Rename load1DualHsimd to loadU1DualHsimd

Documentation didn't require any alignment, test didn't use
alignment and all implementations didn't require any alignment.
But name suggested that alignment is required.
Only current usage had 2-wide alignment but requiring that
would make the function less general without any advantage.

Change-Id: I651c1327a3febc368cb4b039ad226d0771770e60

diff --git a/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_outer.h b/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_outer.h
index 337f9860c48378a1c2c933150609789b522c1c80..2aa994455f01171c3eacf1a4e5e237ad728ae9f2 100644 (file)
--- a/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_outer.h
+++ b/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_outer.h
@@ -475,12 +475,12 @@
         /* Load i atom data */
         int sciy             = scix + STRIDE;
         int sciz             = sciy + STRIDE;
-        ix_S0          = load1DualHsimd(x+scix);
-        ix_S2          = load1DualHsimd(x+scix+2);
-        iy_S0          = load1DualHsimd(x+sciy);
-        iy_S2          = load1DualHsimd(x+sciy+2);
-        iz_S0          = load1DualHsimd(x+sciz);
-        iz_S2          = load1DualHsimd(x+sciz+2);
+        ix_S0          = loadU1DualHsimd(x+scix);
+        ix_S2          = loadU1DualHsimd(x+scix+2);
+        iy_S0          = loadU1DualHsimd(x+sciy);
+        iy_S2          = loadU1DualHsimd(x+sciy+2);
+        iz_S0          = loadU1DualHsimd(x+sciz);
+        iz_S2          = loadU1DualHsimd(x+sciz+2);
         ix_S0          = ix_S0 + shX_S;
         ix_S2          = ix_S2 + shX_S;
         iy_S0          = iy_S0 + shY_S;
@@ -494,32 +494,32 @@
 
             facel_S    = SimdReal(facel);
 
-            iq_S0      = load1DualHsimd(q+sci);
-            iq_S2      = load1DualHsimd(q+sci+2);
+            iq_S0      = loadU1DualHsimd(q+sci);
+            iq_S2      = loadU1DualHsimd(q+sci+2);
             iq_S0      = facel_S * iq_S0;
             iq_S2      = facel_S * iq_S2;
         }
 
 #ifdef LJ_COMB_LB
-        hsig_i_S0 = load1DualHsimd(ljc+sci2);
-        hsig_i_S2 = load1DualHsimd(ljc+sci2+2);
-        seps_i_S0 = load1DualHsimd(ljc+sci2+STRIDE);
-        seps_i_S2 = load1DualHsimd(ljc+sci2+STRIDE+2);
+        hsig_i_S0 = loadU1DualHsimd(ljc+sci2);
+        hsig_i_S2 = loadU1DualHsimd(ljc+sci2+2);
+        seps_i_S0 = loadU1DualHsimd(ljc+sci2+STRIDE);
+        seps_i_S2 = loadU1DualHsimd(ljc+sci2+STRIDE+2);
 #else
 #ifdef LJ_COMB_GEOM
         SimdReal   c6s_S0, c12s_S0;
         SimdReal   c6s_S2, c12s_S2;
 
-        c6s_S0 = load1DualHsimd(ljc+sci2);
+        c6s_S0 = loadU1DualHsimd(ljc+sci2);
 
         if (!half_LJ)
         {
-            c6s_S2 = load1DualHsimd(ljc+sci2+2);
+            c6s_S2 = loadU1DualHsimd(ljc+sci2+2);
         }
-        c12s_S0 = load1DualHsimd(ljc+sci2+STRIDE);
+        c12s_S0 = loadU1DualHsimd(ljc+sci2+STRIDE);
         if (!half_LJ)
         {
-            c12s_S2 = load1DualHsimd(ljc+sci2+STRIDE+2);
+            c12s_S2 = loadU1DualHsimd(ljc+sci2+STRIDE+2);
         }
 #elif !defined LJ_COMB_LB && !defined FIX_LJ_C
         const real *nbfp0     = nbfp_ptr + type[sci  ]*nbat->ntype*c_simdBestPairAlignment;
@@ -535,10 +535,10 @@
 #ifdef LJ_EWALD_GEOM
         /* We need the geometrically combined C6 for the PME grid correction */
         SimdReal c6s_S0, c6s_S2;
-        c6s_S0 = load1DualHsimd(ljc+sci2);
+        c6s_S0 = loadU1DualHsimd(ljc+sci2);
         if (!half_LJ)
         {
-            c6s_S2 = load1DualHsimd(ljc+sci2+2);
+            c6s_S2 = loadU1DualHsimd(ljc+sci2+2);
         }
 #endif
 

diff --git a/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_prune.cpp b/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_prune.cpp
index 29e22c5615d5d6f0d1db1a251a4ab91ba17f549f..034d5b31dea185e812995e867d44fcc754f30d0e 100644 (file)
--- a/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_prune.cpp
+++ b/src/gromacs/mdlib/nbnxn_kernels/simd_2xnn/nbnxn_kernel_simd_2xnn_prune.cpp
@@ -98,12 +98,12 @@ nbnxn_kernel_prune_2xnn(nbnxn_pairlist_t *         nbl,
         /* Load i atom data */
         int      sciy    = scix + STRIDE;
         int      sciz    = sciy + STRIDE;
-        SimdReal ix_S0   = load1DualHsimd(x + scix    ) + shX_S;
-        SimdReal ix_S2   = load1DualHsimd(x + scix + 2) + shX_S;
-        SimdReal iy_S0   = load1DualHsimd(x + sciy    ) + shY_S;
-        SimdReal iy_S2   = load1DualHsimd(x + sciy + 2) + shY_S;
-        SimdReal iz_S0   = load1DualHsimd(x + sciz    ) + shZ_S;
-        SimdReal iz_S2   = load1DualHsimd(x + sciz + 2) + shZ_S;
+        SimdReal ix_S0   = loadU1DualHsimd(x + scix    ) + shX_S;
+        SimdReal ix_S2   = loadU1DualHsimd(x + scix + 2) + shX_S;
+        SimdReal iy_S0   = loadU1DualHsimd(x + sciy    ) + shY_S;
+        SimdReal iy_S2   = loadU1DualHsimd(x + sciy + 2) + shY_S;
+        SimdReal iz_S0   = loadU1DualHsimd(x + sciz    ) + shZ_S;
+        SimdReal iz_S2   = loadU1DualHsimd(x + sciz + 2) + shZ_S;
 
         for (int cjind = ciEntry->cj_ind_start; cjind < ciEntry->cj_ind_end; cjind++)
         {

diff --git a/src/gromacs/mdlib/nbnxn_search_simd_2xnn.h b/src/gromacs/mdlib/nbnxn_search_simd_2xnn.h
index 733c030f678ee67527c9b7a39a93d9658516ad48..6d1430584f638d3fe3ff09cc84f38879a3d31672 100644 (file)
--- a/src/gromacs/mdlib/nbnxn_search_simd_2xnn.h
+++ b/src/gromacs/mdlib/nbnxn_search_simd_2xnn.h
@@ -53,12 +53,12 @@ icell_set_x_simd_2xnn(int ci,
 
     ia = xIndexFromCi<NbnxnLayout::Simd2xNN>(ci);
 
-    store(x_ci_simd + 0*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 0*STRIDE_S + 0) + SimdReal(shx) );
-    store(x_ci_simd + 1*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 1*STRIDE_S + 0) + SimdReal(shy) );
-    store(x_ci_simd + 2*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 2*STRIDE_S + 0) + SimdReal(shz) );
-    store(x_ci_simd + 3*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 0*STRIDE_S + 2) + SimdReal(shx) );
-    store(x_ci_simd + 4*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 1*STRIDE_S + 2) + SimdReal(shy) );
-    store(x_ci_simd + 5*GMX_SIMD_REAL_WIDTH, load1DualHsimd(x + ia + 2*STRIDE_S + 2) + SimdReal(shz) );
+    store(x_ci_simd + 0*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 0*STRIDE_S + 0) + SimdReal(shx) );
+    store(x_ci_simd + 1*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 1*STRIDE_S + 0) + SimdReal(shy) );
+    store(x_ci_simd + 2*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 2*STRIDE_S + 0) + SimdReal(shz) );
+    store(x_ci_simd + 3*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 0*STRIDE_S + 2) + SimdReal(shx) );
+    store(x_ci_simd + 4*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 1*STRIDE_S + 2) + SimdReal(shy) );
+    store(x_ci_simd + 5*GMX_SIMD_REAL_WIDTH, loadU1DualHsimd(x + ia + 2*STRIDE_S + 2) + SimdReal(shz) );
 }
 
 /* SIMD code for checking and adding cluster-pairs to the list using coordinates in packed format.

diff --git a/src/gromacs/simd/impl_reference/impl_reference_util_double.h b/src/gromacs/simd/impl_reference/impl_reference_util_double.h
index a7256854b52b7f3c70062230b3d4c6dad5090337..e4e4f92b1d17767933115206a5fe41a8c1c0a92b 100644 (file)
--- a/src/gromacs/simd/impl_reference/impl_reference_util_double.h
+++ b/src/gromacs/simd/impl_reference/impl_reference_util_double.h
@@ -1,7 +1,7 @@
 /*
  * This file is part of the GROMACS molecular simulation package.
  *
- * Copyright (c) 2014,2015, by the GROMACS development team, led by
+ * Copyright (c) 2014,2015,2017, 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.
@@ -703,7 +703,7 @@ loadDuplicateHsimd(const double *  m)
  * Available if \ref GMX_SIMD_HAVE_HSIMD_UTIL_DOUBLE is 1.
  */
 static inline SimdDouble gmx_simdcall
-load1DualHsimd(const double *  m)
+loadU1DualHsimd(const double *  m)
 {
     SimdDouble        a;
 

diff --git a/src/gromacs/simd/impl_reference/impl_reference_util_float.h b/src/gromacs/simd/impl_reference/impl_reference_util_float.h
index 2254d7f967c536593dbd0d2fefdf59667f54bd77..6a0f4156e4fd85d1932d68fb59a32a3842ba082b 100644 (file)
--- a/src/gromacs/simd/impl_reference/impl_reference_util_float.h
+++ b/src/gromacs/simd/impl_reference/impl_reference_util_float.h
@@ -1,7 +1,7 @@
 /*
  * This file is part of the GROMACS molecular simulation package.
  *
- * Copyright (c) 2014,2015, by the GROMACS development team, led by
+ * Copyright (c) 2014,2015,2017, 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.
@@ -752,7 +752,7 @@ loadDuplicateHsimd(const float *  m)
  * Available if \ref GMX_SIMD_HAVE_HSIMD_UTIL_FLOAT is 1.
  */
 static inline SimdFloat gmx_simdcall
-load1DualHsimd(const float *  m)
+loadU1DualHsimd(const float *  m)
 {
     SimdFloat        a;
 

diff --git a/src/gromacs/simd/impl_x86_avx_256/impl_x86_avx_256_util_float.h b/src/gromacs/simd/impl_x86_avx_256/impl_x86_avx_256_util_float.h
index 9507128f18db98396a70728670826e786e70b04f..ae81512196377d6334c6a4f3d14133a317377604 100644 (file)
--- a/src/gromacs/simd/impl_x86_avx_256/impl_x86_avx_256_util_float.h
+++ b/src/gromacs/simd/impl_x86_avx_256/impl_x86_avx_256_util_float.h
@@ -586,7 +586,7 @@ loadDuplicateHsimd(const float * m)
 }
 
 static inline SimdFloat gmx_simdcall
-load1DualHsimd(const float * m)
+loadU1DualHsimd(const float * m)
 {
     __m128 t0, t1;
     t0 = _mm_broadcast_ss(m);

diff --git a/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_double.h b/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_double.h
index f65bcf21349f09cd36d63dd0e8f296a70ab6864a..e14c32e3ee6890fd081cb835b2db58522f75fdae 100644 (file)
--- a/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_double.h
+++ b/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_double.h
@@ -320,7 +320,7 @@ loadDuplicateHsimd(const double * m)
 }
 
 static inline SimdDouble gmx_simdcall
-load1DualHsimd(const double * m)
+loadU1DualHsimd(const double * m)
 {
     return {
                _mm512_insertf64x4(_mm512_broadcastsd_pd(_mm_load_sd(m)),

diff --git a/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_float.h b/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_float.h
index 83044fe11c9782a5a28057d1fe7b268dd122e025..991e82cc0b834d143fc4800d3222a5f1d9d5cfac 100644 (file)
--- a/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_float.h
+++ b/src/gromacs/simd/impl_x86_avx_512/impl_x86_avx_512_util_float.h
@@ -360,7 +360,7 @@ loadDuplicateHsimd(const float * m)
 }
 
 static inline SimdFloat gmx_simdcall
-load1DualHsimd(const float * m)
+loadU1DualHsimd(const float * m)
 {
     return {
                _mm512_shuffle_f32x4(_mm512_broadcastss_ps(_mm_load_ss(m)),

diff --git a/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_double.h b/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_double.h
index 476cad836fcce864ed1ec8bd5ae5d2b36a297840..fe93fdbdea31dee8abb12373455cbe0b450ced9e 100644 (file)
--- a/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_double.h
+++ b/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_double.h
@@ -346,7 +346,7 @@ loadDuplicateHsimd(const double * m)
 }
 
 static inline SimdDouble gmx_simdcall
-load1DualHsimd(const double * m)
+loadU1DualHsimd(const double * m)
 {
     return _mm512_mask_extload_pd(_mm512_extload_pd(m, _MM_UPCONV_PD_NONE, _MM_BROADCAST_1X8, _MM_HINT_NONE), _mm512_int2mask(0xF0),
                                   m+1, _MM_UPCONV_PD_NONE, _MM_BROADCAST_1X8, _MM_HINT_NONE);

diff --git a/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_float.h b/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_float.h
index 6912337067cff2c84da0b44e5a8ab3cf6642171e..ad68228dd1dbee3e77618d0b519ad421c67b89e1 100644 (file)
--- a/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_float.h
+++ b/src/gromacs/simd/impl_x86_mic/impl_x86_mic_util_float.h
@@ -347,7 +347,7 @@ loadDuplicateHsimd(const float * m)
 }
 
 static inline SimdFloat gmx_simdcall
-load1DualHsimd(const float * m)
+loadU1DualHsimd(const float * m)
 {
     return _mm512_mask_extload_ps(_mm512_extload_ps(m, _MM_UPCONV_PS_NONE, _MM_BROADCAST_1X16, _MM_HINT_NONE), _mm512_int2mask(0xFF00),
                                   m+1, _MM_UPCONV_PS_NONE, _MM_BROADCAST_1X16, _MM_HINT_NONE);

diff --git a/src/gromacs/simd/tests/simd_floatingpoint_util.cpp b/src/gromacs/simd/tests/simd_floatingpoint_util.cpp
index e9205eab0e55fb0af9e76dd52c6dc219034a9052..623680332cf73c76d1553f46fe67cb574567dd61 100644 (file)
--- a/src/gromacs/simd/tests/simd_floatingpoint_util.cpp
+++ b/src/gromacs/simd/tests/simd_floatingpoint_util.cpp
@@ -724,7 +724,7 @@ TEST_F(SimdFloatingpointUtilTest, loadDuplicateHsimd)
 }
 
 
-TEST_F(SimdFloatingpointUtilTest, load1DualHsimd)
+TEST_F(SimdFloatingpointUtilTest, loadU1DualHsimd)
 {
     SimdReal        v0, v1;
     int             i;
@@ -740,7 +740,7 @@ TEST_F(SimdFloatingpointUtilTest, load1DualHsimd)
     }
 
     v0 = load<SimdReal>(val0_);
-    v1 = load1DualHsimd(data);
+    v1 = loadU1DualHsimd(data);
 
     GMX_EXPECT_SIMD_REAL_EQ(v0, v1);
 }