Merge release-5-0 into master

Change-Id: I23579c289b8a057a94cf64b142759934b9402ee5

diff --git a/cmake/cmake_uninstall.cmake.in b/cmake/cmake_uninstall.cmake.in
index 17376d2e8236182ee62c1934630b1da7412e97f8..cb708819ade0750fc443f32a413010d21f03e4c2 100644 (file)
--- a/cmake/cmake_uninstall.cmake.in
+++ b/cmake/cmake_uninstall.cmake.in
@@ -6,7 +6,7 @@ FILE(READ "@CMAKE_CURRENT_BINARY_DIR@/install_manifest.txt" files)
 STRING(REGEX REPLACE "\n" ";" files "${files}")
 FOREACH(file ${files})
   MESSAGE(STATUS "Uninstalling \"$ENV{DESTDIR}${file}\"")
-  IF(EXISTS "$ENV{DESTDIR}${file}")
+  IF(EXISTS "$ENV{DESTDIR}${file}" OR IS_SYMLINK "$ENV{DESTDIR}${file}")
     EXEC_PROGRAM(
       "@CMAKE_COMMAND@" ARGS "-E remove \"$ENV{DESTDIR}${file}\""
       OUTPUT_VARIABLE rm_out
@@ -15,8 +15,8 @@ FOREACH(file ${files})
     IF(NOT "${rm_retval}" STREQUAL 0)
       MESSAGE(FATAL_ERROR "Problem when removing \"$ENV{DESTDIR}${file}\"")
     ENDIF(NOT "${rm_retval}" STREQUAL 0)
-  ELSE(EXISTS "$ENV{DESTDIR}${file}")
+  ELSE(EXISTS "$ENV{DESTDIR}${file}" OR IS_SYMLINK "$ENV{DESTDIR}${file}")
     MESSAGE(STATUS "File \"$ENV{DESTDIR}${file}\" does not exist.")
-  ENDIF(EXISTS "$ENV{DESTDIR}${file}")
+  ENDIF(EXISTS "$ENV{DESTDIR}${file}" OR IS_SYMLINK "$ENV{DESTDIR}${file}")
 ENDFOREACH(file)
 

diff --git a/src/gromacs/ewald/pme.c b/src/gromacs/ewald/pme.c
index 965111e4666f1fbda7cf766e3571988b058b2311..3a5c710a957c3eb69330a6c192234c5371fc59d8 100644 (file)
--- a/src/gromacs/ewald/pme.c
+++ b/src/gromacs/ewald/pme.c
@@ -3,7 +3,7 @@
  *
  * Copyright (c) 1991-2000, University of Groningen, The Netherlands.
  * Copyright (c) 2001-2004, The GROMACS development team.
- * Copyright (c) 2013,2014, by the GROMACS development team, led by
+ * Copyright (c) 2013,2014,2015, by the GROMACS development team, led by
  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
  * and including many others, as listed in the AUTHORS file in the
  * top-level source directory and at http://www.gromacs.org.
@@ -3854,7 +3854,7 @@ reduce_threadgrid_overlap(gmx_pme_t pme,
     int  fft_my, fft_mz;
     int  buf_my = -1;
     int  nsx, nsy, nsz;
-    ivec localcopy_end;
+    ivec localcopy_end, commcopy_end;
     int  offx, offy, offz, x, y, z, i0, i0t;
     int  sx, sy, sz, fx, fy, fz, tx1, ty1, tz1, ox, oy, oz;
     gmx_bool bClearBufX, bClearBufY, bClearBufXY, bClearBuf;
@@ -3896,8 +3896,22 @@ reduce_threadgrid_overlap(gmx_pme_t pme,
          * not beyond the local FFT grid.
          */
         localcopy_end[d] =
-            min(pmegrid->offset[d]+pmegrid->n[d]-(pmegrid->order-1),
+            min(pmegrid->offset[d] + pmegrid->n[d] - (pmegrid->order - 1),
                 local_fft_ndata[d]);
+
+        /* Determine up to where our thread needs to copy from the
+         * thread-local charge spreading grid to the communication buffer.
+         * Note: only relevant with communication, ignored otherwise.
+         */
+        commcopy_end[d]  = localcopy_end[d];
+        if (pmegrid->ci[d] == pmegrids->nc[d] - 1)
+        {
+            /* The last thread should copy up to the last pme grid line.
+             * When the rank-local FFT grid is narrower than pme-order,
+             * we need the max below to ensure copying of all data.
+             */
+            commcopy_end[d] = max(commcopy_end[d], pme->pme_order);
+        }
     }
 
     offx = pmegrid->offset[XX];
@@ -3928,17 +3942,11 @@ reduce_threadgrid_overlap(gmx_pme_t pme,
         }
         pmegrid_g = &pmegrids->grid_th[fx*pmegrids->nc[YY]*pmegrids->nc[ZZ]];
         ox       += pmegrid_g->offset[XX];
-        /* Determine the end of our part of the source grid */
-        if (!bCommX)
-        {
-            /* Use our thread local source grid and target grid part */
-            tx1 = min(ox + pmegrid_g->n[XX], localcopy_end[XX]);
-        }
-        else
-        {
-            /* Use our thread local source grid and the spreading range */
-            tx1 = min(ox + pmegrid_g->n[XX], pme->pme_order);
-        }
+        /* Determine the end of our part of the source grid.
+         * Use our thread local source grid and target grid part
+         */
+        tx1 = min(ox + pmegrid_g->n[XX],
+                  !bCommX ? localcopy_end[XX] : commcopy_end[XX]);
 
         for (sy = 0; sy >= -pmegrids->nthread_comm[YY]; sy--)
         {
@@ -3953,17 +3961,11 @@ reduce_threadgrid_overlap(gmx_pme_t pme,
             }
             pmegrid_g = &pmegrids->grid_th[fy*pmegrids->nc[ZZ]];
             oy       += pmegrid_g->offset[YY];
-            /* Determine the end of our part of the source grid */
-            if (!bCommY)
-            {
-                /* Use our thread local source grid and target grid part */
-                ty1 = min(oy + pmegrid_g->n[YY], localcopy_end[YY]);
-            }
-            else
-            {
-                /* Use our thread local source grid and the spreading range */
-                ty1 = min(oy + pmegrid_g->n[YY], pme->pme_order);
-            }
+            /* Determine the end of our part of the source grid.
+             * Use our thread local source grid and target grid part
+             */
+            ty1 = min(oy + pmegrid_g->n[YY],
+                      !bCommY ? localcopy_end[YY] : commcopy_end[YY]);
 
             for (sz = 0; sz >= -pmegrids->nthread_comm[ZZ]; sz--)
             {

diff --git a/src/gromacs/mdlib/nbnxn_cuda/nbnxn_cuda.cu b/src/gromacs/mdlib/nbnxn_cuda/nbnxn_cuda.cu
index 793f13fc7c8e1bf8dbf6ad8d28a26e2990d9f61c..9ee57c656da43a53753988d19d5df8a34fab331a 100644 (file)
--- a/src/gromacs/mdlib/nbnxn_cuda/nbnxn_cuda.cu
+++ b/src/gromacs/mdlib/nbnxn_cuda/nbnxn_cuda.cu
@@ -1,7 +1,7 @@
 /*
  * This file is part of the GROMACS molecular simulation package.
  *
- * Copyright (c) 2012,2013,2014, by the GROMACS development team, led by
+ * Copyright (c) 2012,2013,2014,2015, by the GROMACS development team, led by
  * Mark Abraham, David van der Spoel, Berk Hess, and Erik Lindahl,
  * and including many others, as listed in the AUTHORS file in the
  * top-level source directory and at http://www.gromacs.org.
@@ -327,7 +327,7 @@ void nbnxn_cuda_launch_kernel(nbnxn_cuda_ptr_t        cu_nb,
     cu_timers_t         *t       = cu_nb->timers;
     cudaStream_t         stream  = cu_nb->stream[iloc];
 
-    bool                 bCalcEner   = flags & GMX_FORCE_VIRIAL;
+    bool                 bCalcEner   = flags & GMX_FORCE_ENERGY;
     bool                 bCalcFshift = flags & GMX_FORCE_VIRIAL;
     bool                 bDoTime     = cu_nb->bDoTime;
 
@@ -465,7 +465,7 @@ void nbnxn_cuda_launch_cpyback(nbnxn_cuda_ptr_t        cu_nb,
     bool             bDoTime = cu_nb->bDoTime;
     cudaStream_t     stream  = cu_nb->stream[iloc];
 
-    bool             bCalcEner   = flags & GMX_FORCE_VIRIAL;
+    bool             bCalcEner   = flags & GMX_FORCE_ENERGY;
     bool             bCalcFshift = flags & GMX_FORCE_VIRIAL;
 
     /* don't launch copy-back if there was no work to do */
@@ -621,7 +621,7 @@ void nbnxn_cuda_wait_gpu(nbnxn_cuda_ptr_t cu_nb,
     wallclock_gpu_t *timings  = cu_nb->timings;
     nb_staging       nbst     = cu_nb->nbst;
 
-    bool             bCalcEner   = flags & GMX_FORCE_VIRIAL;
+    bool             bCalcEner   = flags & GMX_FORCE_ENERGY;
     bool             bCalcFshift = flags & GMX_FORCE_VIRIAL;
 
     /* turn energy calculation always on/off (for debugging/testing only) */