);
#endif
}//verify_results
//format change for fatlink and longlink
void* myfatlink[4];
void* mylonglink[4];
for(int i=0;i < 4;i++){
myfatlink[i] = malloc(V*gaugeSiteSize*gSize);
if(myfatlink[i] == NULL){
printf("Error: malloc failed for myfatlink[%d]\n", i);
exit(1);
}
mylonglink[i] = malloc(V*gaugeSiteSize*gSize);
if(mylonglink[i] == NULL){
printf("Error: malloc failed for mylonglink[%d]\n", i);
exit(1);
}
memset(myfatlink[i], 0, V*gaugeSiteSize*gSize);
memset(mylonglink[i], 0, V*gaugeSiteSize*gSize);
}
for(int i=0;i < V; i++){
for(int dir=0; dir< 4; dir++){
char* src = ((char*)fatlink)+ (4*i+dir)*gaugeSiteSize*gSize;
char* dst = ((char*)myfatlink[dir]) + i*gaugeSiteSize*gSize;
memcpy(dst, src, gaugeSiteSize*gSize);
src = ((char*)longlink)+ (4*i+dir)*gaugeSiteSize*gSize;
dst = ((char*)mylonglink[dir]) + i*gaugeSiteSize*gSize;
memcpy(dst, src, gaugeSiteSize*gSize);
}
}
if (verify_results) {
printfQuda("Checking fat links...\n");
int res=1;
for(int dir=0; dir<4; dir++){
res &= compare_floats(fat_reflink[dir], myfatlink[dir], V*gaugeSiteSize, 1e-3, qudaGaugeParam.cpu_prec);
}
strong_check_link(myfatlink, "GPU results: ",
fat_reflink, "CPU reference results:",
V, qudaGaugeParam.cpu_prec);
printfQuda("Fat-link test %s\n\n",(1 == res) ? "PASSED" : "FAILED");
#ifdef MULTI_GPU
if(test){
#endif
printfQuda("Checking long links...\n");
res = 1;
for(int dir=0; dir<4; ++dir){
res &= compare_floats(long_reflink[dir], mylonglink[dir], V*gaugeSiteSize, 1e-3, qudaGaugeParam.cpu_prec);
}
strong_check_link(mylonglink, "GPU results: ",
long_reflink, "CPU reference results:",
V, qudaGaugeParam.cpu_prec);
printfQuda("Long-link test %s\n\n",(1 == res) ? "PASSED" : "FAILED");
#ifdef MULTI_GPU
}else{ // !test
printfQuda("Extended volume is required for multi-GPU long-link construction\n");
}
#endif
}
int volume = qudaGaugeParam.X[0]*qudaGaugeParam.X[1]*qudaGaugeParam.X[2]*qudaGaugeParam.X[3];
int flops= 61632;
#ifdef MULTI_GPU
if(test) flops += (252*4); // long-link contribution
#else
flops += (252*4); // 2*117 + 18 (two matrix-matrix multiplications and a matrix rescale)
#endif
double perf = 1.0* flops*volume/(secs*1024*1024*1024);
printfQuda("link computation time =%.2f ms, flops= %.2f Gflops\n", secs*1000, perf);
for(int i=0;i < 4;i++){
free(myfatlink[i]);
}
#ifdef MULTI_GPU
if (verify_results){
int i;
for(i=0;i < 4;i++){
free(ghost_sitelink[i]);
}
for(i=0;i <4; i++){
for(int j=0;j <4; j++){
if (i==j){
continue;
}
free(ghost_sitelink_diag[i*4+j]);
}
}
}
#endif
for(int i=0;i < 4; i++){
cudaFreeHost(sitelink[i]);
cudaFreeHost(sitelink_ex[i]);
free(fat_reflink[i]);
}
cudaFreeHost(fatlink);
cudaFreeHost(longlink);
if(milc_sitelink) free(milc_sitelink);
if(milc_sitelink_ex) free(milc_sitelink_ex);
#ifdef MULTI_GPU
exchange_llfat_cleanup();
#endif
endQuda();
}