biod.pnpi.spb.ru / alexxy / gromacs.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
Apply clang-format to source tree
[alexxy/gromacs.git] / src / gromacs / ewald / pme_simd4.h
diff --git a/src/gromacs/ewald/pme_simd4.h b/src/gromacs/ewald/pme_simd4.h
index 0fdc3170b1ed147e36c8ceaf536f7ab899ae3143..8b948c2fab07fbbb414f02e70fa7e82df64b3688 100644 (file)
--- a/src/gromacs/ewald/pme_simd4.h
+++ b/src/gromacs/ewald/pme_simd4.h
@@ -57,34 +57,34 @@
     Simd4Real gri_S0, gri_S1, gri_S2, gri_S3;
 
     /* With order 4 the z-spline is actually aligned */
-    tz_S  = load4(thz);
+    tz_S = load4(thz);
 
     for (ithx = 0; (ithx < 4); ithx++)
     {
-        index_x = (i0+ithx)*pny*pnz;
-        valx    = coefficient*thx[ithx];
+        index_x = (i0 + ithx) * pny * pnz;
+        valx    = coefficient * thx[ithx];
 
-        vx_S   = Simd4Real(valx);
+        vx_S = Simd4Real(valx);
 
         vx_tz_S = vx_S * tz_S;
 
-        gri_S0 = load4U(grid+index_x+(j0+0)*pnz+k0);
-        gri_S1 = load4U(grid+index_x+(j0+1)*pnz+k0);
-        gri_S2 = load4U(grid+index_x+(j0+2)*pnz+k0);
-        gri_S3 = load4U(grid+index_x+(j0+3)*pnz+k0);
+        gri_S0 = load4U(grid + index_x + (j0 + 0) * pnz + k0);
+        gri_S1 = load4U(grid + index_x + (j0 + 1) * pnz + k0);
+        gri_S2 = load4U(grid + index_x + (j0 + 2) * pnz + k0);
+        gri_S3 = load4U(grid + index_x + (j0 + 3) * pnz + k0);
 
         sum_S0 = fma(vx_tz_S, ty_S0, gri_S0);
         sum_S1 = fma(vx_tz_S, ty_S1, gri_S1);
         sum_S2 = fma(vx_tz_S, ty_S2, gri_S2);
         sum_S3 = fma(vx_tz_S, ty_S3, gri_S3);
 
-        store4U(grid+index_x+(j0+0)*pnz+k0, sum_S0);
-        store4U(grid+index_x+(j0+1)*pnz+k0, sum_S1);
-        store4U(grid+index_x+(j0+2)*pnz+k0, sum_S2);
-        store4U(grid+index_x+(j0+3)*pnz+k0, sum_S3);
+        store4U(grid + index_x + (j0 + 0) * pnz + k0, sum_S0);
+        store4U(grid + index_x + (j0 + 1) * pnz + k0, sum_S1);
+        store4U(grid + index_x + (j0 + 2) * pnz + k0, sum_S2);
+        store4U(grid + index_x + (j0 + 3) * pnz + k0, sum_S3);
     }
 }
-#undef PME_SPREAD_SIMD4_ORDER4
+#    undef PME_SPREAD_SIMD4_ORDER4
 #endif
 
 
@@ -95,105 +95,105 @@
  */
 {
     using namespace gmx;
-    int              offset;
-    int              index;
-    Simd4Real        ty_S0(thy[0]);
-    Simd4Real        ty_S1(thy[1]);
-    Simd4Real        ty_S2(thy[2]);
-    Simd4Real        ty_S3(thy[3]);
-    Simd4Real        tz_S0;
-    Simd4Real        tz_S1;
-    Simd4Real        vx_S;
-    Simd4Real        vx_tz_S0;
-    Simd4Real        vx_tz_S1;
-    Simd4Real        sum_S00, sum_S01, sum_S02, sum_S03;
-    Simd4Real        sum_S10, sum_S11, sum_S12, sum_S13;
-    Simd4Real        gri_S00, gri_S01, gri_S02, gri_S03;
-    Simd4Real        gri_S10, gri_S11, gri_S12, gri_S13;
-#if PME_ORDER == 5
-    Simd4Real        ty_S4(thy[4]);
-    Simd4Real        sum_S04;
-    Simd4Real        sum_S14;
-    Simd4Real        gri_S04;
-    Simd4Real        gri_S14;
-#endif
+    int       offset;
+    int       index;
+    Simd4Real ty_S0(thy[0]);
+    Simd4Real ty_S1(thy[1]);
+    Simd4Real ty_S2(thy[2]);
+    Simd4Real ty_S3(thy[3]);
+    Simd4Real tz_S0;
+    Simd4Real tz_S1;
+    Simd4Real vx_S;
+    Simd4Real vx_tz_S0;
+    Simd4Real vx_tz_S1;
+    Simd4Real sum_S00, sum_S01, sum_S02, sum_S03;
+    Simd4Real sum_S10, sum_S11, sum_S12, sum_S13;
+    Simd4Real gri_S00, gri_S01, gri_S02, gri_S03;
+    Simd4Real gri_S10, gri_S11, gri_S12, gri_S13;
+#    if PME_ORDER == 5
+    Simd4Real ty_S4(thy[4]);
+    Simd4Real sum_S04;
+    Simd4Real sum_S14;
+    Simd4Real gri_S04;
+    Simd4Real gri_S14;
+#    endif
 
     offset = k0 & 3;
 
-#ifdef PME_SIMD4_UNALIGNED
-    tz_S0 = load4U(thz-offset);
-    tz_S1 = load4U(thz-offset+4);
-#else
+#    ifdef PME_SIMD4_UNALIGNED
+    tz_S0 = load4U(thz - offset);
+    tz_S1 = load4U(thz - offset + 4);
+#    else
     {
         int i;
         /* Copy thz to an aligned buffer (unused buffer parts are masked) */
         for (i = 0; i < PME_ORDER; i++)
         {
-            thz_aligned[offset+i] = thz[i];
+            thz_aligned[offset + i] = thz[i];
         }
         tz_S0 = load4(thz_aligned);
-        tz_S1 = load4(thz_aligned+4);
+        tz_S1 = load4(thz_aligned + 4);
     }
-#endif
+#    endif
     tz_S0 = selectByMask(tz_S0, work->mask_S0[offset]);
     tz_S1 = selectByMask(tz_S1, work->mask_S1[offset]);
 
     for (ithx = 0; (ithx < PME_ORDER); ithx++)
     {
-        index = (i0+ithx)*pny*pnz + j0*pnz + k0 - offset;
-        valx  = coefficient*thx[ithx];
+        index = (i0 + ithx) * pny * pnz + j0 * pnz + k0 - offset;
+        valx  = coefficient * thx[ithx];
 
-        vx_S   = Simd4Real(valx);
+        vx_S = Simd4Real(valx);
 
         vx_tz_S0 = vx_S * tz_S0;
         vx_tz_S1 = vx_S * tz_S1;
 
-        gri_S00 = load4(grid+index+0*pnz);
-        gri_S01 = load4(grid+index+1*pnz);
-        gri_S02 = load4(grid+index+2*pnz);
-        gri_S03 = load4(grid+index+3*pnz);
-#if PME_ORDER == 5
-        gri_S04 = load4(grid+index+4*pnz);
-#endif
-        gri_S10 = load4(grid+index+0*pnz+4);
-        gri_S11 = load4(grid+index+1*pnz+4);
-        gri_S12 = load4(grid+index+2*pnz+4);
-        gri_S13 = load4(grid+index+3*pnz+4);
-#if PME_ORDER == 5
-        gri_S14 = load4(grid+index+4*pnz+4);
-#endif
+        gri_S00 = load4(grid + index + 0 * pnz);
+        gri_S01 = load4(grid + index + 1 * pnz);
+        gri_S02 = load4(grid + index + 2 * pnz);
+        gri_S03 = load4(grid + index + 3 * pnz);
+#    if PME_ORDER == 5
+        gri_S04 = load4(grid + index + 4 * pnz);
+#    endif
+        gri_S10 = load4(grid + index + 0 * pnz + 4);
+        gri_S11 = load4(grid + index + 1 * pnz + 4);
+        gri_S12 = load4(grid + index + 2 * pnz + 4);
+        gri_S13 = load4(grid + index + 3 * pnz + 4);
+#    if PME_ORDER == 5
+        gri_S14 = load4(grid + index + 4 * pnz + 4);
+#    endif
 
         sum_S00 = fma(vx_tz_S0, ty_S0, gri_S00);
         sum_S01 = fma(vx_tz_S0, ty_S1, gri_S01);
         sum_S02 = fma(vx_tz_S0, ty_S2, gri_S02);
         sum_S03 = fma(vx_tz_S0, ty_S3, gri_S03);
-#if PME_ORDER == 5
+#    if PME_ORDER == 5
         sum_S04 = fma(vx_tz_S0, ty_S4, gri_S04);
-#endif
+#    endif
         sum_S10 = fma(vx_tz_S1, ty_S0, gri_S10);
         sum_S11 = fma(vx_tz_S1, ty_S1, gri_S11);
         sum_S12 = fma(vx_tz_S1, ty_S2, gri_S12);
         sum_S13 = fma(vx_tz_S1, ty_S3, gri_S13);
-#if PME_ORDER == 5
+#    if PME_ORDER == 5
         sum_S14 = fma(vx_tz_S1, ty_S4, gri_S14);
-#endif
-
-        store4(grid+index+0*pnz, sum_S00);
-        store4(grid+index+1*pnz, sum_S01);
-        store4(grid+index+2*pnz, sum_S02);
-        store4(grid+index+3*pnz, sum_S03);
-#if PME_ORDER == 5
-        store4(grid+index+4*pnz, sum_S04);
-#endif
-        store4(grid+index+0*pnz+4, sum_S10);
-        store4(grid+index+1*pnz+4, sum_S11);
-        store4(grid+index+2*pnz+4, sum_S12);
-        store4(grid+index+3*pnz+4, sum_S13);
-#if PME_ORDER == 5
-        store4(grid+index+4*pnz+4, sum_S14);
-#endif
+#    endif
+
+        store4(grid + index + 0 * pnz, sum_S00);
+        store4(grid + index + 1 * pnz, sum_S01);
+        store4(grid + index + 2 * pnz, sum_S02);
+        store4(grid + index + 3 * pnz, sum_S03);
+#    if PME_ORDER == 5
+        store4(grid + index + 4 * pnz, sum_S04);
+#    endif
+        store4(grid + index + 0 * pnz + 4, sum_S10);
+        store4(grid + index + 1 * pnz + 4, sum_S11);
+        store4(grid + index + 2 * pnz + 4, sum_S12);
+        store4(grid + index + 3 * pnz + 4, sum_S13);
+#    if PME_ORDER == 5
+        store4(grid + index + 4 * pnz + 4, sum_S14);
+#    endif
     }
 }
-#undef PME_ORDER
-#undef PME_SPREAD_SIMD4_ALIGNED
+#    undef PME_ORDER
+#    undef PME_SPREAD_SIMD4_ALIGNED
 #endif
Local GROMACS mirror with custom stuff