Merge branch 'develop' into mtblas

Author: mohanchen

docs/advanced/input_files/kpt.md

@@ -8,7 +8,7 @@ ABACUS uses periodic boundary conditions for both crystals and finite systems. F
 
 ## Gamma-only Calculations
 
-In ABACUS, we offer th option of running gamma-only calculations for LCAO basis by setting [gamma_only](./input-main.md#gamma_only) to be 1. Due to details of implementation, gamma-only calculation will be slightly faster than running a non gamma-only calculation and explicitly setting gamma point to be the only the k-point, but the results should be consistent.
+In ABACUS, we offer the option of running gamma-only calculations for LCAO basis by setting [gamma_only](./input-main.md#gamma_only) to be 1. Due to details of implementation, gamma-only calculation will be slightly faster than running a non gamma-only calculation and explicitly setting gamma point to be the only the k-point, but the results should be consistent.
 
 > If gamma_only is set to 1, the KPT file will be overwritten. So make sure to turn off gamma_only for multi-k calculations.
 
@@ -23,7 +23,7 @@ method to generate k-mesh, and the following is an example input k-point (`KPT`)
 K_POINTS //keyword for start
 0 //total number of k-point, `0' means generate automatically
 Gamma //which kind of Monkhorst-Pack method, `Gamma' or `MP'
-2 2 2 0 0 0 //first three number: subdivisions along recpri. vectors
+2 2 2 0 0 0 //first three number: subdivisions along reciprocal vectors
             //last three number: shift of the mesh
 ```
 
@@ -63,8 +63,8 @@ Direct //`Direct' or `Cartesian' coordinate
 ## Band structure calculations
 
 ABACUS uses specified high-symmetry directions of the Brillouin zone for band structure
-calculations. The third line of k-point file should start with ‘Line’ or ‘Line_Cartesian’ for
-line mode. ‘Line’ means the positions below are in Direct coordinates, while ‘Line_Cartesian’
+calculations. The third line of k-point file should start with 'Line' or 'Line_Cartesian' for
+line mode. 'Line' means the positions below are in Direct coordinates, while 'Line_Cartesian'
 means in Cartesian coordinates:
 
 ```

source/module_hamilt_lcao/module_gint/gint_vl_gpu.cu

@@ -47,7 +47,8 @@ void gint_vl_gpu(hamilt::HContainer<double>* hRGint,
         checkCuda(cudaStreamCreate(&streams[i]));
     }
 
-    Cuda_Mem_Wrapper<double> grid_vlocal_g(hRGint->get_nnr(), 1, false);
+    const int nnrg = is_gamma_only ? hRGint->get_nnr() : gridt.nnrg;
+    Cuda_Mem_Wrapper<double> grid_vlocal_g(nnrg, 1, false);
     grid_vlocal_g.memset_device_sync();
 
     Cuda_Mem_Wrapper<double> dr_part(max_atom_per_z * 3, num_streams, true);
@@ -189,15 +190,15 @@ void gint_vl_gpu(hamilt::HContainer<double>* hRGint,
         checkCuda(cudaMemcpy(
             hRGint->get_wrapper(),
             grid_vlocal_g.get_device_pointer(),
-            hRGint->get_nnr() * sizeof(double),
+            nnrg * sizeof(double),
             cudaMemcpyDeviceToHost));
     }
     else
     {
         checkCuda(cudaMemcpy(
             pvpR,
             grid_vlocal_g.get_device_pointer(),
-            hRGint->get_nnr() * sizeof(double),
+            nnrg * sizeof(double),
             cudaMemcpyDeviceToHost));
     }
     for (int i = 0; i < num_streams; i++)

source/module_hsolver/diago_elpa_native.cpp

@@ -81,7 +81,11 @@ void DiagoElpaNative<T>::diag_pool(hamilt::MatrixBlock<T>& h_mat,
     // elpa_init(20210430);
     int success;
     elpa_t handle = elpa_allocate(&success);
+#ifdef _OPENMP
     int num_threads = omp_get_max_threads();
+#else
+    int num_threads = 1;
+#endif
     elpa_set(handle, "omp_threads", num_threads, &success);
     elpa_set(handle, "na", (int)nFull, &success);
     elpa_set(handle, "nev", (int)nev, &success);