Open MP on host enabled

Grid/lattice/Lattice_reduction.h

@@ -28,6 +28,9 @@ Author: Christoph Lehner <christoph@lhnr.de>
 #if defined(GRID_CUDA)||defined(GRID_HIP)
 #include <Grid/lattice/Lattice_reduction_gpu.h>
 #endif
+#if defined(GRID_SYCL)
+#include <Grid/lattice/Lattice_reduction_sycl.h>
+#endif
 
 NAMESPACE_BEGIN(Grid);
 
@@ -127,7 +130,7 @@ inline Double max(const Double *arg, Integer osites)
 template<class vobj>
 inline typename vobj::scalar_object sum(const vobj *arg, Integer osites)
 {
-#if defined(GRID_CUDA)||defined(GRID_HIP)
+#if defined(GRID_CUDA)||defined(GRID_HIP)||defined(GRID_SYCL)
   return sum_gpu(arg,osites);
 #else
   return sum_cpu(arg,osites);
@@ -136,7 +139,7 @@ inline typename vobj::scalar_object sum(const vobj *arg, Integer osites)
 template<class vobj>
 inline typename vobj::scalar_objectD sumD(const vobj *arg, Integer osites)
 {
-#if defined(GRID_CUDA)||defined(GRID_HIP)
+#if defined(GRID_CUDA)||defined(GRID_HIP)||defined(GRID_SYCL)
   return sumD_gpu(arg,osites);
 #else
   return sumD_cpu(arg,osites);
@@ -145,7 +148,7 @@ inline typename vobj::scalar_objectD sumD(const vobj *arg, Integer osites)
 template<class vobj>
 inline typename vobj::scalar_objectD sumD_large(const vobj *arg, Integer osites)
 {
-#if defined(GRID_CUDA)||defined(GRID_HIP)
+#if defined(GRID_CUDA)||defined(GRID_HIP)||defined(GRID_SYCL)
   return sumD_gpu_large(arg,osites);
 #else
   return sumD_cpu(arg,osites);
@@ -155,13 +158,13 @@ inline typename vobj::scalar_objectD sumD_large(const vobj *arg, Integer osites)
 template<class vobj>
 inline typename vobj::scalar_object sum(const Lattice<vobj> &arg)
 {
-#if defined(GRID_CUDA)||defined(GRID_HIP)
-  autoView( arg_v, arg, AcceleratorRead);
   Integer osites = arg.Grid()->oSites();
-  auto ssum= sum_gpu(&arg_v[0],osites);
+#if defined(GRID_CUDA)||defined(GRID_HIP)||defined(GRID_SYCL)
+  typename vobj::scalar_object ssum;
+  autoView( arg_v, arg, AcceleratorRead);
+  ssum= sum_gpu(&arg_v[0],osites);
 #else
   autoView(arg_v, arg, CpuRead);
-  Integer osites = arg.Grid()->oSites();
   auto ssum= sum_cpu(&arg_v[0],osites);
 #endif  
   arg.Grid()->GlobalSum(ssum);
@@ -171,7 +174,7 @@ inline typename vobj::scalar_object sum(const Lattice<vobj> &arg)
 template<class vobj>
 inline typename vobj::scalar_object sum_large(const Lattice<vobj> &arg)
 {
-#if defined(GRID_CUDA)||defined(GRID_HIP)
+#if defined(GRID_CUDA)||defined(GRID_HIP)||defined(GRID_SYCL)
   autoView( arg_v, arg, AcceleratorRead);
   Integer osites = arg.Grid()->oSites();
   auto ssum= sum_gpu_large(&arg_v[0],osites);
@@ -235,11 +238,10 @@ inline ComplexD rankInnerProduct(const Lattice<vobj> &left,const Lattice<vobj> &
   typedef decltype(innerProductD(vobj(),vobj())) inner_t;
   Vector<inner_t> inner_tmp(sites);
   auto inner_tmp_v = &inner_tmp[0];
-    
   {
     autoView( left_v , left, AcceleratorRead);
     autoView( right_v,right, AcceleratorRead);
-
+    // This code could read coalesce
     // GPU - SIMT lane compliance...
     accelerator_for( ss, sites, 1,{
 	auto x_l = left_v[ss];

Grid/lattice/Lattice_reduction_sycl.h

@@ -0,0 +1,125 @@
+NAMESPACE_BEGIN(Grid);
+
+/////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Possibly promote to double and sum
+/////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+template <class vobj>
+inline typename vobj::scalar_objectD sumD_gpu_tensor(const vobj *lat, Integer osites) 
+{
+  typedef typename vobj::scalar_object sobj;
+  typedef typename vobj::scalar_objectD sobjD;
+  sobj *mysum =(sobj *) malloc_shared(sizeof(sobj),*theGridAccelerator);
+  sobj identity; zeroit(identity);
+  sobj ret ; 
+
+  Integer nsimd= vobj::Nsimd();
+  
+  theGridAccelerator->submit([&](cl::sycl::handler &cgh) {
+     auto Reduction = cl::sycl::reduction(mysum,identity,std::plus<>());
+     cgh.parallel_for(cl::sycl::range<1>{osites},
+		      Reduction,
+		      [=] (cl::sycl::id<1> item, auto &sum) {
+      auto osite   = item[0];
+      sum +=Reduce(lat[osite]);
+     });
+   });
+  theGridAccelerator->wait();
+  ret = mysum[0];
+  free(mysum,*theGridAccelerator);
+  sobjD dret; convertType(dret,ret);
+  return dret;
+}
+
+template <class vobj>
+inline typename vobj::scalar_objectD sumD_gpu_large(const vobj *lat, Integer osites)
+{
+  return sumD_gpu_tensor(lat,osites);
+}
+template <class vobj>
+inline typename vobj::scalar_objectD sumD_gpu_small(const vobj *lat, Integer osites)
+{
+  return sumD_gpu_large(lat,osites);
+}
+
+template <class vobj>
+inline typename vobj::scalar_objectD sumD_gpu(const vobj *lat, Integer osites)
+{
+  return sumD_gpu_large(lat,osites);
+}
+
+/////////////////////////////////////////////////////////////////////////////////////////////////////////
+// Return as same precision as input performing reduction in double precision though
+/////////////////////////////////////////////////////////////////////////////////////////////////////////
+template <class vobj>
+inline typename vobj::scalar_object sum_gpu(const vobj *lat, Integer osites) 
+{
+  typedef typename vobj::scalar_object sobj;
+  sobj result;
+  result = sumD_gpu(lat,osites);
+  return result;
+}
+
+template <class vobj>
+inline typename vobj::scalar_object sum_gpu_large(const vobj *lat, Integer osites)
+{
+  typedef typename vobj::scalar_object sobj;
+  sobj result;
+  result = sumD_gpu_large(lat,osites);
+  return result;
+}
+
+NAMESPACE_END(Grid);
+
+/*
+template<class Double> Double svm_reduce(Double *vec,uint64_t L)
+{
+  Double sumResult; zeroit(sumResult);
+  Double *d_sum =(Double *)cl::sycl::malloc_shared(sizeof(Double),*theGridAccelerator);
+  Double identity;  zeroit(identity);
+  theGridAccelerator->submit([&](cl::sycl::handler &cgh) {
+     auto Reduction = cl::sycl::reduction(d_sum,identity,std::plus<>());
+     cgh.parallel_for(cl::sycl::range<1>{L},
+		      Reduction,
+		      [=] (cl::sycl::id<1> index, auto &sum) {
+	 sum +=vec[index];
+     });
+   });
+  theGridAccelerator->wait();
+  Double ret = d_sum[0];
+  free(d_sum,*theGridAccelerator);
+  std::cout << " svm_reduce finished "<<L<<" sites sum = " << ret <<std::endl;
+  return ret;
+}
+
+template <class vobj>
+inline typename vobj::scalar_objectD sumD_gpu_repack(const vobj *lat, Integer osites)
+{
+  typedef typename vobj::vector_type  vector;
+  typedef typename vobj::scalar_type  scalar;
+
+  typedef typename vobj::scalar_typeD scalarD;
+  typedef typename vobj::scalar_objectD sobjD;
+
+  sobjD ret;
+  scalarD *ret_p = (scalarD *)&ret;
+  
+  const int nsimd = vobj::Nsimd();
+  const int words = sizeof(vobj)/sizeof(vector);
+
+  Vector<scalar> buffer(osites*nsimd);
+  scalar *buf = &buffer[0];
+  vector *dat = (vector *)lat;
+
+  for(int w=0;w<words;w++) {
+
+    accelerator_for(ss,osites,nsimd,{
+	int lane = acceleratorSIMTlane(nsimd);
+	buf[ss*nsimd+lane] = dat[ss*words+w].getlane(lane);
+    });
+    //Precision change at this point is to late to gain precision
+    ret_p[w] = svm_reduce(buf,nsimd*osites);
+  }
+  return ret;
+}
+*/

systems/PVC/benchmarks/run-1tile.sh

@@ -6,7 +6,7 @@
 
 source /nfs/site/home/paboylex/ATS/GridNew/Grid/systems/PVC-nightly/setup.sh
 
-export NT=16
+export NT=8
 
 export I_MPI_OFFLOAD=1
 export I_MPI_OFFLOAD_TOPOLIB=level_zero
@@ -21,7 +21,7 @@ export I_MPI_OFFLOAD_CELL=tile
 export EnableImplicitScaling=0
 export EnableWalkerPartition=0
 export ZE_AFFINITY_MASK=0.0
-mpiexec -launcher ssh -n 1 -host localhost  ./Benchmark_dwf_fp32 --mpi 1.1.1.1 --grid 32.32.32.32 --accelerator-threads $NT --comms-sequential --shm-mpi 1 --cacheblocking 8.8.8.8
+mpiexec -launcher ssh -n 1 -host localhost  ./Benchmark_dwf_fp32 --mpi 1.1.1.1 --grid 32.32.32.32 --accelerator-threads $NT --comms-sequential --shm-mpi 1 --device-mem 32768
 
 export ZE_AFFINITY_MASK=0
 export I_MPI_OFFLOAD_CELL=device

systems/PVC/benchmarks/run-2tile-mpi.sh

@@ -8,7 +8,6 @@ source /nfs/site/home/paboylex/ATS/GridNew/Grid/systems/PVC-nightly/setup.sh
 
 export NT=16
 
-
 # export IGC_EnableLSCFenceUGMBeforeEOT=0
 # export SYCL_PROGRAM_COMPILE_OPTIONS="-ze-opt-large-register-file=False"
 #export IGC_ShaderDumpEnable=1 
@@ -20,14 +19,16 @@ export SYCL_DEVICE_FILTER=gpu,level_zero
 export I_MPI_OFFLOAD_CELL=tile
 export EnableImplicitScaling=0
 export EnableWalkerPartition=0
-#export SYCL_PI_LEVEL_ZERO_DEVICE_SCOPE_EVENTS=1
+export SYCL_PI_LEVEL_ZERO_DEVICE_SCOPE_EVENTS=1
 export SYCL_PI_LEVEL_ZERO_USE_IMMEDIATE_COMMANDLISTS=1
-#export SYCL_PI_LEVEL_ZERO_USE_COPY_ENGINE=0
+export SYCL_PI_LEVEL_ZERO_USE_COPY_ENGINE=0
 
-mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_dwf_fp32 --mpi 1.1.1.2 --grid 32.32.32.64 --accelerator-threads $NT  --shm-mpi 1 > dw.2tile.1x2.log
+for i in 0 
-mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_dwf_fp32 --mpi 2.1.1.1 --grid 64.32.32.32 --accelerator-threads $NT  --shm-mpi 1 > dw.2tile.2x1.log
+do
-
+mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_dwf_fp32 --mpi 1.1.1.2 --grid 32.32.32.64 --accelerator-threads $NT  --shm-mpi 1  --device-mem 32768
-mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_halo --mpi 1.1.1.2 --grid 32.32.32.64 --accelerator-threads $NT  --shm-mpi 1 > halo.2tile.1x2.log
+mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_dwf_fp32 --mpi 2.1.1.1 --grid 64.32.32.32 --accelerator-threads $NT  --shm-mpi 1  --device-mem 32768
-mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_halo --mpi 2.1.1.1 --grid 64.32.32.32 --accelerator-threads $NT  --shm-mpi 1 > halo.2tile.2x1.log
+done
+#mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_halo --mpi 1.1.1.2 --grid 32.32.32.64 --accelerator-threads $NT  --shm-mpi 1 > halo.2tile.1x2.log
+#mpiexec -launcher ssh -n 2 -host localhost  ./wrap4gpu.sh ./Benchmark_halo --mpi 2.1.1.1 --grid 64.32.32.32 --accelerator-threads $NT  --shm-mpi 1 > halo.2tile.2x1.log
 
 

systems/PVC/config-command

@@ -2,14 +2,15 @@ INSTALL=/nfs/site/home/azusayax/install
 ../../configure \
 	--enable-simd=GPU \
 	--enable-gen-simd-width=64 \
-	--enable-comms=mpi \
+	--enable-comms=mpi-auto \
 	--disable-accelerator-cshift \
 	--disable-gparity \
 	--disable-fermion-reps \
 	--enable-shm=nvlink \
 	--enable-accelerator=sycl \
-	--enable-unified=yes \
+	--enable-unified=no \
-	CXX=mpicxx \
+	MPICXX=mpicxx \
+	CXX=dpcpp \
 	LDFLAGS="-fsycl-device-code-split=per_kernel -fsycl-device-lib=all -lze_loader -L$INSTALL/lib" \
-	CXXFLAGS="-cxx=dpcpp -fsycl-unnamed-lambda -fsycl -no-fma -I$INSTALL/include -Wtautological-constant-compare"
+	CXXFLAGS="-fsycl-unnamed-lambda -fsycl -no-fma -I$INSTALL/include -Wno-tautological-compare"