updates debug trace outputs

src/generate_databases/save_arrays_solver.F90

@@ -729,6 +729,19 @@ subroutine save_arrays_solver_files(nspec,ibool)
     endif ! NPROC > 1
   endif  !if (SAVE_MESH_FILES_ADDITIONAL)
 
+  !debug
+  ! saves ispec number
+  !allocate(v_tmp_i(nspec),stat=ier)
+  !if (ier /= 0) call exit_MPI_without_rank('error allocating array 656')
+  !if (ier /= 0) stop 'error allocating array v_tmp_i'
+  !do i = 1,nspec
+  !  v_tmp_i(i) = i
+  !enddo
+  !filename = prname(1:len_trim(prname))//'ispec_number'
+  !call write_VTK_data_elem_i(nspec,nglob_unique,xstore_unique,ystore_unique,zstore_unique,ibool, &
+  !                           v_tmp_i,filename)
+  !deallocate(v_tmp_i)
+
   end subroutine save_arrays_solver_files
 
 !

src/gpu/assemble_MPI_vector_cuda.cu

@@ -43,7 +43,7 @@ void FC_FUNC_(transfer_boun_accel_from_device,
               TRANSFER_BOUN_ACCEL_FROM_DEVICE)(long* Mesh_pointer,
                                                realw* send_accel_buffer,
                                                const int* FORWARD_OR_ADJOINT){
-TRACE("\ttransfer_boun_accel_from_device");
+TRACE("transfer_boun_accel_from_device");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 
@@ -123,7 +123,7 @@ void FC_FUNC_(transfer_boundary_from_device_a,
 
 // asynchronous transfer from device to host
 
-  TRACE("\ttransfer_boundary_from_device_a");
+  TRACE("transfer_boundary_from_device_a");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); // get Mesh from fortran integer wrapper
 
@@ -210,7 +210,7 @@ void FC_FUNC_(transfer_asmbl_accel_to_device,
                                               const int* nibool_interfaces_ext_mesh,
                                               const int* ibool_interfaces_ext_mesh,
                                               const int* FORWARD_OR_ADJOINT) {
-TRACE("\ttransfer_asmbl_accel_to_device");
+TRACE("transfer_asmbl_accel_to_device");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 
@@ -310,7 +310,7 @@ void FC_FUNC_(transfer_sync_accel_to_device,
                                               const int* nibool_interfaces_ext_mesh,
                                               const int* ibool_interfaces_ext_mesh,
                                               const int* FORWARD_OR_ADJOINT) {
-TRACE("\ttransfer_sync_accel_to_device");
+TRACE("transfer_sync_accel_to_device");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 

src/gpu/check_fields_cuda.cu

@@ -377,7 +377,7 @@ void FC_FUNC_(get_norm_elastic_from_device,
                                             long* Mesh_pointer,
                                             int* type) {
 
-  TRACE("\tget_norm_elastic_from_device");
+  TRACE("get_norm_elastic_from_device");
   //double start_time = get_time_val();
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container

src/gpu/compute_add_sources_viscoelastic_cuda.cu

@@ -41,7 +41,7 @@ void FC_FUNC_(compute_add_sources_el_cuda,
                                            double* h_stf_pre_compute,
                                            int* h_NSOURCES) {
 
-  TRACE("\tcompute_add_sources_el_cuda");
+  TRACE("compute_add_sources_el_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 
@@ -107,7 +107,7 @@ void FC_FUNC_(compute_add_sources_el_s3_cuda,
                                               double* h_stf_pre_compute,
                                               int* h_NSOURCES) {
 
-  TRACE("\tcompute_add_sources_el_s3_cuda");
+  TRACE("compute_add_sources_el_s3_cuda");
   // EPIK_TRACER("compute_add_sources_el_s3_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
@@ -176,7 +176,7 @@ void FC_FUNC_(add_source_main_rec_noise_cu,
                                             int* irec_main_noise_f,
                                             int* islice_selected_rec) {
 
-TRACE("\tadd_source_main_rec_noise_cu");
+TRACE("add_source_main_rec_noise_cu");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 
@@ -229,7 +229,7 @@ void FC_FUNC_(add_sources_el_sim_type_2_or_3,
                                               int* NTSTEP_BETWEEN_READ_ADJSRC,
                                               int* it) {
 
-  TRACE("\tadd_sources_el_sim_type_2_or_3");
+  TRACE("add_sources_el_sim_type_2_or_3");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 

src/gpu/compute_coupling_cuda.cu

@@ -159,6 +159,7 @@ void FC_FUNC_(compute_coupling_el_ac_cuda,
   field* potential_dot_dot;
   realw *accel,*displ;
   int backward_simulation;
+
   if (*FORWARD_OR_ADJOINT == 1) {
     // forward fields
     backward_simulation = 0;
@@ -237,7 +238,7 @@ void FC_FUNC_(compute_coupling_ocean_cuda,
               COMPUTE_COUPLING_OCEAN_CUDA)(long* Mesh_pointer,
                                            int* FORWARD_OR_ADJOINT) {
 
-  TRACE("\tcompute_coupling_ocean_cuda");
+  TRACE("compute_coupling_ocean_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
 

src/gpu/compute_forces_viscoelastic_cuda.cu

@@ -68,7 +68,7 @@ void Kernel_2(int nb_blocks_to_compute,Mesh* mp,int d_iphase,realw d_deltat,
               realw* d_rhostore,
               const int FORWARD_OR_ADJOINT){
 
-  TRACE("\tKernel_2");
+  TRACE("Kernel_2");
 
   GPU_ERROR_CHECKING("before kernel Kernel 2");
 
@@ -1066,7 +1066,7 @@ void FC_FUNC_(compute_forces_viscoelastic_cuda,
                                                 int* ATTENUATION,
                                                 int* FORWARD_OR_ADJOINT_f) {
 
-  TRACE("\tcompute_forces_viscoelastic_cuda");
+  TRACE("compute_forces_viscoelastic_cuda");
   // EPIK_TRACER("compute_forces_viscoelastic_cuda");
   //printf("Running compute_forces\n");
   //double start_time = get_time_val();

src/gpu/compute_stacey_viscoelastic_cuda.cu

@@ -42,7 +42,7 @@ void FC_FUNC_(compute_stacey_viscoelastic_cuda,
                                                 realw* b_absorb_field,
                                                 int* FORWARD_OR_ADJOINT_f) {
 
-  TRACE("\tcompute_stacey_viscoelastic_cuda");
+  TRACE("compute_stacey_viscoelastic_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
   int FORWARD_OR_ADJOINT = *FORWARD_OR_ADJOINT_f;
@@ -226,7 +226,7 @@ void FC_FUNC_(compute_stacey_viscoelastic_undoatt_cuda,
                                                         int* iphasef,
                                                         int* FORWARD_OR_ADJOINT_f) {
 
-  TRACE("\tcompute_stacey_viscoelastic_undoatt_cuda");
+  TRACE("compute_stacey_viscoelastic_undoatt_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); //get mesh pointer out of fortran integer container
   int FORWARD_OR_ADJOINT = *FORWARD_OR_ADJOINT_f;

src/gpu/kernels/kernel_3_accel_cuda_device.cu

@@ -39,7 +39,7 @@ __global__ void kernel_3_accel_cuda_device(realw_p accel,
   // because of block and grid sizing problems, there is a small
   // amount of buffer at the end of the calculation
   if (id < size) {
-    accel[3*id  ] = accel[3*id  ] * rmassx[id];;
+    accel[3*id  ] = accel[3*id  ] * rmassx[id];
     accel[3*id+1] = accel[3*id+1] * rmassy[id];
     accel[3*id+2] = accel[3*id+2] * rmassz[id];
   }

src/gpu/mesh_constants_gpu.h

@@ -780,7 +780,7 @@ static inline void get_blocks_xy(int num_blocks, int* num_blocks_x, int* num_blo
   }
 
 #if DEBUG == 1
-  printf("work group - total %d has group size x = %d / y = %d\n",
+  printf("\twork group - total %d has group size x = %d / y = %d\n",
          num_blocks,*num_blocks_x,*num_blocks_y);
 #endif
 
@@ -794,7 +794,7 @@ static inline void get_blocks_xy(int num_blocks, int* num_blocks_x, int* num_blo
   }
 
 #if DEBUG == 1
-  printf("balancing work group with limit size %d - total %d has group size x = %d / y = %d\n",
+  printf("\tbalancing work group with limit size %d - total %d has group size x = %d / y = %d\n",
          BALANCE_WORK_GROUP_UNITS,num_blocks,*num_blocks_x,*num_blocks_y);
 #endif
 

src/gpu/update_displacement_cuda.cu

@@ -43,7 +43,7 @@ void FC_FUNC_(update_displacement_cuda,
                                           realw* deltatsqover2_F,
                                           int* FORWARD_OR_ADJOINT) {
 
-  TRACE("\tupdate_displacement_cuda");
+  TRACE("update_displacement_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); // get Mesh from fortran integer wrapper
 
@@ -130,7 +130,7 @@ void FC_FUNC_(update_displacement_ac_cuda,
                                            realw* deltatover2_F,
                                            realw* deltatsqover2_F,
                                            int* FORWARD_OR_ADJOINT) {
-  TRACE("\tupdate_displacement_ac_cuda");
+  TRACE("update_displacement_ac_cuda");
   Mesh* mp = (Mesh*)(*Mesh_pointer); // get Mesh from fortran integer wrapper
 
   // safety check
@@ -226,7 +226,7 @@ void FC_FUNC_(kernel_3_a_cuda,
                                int* APPROXIMATE_OCEAN_LOAD,
                                int* FORWARD_OR_ADJOINT) {
 
-  TRACE("\tkernel_3_a_cuda");
+  TRACE("kernel_3_a_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); // get Mesh from fortran integer wrapper
 
@@ -319,7 +319,7 @@ void FC_FUNC_(kernel_3_b_cuda,
                                realw* deltatover2_F,
                                realw* b_deltatover2_F,
                                int* FORWARD_OR_ADJOINT) {
-  TRACE("\tkernel_3_b_cuda");
+  TRACE("kernel_3_b_cuda");
 
   Mesh* mp = (Mesh*)(*Mesh_pointer); // get Mesh from fortran integer wrapper