portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-17 23:37:06 +01:00
Code Releases Activity

Dirichlet first cut - wrong answers on dagger multiply.

Browse Source 
Struggling to get a compute node so changing systems
This commit is contained in:
Peter Boyle
2022-02-22 19:58:33 +00:00
parent da06d15f73
commit aab3bcb46f
12 changed files with 523 additions and 440 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
Grid/communicator/Communicator_base.h
View File

@ -53,10 +53,11 @@ public:
// Communicator should know nothing of the physics grid, only processor grid.
////////////////////////////////////////////
int _Nprocessors; // How many in all
Coordinate _processors; // Which dimensions get relayed out over processors lanes.
int _processor; // linear processor rank
Coordinate _processor_coor; // linear processor coordinate
unsigned long _ndimension;
Coordinate _shm_processors; // Which dimensions get relayed out over processors lanes.
Coordinate _processors; // Which dimensions get relayed out over processors lanes.
Coordinate _processor_coor; // linear processor coordinate
static Grid_MPI_Comm communicator_world;
Grid_MPI_Comm communicator;
std::vector<Grid_MPI_Comm> communicator_halo;
@ -97,8 +98,9 @@ public:
int BossRank(void) ;
int ThisRank(void) ;
const Coordinate & ThisProcessorCoor(void) ;
const Coordinate & ShmGrid(void) { return _shm_processors; } ;
const Coordinate & ProcessorGrid(void) ;
int ProcessorCount(void) ;
int ProcessorCount(void) ;
////////////////////////////////////////////////////////////////////////////////
// very VERY rarely (Log, serial RNG) we need world without a grid
@ -142,16 +144,16 @@ public:
int bytes);
double StencilSendToRecvFrom(void *xmit,
int xmit_to_rank,
int xmit_to_rank,int do_xmit,
void *recv,
int recv_from_rank,
int recv_from_rank,int do_recv,
int bytes,int dir);
double StencilSendToRecvFromBegin(std::vector<CommsRequest_t> &list,
void *xmit,
int xmit_to_rank,
int xmit_to_rank,int do_xmit,
void *recv,
int recv_from_rank,
int recv_from_rank,int do_recv,
int bytes,int dir);

58
Grid/communicator/Communicator_mpi3.cc
View File

@ -106,7 +106,7 @@ CartesianCommunicator::CartesianCommunicator(const Coordinate &processors)
// Remap using the shared memory optimising routine
// The remap creates a comm which must be freed
////////////////////////////////////////////////////
GlobalSharedMemory::OptimalCommunicator (processors,optimal_comm);
GlobalSharedMemory::OptimalCommunicator (processors,optimal_comm,_shm_processors);
InitFromMPICommunicator(processors,optimal_comm);
SetCommunicator(optimal_comm);
///////////////////////////////////////////////////
@ -124,12 +124,13 @@ CartesianCommunicator::CartesianCommunicator(const Coordinate &processors,const
int parent_ndimension = parent._ndimension; assert(_ndimension >= parent._ndimension);
Coordinate parent_processor_coor(_ndimension,0);
Coordinate parent_processors (_ndimension,1);
Coordinate shm_processors (_ndimension,1);
// Can make 5d grid from 4d etc...
int pad = _ndimension-parent_ndimension;
for(int d=0;d<parent_ndimension;d++){
parent_processor_coor[pad+d]=parent._processor_coor[d];
parent_processors [pad+d]=parent._processors[d];
shm_processors [pad+d]=parent._shm_processors[d];
}
//////////////////////////////////////////////////////////////////////////////////////////////////////
@ -154,6 +155,7 @@ CartesianCommunicator::CartesianCommunicator(const Coordinate &processors,const
ccoor[d] = parent_processor_coor[d] % processors[d];
scoor[d] = parent_processor_coor[d] / processors[d];
ssize[d] = parent_processors[d] / processors[d];
if ( processors[d] < shm_processors[d] ) shm_processors[d] = processors[d]; // subnode splitting.
}
// rank within subcomm ; srank is rank of subcomm within blocks of subcomms
@ -335,22 +337,22 @@ void CartesianCommunicator::SendToRecvFrom(void *xmit,
}
// Basic Halo comms primitive
double CartesianCommunicator::StencilSendToRecvFrom( void *xmit,
int dest,
int dest, int dox,
void *recv,
int from,
int from, int dor,
int bytes,int dir)
{
std::vector<CommsRequest_t> list;
double offbytes = StencilSendToRecvFromBegin(list,xmit,dest,recv,from,bytes,dir);
double offbytes = StencilSendToRecvFromBegin(list,xmit,dest,dox,recv,from,dor,bytes,dir);
StencilSendToRecvFromComplete(list,dir);
return offbytes;
}
double CartesianCommunicator::StencilSendToRecvFromBegin(std::vector<CommsRequest_t> &list,
void *xmit,
int dest,
int dest,int dox,
void *recv,
int from,
int from,int dor,
int bytes,int dir)
{
int ncomm =communicator_halo.size();
@ -370,28 +372,32 @@ double CartesianCommunicator::StencilSendToRecvFromBegin(std::vector<CommsReques
double off_node_bytes=0.0;
int tag;
if ( (gfrom ==MPI_UNDEFINED) || Stencil_force_mpi ) {
tag= dir+from*32;
ierr=MPI_Irecv(recv, bytes, MPI_CHAR,from,tag,communicator_halo[commdir],&rrq);
assert(ierr==0);
list.push_back(rrq);
off_node_bytes+=bytes;
if ( dox ) {
if ( (gfrom ==MPI_UNDEFINED) || Stencil_force_mpi ) {
tag= dir+from*32;
ierr=MPI_Irecv(recv, bytes, MPI_CHAR,from,tag,communicator_halo[commdir],&rrq);
assert(ierr==0);
list.push_back(rrq);
off_node_bytes+=bytes;
}
}
if ( (gdest == MPI_UNDEFINED) || Stencil_force_mpi ) {
tag= dir+_processor*32;
ierr =MPI_Isend(xmit, bytes, MPI_CHAR,dest,tag,communicator_halo[commdir],&xrq);
assert(ierr==0);
list.push_back(xrq);
off_node_bytes+=bytes;
} else {
if (dor) {
if ( (gdest == MPI_UNDEFINED) || Stencil_force_mpi ) {
tag= dir+_processor*32;
ierr =MPI_Isend(xmit, bytes, MPI_CHAR,dest,tag,communicator_halo[commdir],&xrq);
assert(ierr==0);
list.push_back(xrq);
off_node_bytes+=bytes;
} else {
// TODO : make a OMP loop on CPU, call threaded bcopy
void *shm = (void *) this->ShmBufferTranslate(dest,recv);
assert(shm!=NULL);
// std::cout <<"acceleratorCopyDeviceToDeviceAsynch"<< std::endl;
acceleratorCopyDeviceToDeviceAsynch(xmit,shm,bytes);
void *shm = (void *) this->ShmBufferTranslate(dest,recv);
assert(shm!=NULL);
// std::cout <<"acceleratorCopyDeviceToDeviceAsynch"<< std::endl;
acceleratorCopyDeviceToDeviceAsynch(xmit,shm,bytes);
}
}
if ( CommunicatorPolicy == CommunicatorPolicySequential ) {
this->StencilSendToRecvFromComplete(list,dir);
}

10
Grid/communicator/Communicator_none.cc
View File

@ -45,12 +45,14 @@ void CartesianCommunicator::Init(int *argc, char *** arv)
CartesianCommunicator::CartesianCommunicator(const Coordinate &processors,const CartesianCommunicator &parent,int &srank)
: CartesianCommunicator(processors)
{
_shm_processors = Coordinate(processors.size(),1);
srank=0;
SetCommunicator(communicator_world);
}
CartesianCommunicator::CartesianCommunicator(const Coordinate &processors)
{
_shm_processors = Coordinate(processors.size(),1);
_processors = processors;
_ndimension = processors.size(); assert(_ndimension>=1);
_processor_coor.resize(_ndimension);
@ -111,18 +113,18 @@ void CartesianCommunicator::ShiftedRanks(int dim,int shift,int &source,int &dest
}
double CartesianCommunicator::StencilSendToRecvFrom( void *xmit,
int xmit_to_rank,
int xmit_to_rank,int dox,
void *recv,
int recv_from_rank,
int recv_from_rank,int dor,
int bytes, int dir)
{
return 2.0*bytes;
}
double CartesianCommunicator::StencilSendToRecvFromBegin(std::vector<CommsRequest_t> &list,
void *xmit,
int xmit_to_rank,
int xmit_to_rank,int dox,
void *recv,
int recv_from_rank,
int recv_from_rank,int dor,
int bytes, int dir)
{
return 2.0*bytes;

7
Grid/communicator/SharedMemory.h
View File

@ -93,9 +93,10 @@ public:
// Create an optimal reordered communicator that makes MPI_Cart_create get it right
//////////////////////////////////////////////////////////////////////////////////////
static void Init(Grid_MPI_Comm comm); // Typically MPI_COMM_WORLD
static void OptimalCommunicator (const Coordinate &processors,Grid_MPI_Comm & optimal_comm); // Turns MPI_COMM_WORLD into right layout for Cartesian
static void OptimalCommunicatorHypercube (const Coordinate &processors,Grid_MPI_Comm & optimal_comm); // Turns MPI_COMM_WORLD into right layout for Cartesian
static void OptimalCommunicatorSharedMemory(const Coordinate &processors,Grid_MPI_Comm & optimal_comm); // Turns MPI_COMM_WORLD into right layout for Cartesian
// Turns MPI_COMM_WORLD into right layout for Cartesian
static void OptimalCommunicator (const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &ShmDims);
static void OptimalCommunicatorHypercube (const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &ShmDims);
static void OptimalCommunicatorSharedMemory(const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &ShmDims);
static void GetShmDims(const Coordinate &WorldDims,Coordinate &ShmDims);
///////////////////////////////////////////////////
// Provide shared memory facilities off comm world

15
Grid/communicator/SharedMemoryMPI.cc
View File

@ -152,7 +152,7 @@ int Log2Size(int TwoToPower,int MAXLOG2)
}
return log2size;
}
void GlobalSharedMemory::OptimalCommunicator(const Coordinate &processors,Grid_MPI_Comm & optimal_comm)
void GlobalSharedMemory::OptimalCommunicator(const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &SHM)
{
//////////////////////////////////////////////////////////////////////////////
// Look and see if it looks like an HPE 8600 based on hostname conventions
@ -165,8 +165,8 @@ void GlobalSharedMemory::OptimalCommunicator(const Coordinate &processors,Grid_M
gethostname(name,namelen);
int nscan = sscanf(name,"r%di%dn%d",&R,&I,&N) ;
if(nscan==3 && HPEhypercube ) OptimalCommunicatorHypercube(processors,optimal_comm);
else OptimalCommunicatorSharedMemory(processors,optimal_comm);
if(nscan==3 && HPEhypercube ) OptimalCommunicatorHypercube(processors,optimal_comm,SHM);
else OptimalCommunicatorSharedMemory(processors,optimal_comm,SHM);
}
static inline int divides(int a,int b)
{
@ -221,7 +221,7 @@ void GlobalSharedMemory::GetShmDims(const Coordinate &WorldDims,Coordinate &ShmD
dim=(dim+1) %ndimension;
}
}
void GlobalSharedMemory::OptimalCommunicatorHypercube(const Coordinate &processors,Grid_MPI_Comm & optimal_comm)
void GlobalSharedMemory::OptimalCommunicatorHypercube(const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &SHM)
{
////////////////////////////////////////////////////////////////
// Assert power of two shm_size.
@ -294,7 +294,8 @@ void GlobalSharedMemory::OptimalCommunicatorHypercube(const Coordinate &processo
Coordinate HyperCoor(ndimension);
GetShmDims(WorldDims,ShmDims);
SHM = ShmDims;
////////////////////////////////////////////////////////////////
// Establish torus of processes and nodes with sub-blockings
////////////////////////////////////////////////////////////////
@ -341,7 +342,7 @@ void GlobalSharedMemory::OptimalCommunicatorHypercube(const Coordinate &processo
int ierr= MPI_Comm_split(WorldComm,0,rank,&optimal_comm);
assert(ierr==0);
}
void GlobalSharedMemory::OptimalCommunicatorSharedMemory(const Coordinate &processors,Grid_MPI_Comm & optimal_comm)
void GlobalSharedMemory::OptimalCommunicatorSharedMemory(const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &SHM)
{
////////////////////////////////////////////////////////////////
// Identify subblock of ranks on node spreading across dims
@ -353,6 +354,8 @@ void GlobalSharedMemory::OptimalCommunicatorSharedMemory(const Coordinate &proce
Coordinate ShmCoor(ndimension); Coordinate NodeCoor(ndimension); Coordinate WorldCoor(ndimension);
GetShmDims(WorldDims,ShmDims);
SHM=ShmDims;
////////////////////////////////////////////////////////////////
// Establish torus of processes and nodes with sub-blockings
////////////////////////////////////////////////////////////////

3
Grid/communicator/SharedMemoryNone.cc
View File

@ -48,9 +48,10 @@ void GlobalSharedMemory::Init(Grid_MPI_Comm comm)
_ShmSetup=1;
}
void GlobalSharedMemory::OptimalCommunicator(const Coordinate &processors,Grid_MPI_Comm & optimal_comm)
void GlobalSharedMemory::OptimalCommunicator(const Coordinate &processors,Grid_MPI_Comm & optimal_comm,Coordinate &SHM)
{
optimal_comm = WorldComm;
SHM = Coordinate(processors.size(),1);
}
////////////////////////////////////////////////////////////////////////////////////////////

7
Grid/qcd/action/fermion/WilsonFermion5D.h
View File

@ -173,7 +173,12 @@ public:
GridCartesian &FourDimGrid,
GridRedBlackCartesian &FourDimRedBlackGrid,
double _M5,const ImplParams &p= ImplParams());
void DirichletBlock(std::vector<int> & block){
Stencil.DirichletBlock(block);
StencilEven.DirichletBlock(block);
StencilOdd.DirichletBlock(block);
}
// Constructors
/*
WilsonFermion5D(int simd,

273
Grid/stencil/Stencil.h
View File

@ -131,8 +131,11 @@ class CartesianStencilAccelerator {
int _checkerboard;
int _npoints; // Move to template param?
int _osites;
int _dirichlet;
StencilVector _directions;
StencilVector _distances;
StencilVector _comms_send;
StencilVector _comms_recv;
StencilVector _comm_buf_size;
StencilVector _permute_type;
StencilVector same_node;
@ -226,6 +229,8 @@ public:
void * recv_buf;
Integer to_rank;
Integer from_rank;
Integer do_send;
Integer do_recv;
Integer bytes;
};
struct Merge {
@ -255,7 +260,6 @@ public:
void *recv_buf;
};
protected:
GridBase * _grid;
@ -299,29 +303,6 @@ public:
int u_comm_offset;
int _unified_buffer_size;
/////////////////////////////////////////
// Timing info; ugly; possibly temporary
/////////////////////////////////////////
double commtime;
double mpi3synctime;
double mpi3synctime_g;
double shmmergetime;
double gathertime;
double gathermtime;
double halogtime;
double mergetime;
double decompresstime;
double comms_bytes;
double shm_bytes;
double splicetime;
double nosplicetime;
double calls;
std::vector<double> comm_bytes_thr;
std::vector<double> shm_bytes_thr;
std::vector<double> comm_time_thr;
std::vector<double> comm_enter_thr;
std::vector<double> comm_leave_thr;
////////////////////////////////////////
// Stencil query
////////////////////////////////////////
@ -348,11 +329,12 @@ public:
//////////////////////////////////////////
// Comms packet queue for asynch thread
// Use OpenMP Tasks for cleaner ???
// must be called *inside* parallel region
//////////////////////////////////////////
/*
void CommunicateThreaded()
{
#ifdef GRID_OMP
// must be called in parallel region
int mythread = omp_get_thread_num();
int nthreads = CartesianCommunicator::nCommThreads;
#else
@ -361,65 +343,29 @@ public:
#endif
if (nthreads == -1) nthreads = 1;
if (mythread < nthreads) {
comm_enter_thr[mythread] = usecond();
for (int i = mythread; i < Packets.size(); i += nthreads) {
uint64_t bytes = _grid->StencilSendToRecvFrom(Packets[i].send_buf,
Packets[i].to_rank,
Packets[i].recv_buf,
Packets[i].from_rank,
Packets[i].bytes,i);
comm_bytes_thr[mythread] += bytes;
shm_bytes_thr[mythread] += 2*Packets[i].bytes-bytes; // Send + Recv.
}
comm_leave_thr[mythread]= usecond();
comm_time_thr[mythread] += comm_leave_thr[mythread] - comm_enter_thr[mythread];
}
}
void CollateThreads(void)
{
int nthreads = CartesianCommunicator::nCommThreads;
double first=0.0;
double last =0.0;
for(int t=0;t<nthreads;t++) {
double t0 = comm_enter_thr[t];
double t1 = comm_leave_thr[t];
comms_bytes+=comm_bytes_thr[t];
shm_bytes +=shm_bytes_thr[t];
comm_enter_thr[t] = 0.0;
comm_leave_thr[t] = 0.0;
comm_time_thr[t] = 0.0;
comm_bytes_thr[t]=0;
shm_bytes_thr[t]=0;
if ( first == 0.0 ) first = t0; // first is t0
if ( (t0 > 0.0) && ( t0 < first ) ) first = t0; // min time seen
if ( t1 > last ) last = t1; // max time seen
}
commtime+= last-first;
}
*/
////////////////////////////////////////////////////////////////////////
// Non blocking send and receive. Necessarily parallel.
////////////////////////////////////////////////////////////////////////
void CommunicateBegin(std::vector<std::vector<CommsRequest_t> > &reqs)
{
reqs.resize(Packets.size());
commtime-=usecond();
for(int i=0;i<Packets.size();i++){
uint64_t bytes=_grid->StencilSendToRecvFromBegin(reqs[i],
Packets[i].send_buf,
Packets[i].to_rank,
Packets[i].recv_buf,
Packets[i].from_rank,
Packets[i].bytes,i);
comms_bytes+=bytes;
shm_bytes +=2*Packets[i].bytes-bytes;
_grid->StencilSendToRecvFromBegin(reqs[i],
Packets[i].send_buf,
Packets[i].to_rank,Packets[i].do_send,
Packets[i].recv_buf,
Packets[i].from_rank,Packets[i].do_recv,
Packets[i].bytes,i);
}
}
@ -428,7 +374,6 @@ public:
for(int i=0;i<Packets.size();i++){
_grid->StencilSendToRecvFromComplete(reqs[i],i);
}
commtime+=usecond();
}
////////////////////////////////////////////////////////////////////////
// Blocking send and receive. Either sequential or parallel.
@ -436,28 +381,27 @@ public:
void Communicate(void)
{
if ( CartesianCommunicator::CommunicatorPolicy == CartesianCommunicator::CommunicatorPolicySequential ){
thread_region {
// must be called in parallel region
int mythread = thread_num();
int maxthreads= thread_max();
int nthreads = CartesianCommunicator::nCommThreads;
assert(nthreads <= maxthreads);
if (nthreads == -1) nthreads = 1;
if (mythread < nthreads) {
for (int i = mythread; i < Packets.size(); i += nthreads) {
double start = usecond();
uint64_t bytes= _grid->StencilSendToRecvFrom(Packets[i].send_buf,
Packets[i].to_rank,
Packets[i].recv_buf,
Packets[i].from_rank,
Packets[i].bytes,i);
comm_bytes_thr[mythread] += bytes;
shm_bytes_thr[mythread] += Packets[i].bytes - bytes;
comm_time_thr[mythread] += usecond() - start;
}
}
}
} else { // Concurrent and non-threaded asynch calls to MPI
/////////////////////////////////////////////////////////
// several way threaded on different communicators.
// Cannot combine with Dirichlet operators
// This scheme is needed on Intel Omnipath for best performance
// Deprecate once there are very few omnipath clusters
/////////////////////////////////////////////////////////
int nthreads = CartesianCommunicator::nCommThreads;
int old = GridThread::GetThreads();
GridThread::SetThreads(nthreads);
thread_for(i,Packets.size(),{
_grid->StencilSendToRecvFrom(Packets[i].send_buf,
Packets[i].to_rank,Packets[i].do_send,
Packets[i].recv_buf,
Packets[i].from_rank,Packets[i].do_recv,
Packets[i].bytes,i);
});
GridThread::SetThreads(old);
} else {
/////////////////////////////////////////////////////////
// Concurrent and non-threaded asynch calls to MPI
/////////////////////////////////////////////////////////
std::vector<std::vector<CommsRequest_t> > reqs;
this->CommunicateBegin(reqs);
this->CommunicateComplete(reqs);
@ -499,31 +443,23 @@ public:
sshift[1] = _grid->CheckerBoardShiftForCB(this->_checkerboard,dimension,shift,Odd);
if ( sshift[0] == sshift[1] ) {
if (splice_dim) {
splicetime-=usecond();
auto tmp = GatherSimd(source,dimension,shift,0x3,compress,face_idx);
auto tmp = GatherSimd(source,dimension,shift,0x3,compress,face_idx,point);
is_same_node = is_same_node && tmp;
splicetime+=usecond();
} else {
nosplicetime-=usecond();
auto tmp = Gather(source,dimension,shift,0x3,compress,face_idx);
auto tmp = Gather(source,dimension,shift,0x3,compress,face_idx,point);
is_same_node = is_same_node && tmp;
nosplicetime+=usecond();
}
} else {
if(splice_dim){
splicetime-=usecond();
// if checkerboard is unfavourable take two passes
// both with block stride loop iteration
auto tmp1 = GatherSimd(source,dimension,shift,0x1,compress,face_idx);
auto tmp2 = GatherSimd(source,dimension,shift,0x2,compress,face_idx);
auto tmp1 = GatherSimd(source,dimension,shift,0x1,compress,face_idx,point);
auto tmp2 = GatherSimd(source,dimension,shift,0x2,compress,face_idx,point);
is_same_node = is_same_node && tmp1 && tmp2;
splicetime+=usecond();
} else {
nosplicetime-=usecond();
auto tmp1 = Gather(source,dimension,shift,0x1,compress,face_idx);
auto tmp2 = Gather(source,dimension,shift,0x2,compress,face_idx);
auto tmp1 = Gather(source,dimension,shift,0x1,compress,face_idx,point);
auto tmp2 = Gather(source,dimension,shift,0x2,compress,face_idx,point);
is_same_node = is_same_node && tmp1 && tmp2;
nosplicetime+=usecond();
}
}
}
@ -533,13 +469,10 @@ public:
template<class compressor>
void HaloGather(const Lattice<vobj> &source,compressor &compress)
{
mpi3synctime_g-=usecond();
_grid->StencilBarrier();// Synch shared memory on a single nodes
mpi3synctime_g+=usecond();
// conformable(source.Grid(),_grid);
assert(source.Grid()==_grid);
halogtime-=usecond();
u_comm_offset=0;
@ -553,7 +486,6 @@ public:
assert(u_comm_offset==_unified_buffer_size);
accelerator_barrier();
halogtime+=usecond();
}
/////////////////////////
@ -568,7 +500,6 @@ public:
Packets.resize(0);
CopyReceiveBuffers.resize(0);
CachedTransfers.resize(0);
calls++;
}
void AddCopy(void *from,void * to, Integer bytes)
{
@ -622,12 +553,17 @@ public:
CachedTransfers.push_back(obj);
return 0;
}
void AddPacket(void *xmit,void * rcv, Integer to,Integer from,Integer bytes){
void AddPacket(void *xmit,void * rcv,
Integer to, Integer do_send,
Integer from, Integer do_recv,
Integer bytes){
Packet p;
p.send_buf = xmit;
p.recv_buf = rcv;
p.to_rank = to;
p.from_rank= from;
p.do_send = do_send;
p.do_recv = do_recv;
p.bytes = bytes;
Packets.push_back(p);
}
@ -651,19 +587,13 @@ public:
CommsMerge(decompress,Mergers,Decompressions);
}
template<class decompressor> void CommsMergeSHM(decompressor decompress) {
mpi3synctime-=usecond();
_grid->StencilBarrier();// Synch shared memory on a single nodes
mpi3synctime+=usecond();
shmmergetime-=usecond();
CommsMerge(decompress,MergersSHM,DecompressionsSHM);
shmmergetime+=usecond();
}
template<class decompressor>
void CommsMerge(decompressor decompress,std::vector<Merge> &mm,std::vector<Decompress> &dd)
{
mergetime-=usecond();
for(int i=0;i<mm.size();i++){
auto mp = &mm[i].mpointer[0];
auto vp0= &mm[i].vpointers[0][0];
@ -673,9 +603,7 @@ public:
decompress.Exchange(mp,vp0,vp1,type,o);
});
}
mergetime+=usecond();
decompresstime-=usecond();
for(int i=0;i<dd.size();i++){
auto kp = dd[i].kernel_p;
auto mp = dd[i].mpi_p;
@ -683,7 +611,6 @@ public:
decompress.Decompress(kp,mp,o);
});
}
decompresstime+=usecond();
}
////////////////////////////////////////
// Set up routines
@ -720,19 +647,58 @@ public:
}
}
}
/// Introduce a block structure and switch off comms on boundaries
void DirichletBlock(const std::vector<int> &dirichlet_block)
{
this->_dirichlet = 1;
for(int ii=0;ii<this->_npoints;ii++){
int dimension = this->_directions[ii];
int displacement = this->_distances[ii];
int shift = displacement;
int gd = _grid->_gdimensions[dimension];
int fd = _grid->_fdimensions[dimension];
int pd = _grid->_processors [dimension];
int ld = gd/pd;
int pc = _grid->_processor_coor[dimension];
///////////////////////////////////////////
// Figure out dirichlet send and receive
// on this leg of stencil.
///////////////////////////////////////////
int comm_dim = _grid->_processors[dimension] >1 ;
int block = dirichlet_block[dimension];
this->_comms_send[ii] = comm_dim;
this->_comms_recv[ii] = comm_dim;
if ( block ) {
assert(abs(displacement) < ld );
if( displacement > 0 ) {
// High side, low side
// | <--B--->|
// | | |
// noR
// noS
if ( (ld*(pc+1) ) % block == 0 ) this->_comms_recv[ii] = 0;
if ( ( ld*pc ) % block == 0 ) this->_comms_send[ii] = 0;
} else {
// High side, low side
// | <--B--->|
// | | |
// noS
// noR
if ( (ld*(pc+1) ) % block == 0 ) this->_comms_send[ii] = 0;
if ( ( ld*pc ) % block == 0 ) this->_comms_recv[ii] = 0;
}
}
}
}
CartesianStencil(GridBase *grid,
int npoints,
int checkerboard,
const std::vector<int> &directions,
const std::vector<int> &distances,
Parameters p)
: shm_bytes_thr(npoints),
comm_bytes_thr(npoints),
comm_enter_thr(npoints),
comm_leave_thr(npoints),
comm_time_thr(npoints)
{
this->_dirichlet = 0;
face_table_computed=0;
_grid = grid;
this->parameters=p;
@ -745,6 +711,8 @@ public:
this->_simd_layout = _grid->_simd_layout; // copy simd_layout to give access to Accelerator Kernels
this->_directions = StencilVector(directions);
this->_distances = StencilVector(distances);
this->_comms_send.resize(npoints);
this->_comms_recv.resize(npoints);
this->same_node.resize(npoints);
_unified_buffer_size=0;
@ -763,24 +731,27 @@ public:
int displacement = distances[i];
int shift = displacement;
int gd = _grid->_gdimensions[dimension];
int fd = _grid->_fdimensions[dimension];
int pd = _grid->_processors [dimension];
int ld = gd/pd;
int rd = _grid->_rdimensions[dimension];
int pc = _grid->_processor_coor[dimension];
this->_permute_type[point]=_grid->PermuteType(dimension);
this->_checkerboard = checkerboard;
//////////////////////////
// the permute type
//////////////////////////
int simd_layout = _grid->_simd_layout[dimension];
int comm_dim = _grid->_processors[dimension] >1 ;
int splice_dim = _grid->_simd_layout[dimension]>1 && (comm_dim);
int rotate_dim = _grid->_simd_layout[dimension]>2;
this->_comms_send[ii] = comm_dim;
this->_comms_recv[ii] = comm_dim;
assert ( (rotate_dim && comm_dim) == false) ; // Do not think spread out is supported
int sshift[2];
//////////////////////////
// Underlying approach. For each local site build
// up a table containing the npoint "neighbours" and whether they
@ -881,6 +852,7 @@ public:
GridBase *grid=_grid;
const int Nsimd = grid->Nsimd();
int comms_recv = this->_comms_recv[point];
int fd = _grid->_fdimensions[dimension];
int ld = _grid->_ldimensions[dimension];
int rd = _grid->_rdimensions[dimension];
@ -937,7 +909,9 @@ public:
if ( (shiftpm== 1) && (sx<x) && (grid->_processor_coor[dimension]==grid->_processors[dimension]-1) ) {
wraparound = 1;
}
if (!offnode) {
// Wrap locally dirichlet support case OR node local
if ( (offnode==0) || (comms_recv==0) ) {
int permute_slice=0;
CopyPlane(point,dimension,x,sx,cbmask,permute_slice,wraparound);
@ -1054,11 +1028,14 @@ public:
}
template<class compressor>
int Gather(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,compressor & compress,int &face_idx)
int Gather(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,compressor & compress,int &face_idx, int point)
{
typedef typename cobj::vector_type vector_type;
typedef typename cobj::scalar_type scalar_type;
int comms_send = this->_comms_send[point] ;
int comms_recv = this->_comms_recv[point] ;
assert(rhs.Grid()==_grid);
// conformable(_grid,rhs.Grid());
@ -1124,10 +1101,10 @@ public:
////////////////////////////////////////////////////////
// Gather locally
////////////////////////////////////////////////////////
gathertime-=usecond();
assert(send_buf!=NULL);
Gather_plane_simple_table(face_table[face_idx],rhs,send_buf,compress,u_comm_offset,so); face_idx++;
gathertime+=usecond();
if ( comms_send )
Gather_plane_simple_table(face_table[face_idx],rhs,send_buf,compress,u_comm_offset,so);
face_idx++;
int duplicate = CheckForDuplicate(dimension,sx,comm_proc,(void *)&recv_buf[u_comm_offset],0,bytes,cbmask);
if ( (!duplicate) ) { // Force comms for now
@ -1138,12 +1115,12 @@ public:
///////////////////////////////////////////////////////////
AddPacket((void *)&send_buf[u_comm_offset],
(void *)&recv_buf[u_comm_offset],
xmit_to_rank,
recv_from_rank,
xmit_to_rank, comms_send,
recv_from_rank, comms_recv,
bytes);
}
if ( compress.DecompressionStep() ) {
if ( compress.DecompressionStep() && comms_recv ) {
AddDecompress(&this->u_recv_buf_p[u_comm_offset],
&recv_buf[u_comm_offset],
words,Decompressions);
@ -1155,11 +1132,15 @@ public:
}
template<class compressor>
int GatherSimd(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,compressor &compress,int & face_idx)
int GatherSimd(const Lattice<vobj> &rhs,int dimension,int shift,int cbmask,compressor &compress,int & face_idx,int point)
{
const int Nsimd = _grid->Nsimd();
const int maxl =2;// max layout in a direction
int comms_send = this->_comms_send[point] ;
int comms_recv = this->_comms_recv[point] ;
int fd = _grid->_fdimensions[dimension];
int rd = _grid->_rdimensions[dimension];
int ld = _grid->_ldimensions[dimension];
@ -1224,12 +1205,11 @@ public:
&face_table[face_idx][0],
face_table[face_idx].size()*sizeof(face_table_host[0]));
}
gathermtime-=usecond();
Gather_plane_exchange_table(face_table[face_idx],rhs,spointers,dimension,sx,cbmask,compress,permute_type);
if ( comms_send )
Gather_plane_exchange_table(face_table[face_idx],rhs,spointers,dimension,sx,cbmask,compress,permute_type);
face_idx++;
gathermtime+=usecond();
//spointers[0] -- low
//spointers[1] -- high
@ -1260,7 +1240,10 @@ public:
int duplicate = CheckForDuplicate(dimension,sx,nbr_proc,(void *)rp,i,bytes,cbmask);
if ( (!duplicate) ) { // Force comms for now
AddPacket((void *)sp,(void *)rp,xmit_to_rank,recv_from_rank,bytes);
AddPacket((void *)sp,(void *)rp,
xmit_to_rank,comms_send,
recv_from_rank,comms_recv,
bytes);
}
} else {
@ -1270,7 +1253,9 @@ public:
}
}
AddMerge(&this->u_recv_buf_p[u_comm_offset],rpointers,reduced_buffer_size,permute_type,Mergers);
if ( comms_recv ) {
AddMerge(&this->u_recv_buf_p[u_comm_offset],rpointers,reduced_buffer_size,permute_type,Mergers);
}
u_comm_offset +=buffer_size;
}