Fix AVX double compile error

Introduced by fa181964de6c9

Note that this change will be included in the 4.6 release, therefore
this commit will effectively be the one that the 4.6 release is made
from instead of the parent.

Change-Id: Ieeede7d15109f2c44fcd2e0ef59eff2944206eab

diff --git a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwCSTab_GeomP1P1_avx_256_double.c b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwCSTab_GeomP1P1_avx_256_double.c
index 600922e5406ba453414d9bf09127e5e6c3959ea4..ff480cd0fa2e06645a1eb777ffc9853313eb9e2f 100644 (file)
--- a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwCSTab_GeomP1P1_avx_256_double.c
+++ b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwCSTab_GeomP1P1_avx_256_double.c
@@ -429,7 +429,7 @@ nb_kernel_ElecGB_VdwCSTab_GeomP1P1_VF_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;
@@ -893,7 +893,7 @@ nb_kernel_ElecGB_VdwCSTab_GeomP1P1_F_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;

diff --git a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwLJ_GeomP1P1_avx_256_double.c b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwLJ_GeomP1P1_avx_256_double.c
index 2f1afc6fb62e7590420ff91ce92f72626998081b..930a8c817c3adb974d5f594b398dc08cb827c9eb 100644 (file)
--- a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwLJ_GeomP1P1_avx_256_double.c
+++ b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwLJ_GeomP1P1_avx_256_double.c
@@ -397,7 +397,7 @@ nb_kernel_ElecGB_VdwLJ_GeomP1P1_VF_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;
@@ -810,7 +810,7 @@ nb_kernel_ElecGB_VdwLJ_GeomP1P1_F_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;

diff --git a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwNone_GeomP1P1_avx_256_double.c b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwNone_GeomP1P1_avx_256_double.c
index 5eb2cfd9d91d24f92a52aeede253e7b4557ab6ff..9eba2970a0070d702f1f61e0f67e3af2a49f8b74 100644 (file)
--- a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwNone_GeomP1P1_avx_256_double.c
+++ b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_ElecGB_VdwNone_GeomP1P1_avx_256_double.c
@@ -355,7 +355,7 @@ nb_kernel_ElecGB_VdwNone_GeomP1P1_VF_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;
@@ -720,7 +720,7 @@ nb_kernel_ElecGB_VdwNone_GeomP1P1_F_avx_256_double
             FF               = _mm256_add_pd(Fp,_mm256_mul_pd(gbeps,_mm256_add_pd(G,_mm256_add_pd(Heps,Heps))));
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r00)));
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /* The pointers to scratch make sure that this code with compilers that take gmx_restrict seriously (e.g. icc 13) really can't screw things up. */
             fjptrA             = (jnrlistA>=0) ? dvda+jnrA : scratch;

diff --git a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_template_avx_256_double.pre b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_template_avx_256_double.pre
index 2d1922cface15a4259c813cd10257bcf066cb8ec..65ec4cf35c665b14567b631a9c642a9c94cd33ed 100644 (file)
--- a/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_template_avx_256_double.pre
+++ b/src/gmxlib/nonbonded/nb_kernel_avx_256_double/nb_kernel_template_avx_256_double.pre
@@ -1,7 +1,7 @@
 /* ## */
 /* ## This file is part of the GROMACS molecular simulation package. */
 /* ## */
-/* ## Copyright (c) 2012, by the GROMACS development team, led by */
+/* ## Copyright (c) 2012,2013, by the GROMACS development team, led by */
 /* ## David van der Spoel, Berk Hess, Erik Lindahl, and including many */
 /* ## others, as listed in the AUTHORS file in the top-level source */
 /* ## directory and at http://www.gromacs.org. */
@@ -629,7 +629,7 @@ void
             fgb              = _mm256_mul_pd(gbqqfactor,_mm256_mul_pd(FF,gbscale));
             dvdatmp          = _mm256_mul_pd(minushalf,_mm256_add_pd(vgb,_mm256_mul_pd(fgb,r{I}{J})));
             /*                 #if ROUND == 'Epilogue' */
-            dvdatmp          = _mm256_andnot_ps(dummy_mask,dvdatmp);
+            dvdatmp          = _mm256_andnot_pd(dummy_mask,dvdatmp);
             /*                 #endif */
             dvdasum          = _mm256_add_pd(dvdasum,dvdatmp);
             /*                 #if ROUND == 'Loop' */