Fixed Verlet buffer issue with 2-wide SIMD

The Verlet buffer size for CPUs was always calculated for 4x4.
With 2-wide SIMD the estimate should be for 4x2, which results
in a slighly larger list buffer.
grompp now always sets rlist for a 4x4 list setup; mdrun anyhow
redetermines rlist at run time (added a note for this in grompp).

Fixes #1757.

Change-Id: If4bf9ad17b82b22c9d9f7c1dd3f88e66f2314df4

diff --git a/src/gromacs/gmxpreprocess/calc_verletbuf.c b/src/gromacs/gmxpreprocess/calc_verletbuf.c
index 0387dd1483353b22b6f26afddde41037e57d32f8..28f2d91664fe2ef2e20e8b506cd863546524630a 100644 (file)
--- a/src/gromacs/gmxpreprocess/calc_verletbuf.c
+++ b/src/gromacs/gmxpreprocess/calc_verletbuf.c
@@ -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.
@@ -49,6 +49,7 @@
 #include "coulomb.h"
 #include "calc_verletbuf.h"
 #include "../mdlib/nbnxn_consts.h"
+#include "../mdlib/nbnxn_simd.h"
 
 #ifdef GMX_NBNXN_SIMD
 /* The include below sets the SIMD instruction type (precision+width)
@@ -117,25 +118,28 @@ typedef struct
     int                           n;    /* #atoms of this type in the system */
 } verletbuf_atomtype_t;
 
-void verletbuf_get_list_setup(gmx_bool                bGPU,
+void verletbuf_get_list_setup(gmx_bool gmx_unused     bSIMD,
+                              gmx_bool                bGPU,
                               verletbuf_list_setup_t *list_setup)
 {
-    list_setup->cluster_size_i     = NBNXN_CPU_CLUSTER_I_SIZE;
-
     if (bGPU)
     {
-        list_setup->cluster_size_j = NBNXN_GPU_CLUSTER_SIZE;
+        list_setup->cluster_size_i     = NBNXN_GPU_CLUSTER_SIZE;
+        list_setup->cluster_size_j     = NBNXN_GPU_CLUSTER_SIZE;
     }
     else
     {
-#ifndef GMX_NBNXN_SIMD
-        list_setup->cluster_size_j = NBNXN_CPU_CLUSTER_I_SIZE;
-#else
-        list_setup->cluster_size_j = GMX_SIMD_REAL_WIDTH;
+        list_setup->cluster_size_i     = NBNXN_CPU_CLUSTER_I_SIZE;
+        list_setup->cluster_size_j     = NBNXN_CPU_CLUSTER_I_SIZE;
+#ifdef GMX_NBNXN_SIMD
+        if (bSIMD)
+        {
+            list_setup->cluster_size_j = GMX_SIMD_REAL_WIDTH;
 #ifdef GMX_NBNXN_SIMD_2XNN
-        /* We assume the smallest cluster size to be on the safe side */
-        list_setup->cluster_size_j /= 2;
+            /* We assume the smallest cluster size to be on the safe side */
+            list_setup->cluster_size_j /= 2;
 #endif
+        }
 #endif
     }
 }

diff --git a/src/gromacs/gmxpreprocess/calc_verletbuf.h b/src/gromacs/gmxpreprocess/calc_verletbuf.h
index 65ef4923506c4fe15eb99da7c653d7575670e4cd..d54ecd437c873b963db470b0ed1703908a9d985e 100644 (file)
--- a/src/gromacs/gmxpreprocess/calc_verletbuf.h
+++ b/src/gromacs/gmxpreprocess/calc_verletbuf.h
@@ -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.
@@ -63,8 +63,11 @@ static const real verlet_buffer_ratio_NVE_T0     = 0.10;
 /* Sets the pair-list setup assumed for the current Gromacs configuration.
  * The setup with smallest cluster sizes is return, such that the Verlet
  * buffer size estimated with this setup will be conservative.
+ * bSIMD tells if to take into account SIMD, when supported.
+ * bGPU tells to estimate for GPU kernels (bSIMD is ignored with bGPU=TRUE)
  */
-void verletbuf_get_list_setup(gmx_bool                bGPU,
+void verletbuf_get_list_setup(gmx_bool                bSIMD,
+                              gmx_bool                bGPU,
                               verletbuf_list_setup_t *list_setup);
 
 

diff --git a/src/gromacs/gmxpreprocess/grompp.c b/src/gromacs/gmxpreprocess/grompp.c
index b63ce1020f3155bbc7d11f9c3754c3e81b792767..53bdb699c375df66a65f8880d612556d21447a04 100644 (file)
--- a/src/gromacs/gmxpreprocess/grompp.c
+++ b/src/gromacs/gmxpreprocess/grompp.c
@@ -1374,7 +1374,7 @@ static void set_verlet_buffer(const gmx_mtop_t *mtop,
                             &ls, &n_nonlin_vsite, &rlist_1x1);
 
     /* Set the pair-list buffer size in ir */
-    verletbuf_get_list_setup(FALSE, &ls);
+    verletbuf_get_list_setup(FALSE, FALSE, &ls);
     calc_verlet_buffer_size(mtop, det(box), ir, buffer_temp,
                             &ls, &n_nonlin_vsite, &ir->rlist);
 
@@ -1392,6 +1392,8 @@ static void set_verlet_buffer(const gmx_mtop_t *mtop,
            ls.cluster_size_i, ls.cluster_size_j,
            ir->rlist, ir->rlist-max(ir->rvdw, ir->rcoulomb));
 
+    printf("Note that mdrun will redetermine rlist based on the actual pair-list setup\n");
+
     if (sqr(ir->rlistlong) >= max_cutoff2(ir->ePBC, box))
     {
         gmx_fatal(FARGS, "The pair-list cut-off (%g nm) is longer than half the shortest box vector or longer than the smallest box diagonal element (%g nm). Increase the box size or decrease nstlist or increase verlet-buffer-tolerance.", ir->rlistlong, sqrt(max_cutoff2(ir->ePBC, box)));

diff --git a/src/programs/mdrun/runner.c b/src/programs/mdrun/runner.c
index 744e9f501e301cede3bcaac6434dccf8d4eb2e13..b951255d7c875af3938ab8740abf8035d537cf75 100644 (file)
--- a/src/programs/mdrun/runner.c
+++ b/src/programs/mdrun/runner.c
@@ -602,7 +602,7 @@ static void increase_nstlist(FILE *fp, t_commrec *cr,
         ir->nstlist = nstlist_cmdline;
     }
 
-    verletbuf_get_list_setup(bGPU, &ls);
+    verletbuf_get_list_setup(TRUE, bGPU, &ls);
 
     /* Allow rlist to make the list a given factor larger than the list
      * would be with nstlist=10.
@@ -727,7 +727,7 @@ static void prepare_verlet_scheme(FILE                           *fplog,
          * calc_verlet_buffer_size gives the same results for 4x8 and 4x4
          * and 4x2 gives a larger buffer than 4x4, this is ok.
          */
-        verletbuf_get_list_setup(bUseGPU, &ls);
+        verletbuf_get_list_setup(TRUE, bUseGPU, &ls);
 
         calc_verlet_buffer_size(mtop, det(box), ir, -1, &ls, NULL, &rlist_new);
 
@@ -828,7 +828,7 @@ static void convert_to_verlet_scheme(FILE *fplog,
     {
         verletbuf_list_setup_t ls;
 
-        verletbuf_get_list_setup(FALSE, &ls);
+        verletbuf_get_list_setup(TRUE, FALSE, &ls);
         calc_verlet_buffer_size(mtop, box_vol, ir, -1, &ls, NULL, &ir->rlist);
     }
     else