Merge pull request #2721 from martin-frbg/p8align

Fix alignment errors in the power8 saxpy kernel

1 files changed, 93 insertions, 3 deletions

diff --git a/kernel/power/saxpy.c b/kernel/power/saxpy.c
index d005427b5..3d3b1613c 100644
--- a/kernel/power/saxpy.c
+++ b/kernel/power/saxpy.c
@@ -28,6 +28,22 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include "common.h"
 
+#define  offset_0 0
+#define  offset_1 16
+#define  offset_2 32
+#define  offset_3 48
+#define  offset_4 64
+#define  offset_5 80
+#define  offset_6 96
+#define  offset_7 112
+#define  offset_8 128
+#define  offset_9 144
+#define  offset_10 160
+#define  offset_11 176
+#define  offset_12 192
+#define  offset_13 208
+#define  offset_14 224
+#define  offset_15 240
  
 
 #if defined(__VEC__) || defined(__ALTIVEC__)
@@ -38,12 +54,85 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 static void saxpy_kernel_64(BLASLONG n, FLOAT *x, FLOAT *y, FLOAT alpha)
 {
     BLASLONG  i = 0;
-    __vector float v_a = {alpha,alpha,alpha,alpha}; 
-    __vector float * v_y=(__vector float *)y;
-    __vector float * v_x=(__vector float *)x;
+    __vector float v_a __attribute((aligned(16))) = {alpha,alpha,alpha,alpha}; 
+    __vector float * vptr_y =(__vector float *)y;
+    __vector float * vptr_x =(__vector float *)x;
         
     for(; i<n/4; i+=16){
 
+
+        register __vector float vy_0 = vec_vsx_ld( offset_0 ,vptr_y ) ;
+        register __vector float vy_1 = vec_vsx_ld( offset_1 ,vptr_y ) ;
+        register __vector float vy_2 = vec_vsx_ld( offset_2 ,vptr_y ) ;
+        register __vector float vy_3 = vec_vsx_ld( offset_3 ,vptr_y ) ;
+        register __vector float vy_4 = vec_vsx_ld( offset_4 ,vptr_y ) ;
+        register __vector float vy_5 = vec_vsx_ld( offset_5 ,vptr_y ) ;
+        register __vector float vy_6 = vec_vsx_ld( offset_6 ,vptr_y ) ;
+        register __vector float vy_7 = vec_vsx_ld( offset_7 ,vptr_y ) ;
+        register __vector float vy_8 = vec_vsx_ld( offset_8 ,vptr_y ) ;
+        register __vector float vy_9 = vec_vsx_ld( offset_9 ,vptr_y ) ;
+        register __vector float vy_10 = vec_vsx_ld( offset_10 ,vptr_y ) ;
+        register __vector float vy_11 = vec_vsx_ld( offset_11 ,vptr_y ) ;
+        register __vector float vy_12 = vec_vsx_ld( offset_12 ,vptr_y ) ;
+        register __vector float vy_13 = vec_vsx_ld( offset_13 ,vptr_y ) ;
+        register __vector float vy_14 = vec_vsx_ld( offset_14 ,vptr_y ) ;
+        register __vector float vy_15 = vec_vsx_ld( offset_15 ,vptr_y ) ;
+
+        register __vector float vx_0 = vec_vsx_ld( offset_0 ,vptr_x ) ;
+        register __vector float vx_1 = vec_vsx_ld( offset_1 ,vptr_x ) ;
+        register __vector float vx_2 = vec_vsx_ld( offset_2 ,vptr_x ) ;
+        register __vector float vx_3 = vec_vsx_ld( offset_3 ,vptr_x ) ;
+        register __vector float vx_4 = vec_vsx_ld( offset_4 ,vptr_x ) ;
+        register __vector float vx_5 = vec_vsx_ld( offset_5 ,vptr_x ) ;
+        register __vector float vx_6 = vec_vsx_ld( offset_6 ,vptr_x ) ;
+        register __vector float vx_7 = vec_vsx_ld( offset_7 ,vptr_x ) ;
+        register __vector float vx_8 = vec_vsx_ld( offset_8 ,vptr_x ) ;
+        register __vector float vx_9 = vec_vsx_ld( offset_9 ,vptr_x ) ;
+        register __vector float vx_10 = vec_vsx_ld( offset_10 ,vptr_x ) ;
+        register __vector float vx_11 = vec_vsx_ld( offset_11 ,vptr_x ) ;
+        register __vector float vx_12 = vec_vsx_ld( offset_12 ,vptr_x ) ;
+        register __vector float vx_13 = vec_vsx_ld( offset_13 ,vptr_x ) ;
+        register __vector float vx_14 = vec_vsx_ld( offset_14 ,vptr_x ) ;
+        register __vector float vx_15 = vec_vsx_ld( offset_15 ,vptr_x ) ;
+        vy_0 += vx_0*v_a;
+        vy_1 += vx_1*v_a;
+        vy_2 += vx_2*v_a;
+        vy_3 += vx_3*v_a;
+        vy_4 += vx_4*v_a;
+        vy_5 += vx_5*v_a;
+        vy_6 += vx_6*v_a;
+        vy_7 += vx_7*v_a;
+        vy_8 += vx_8*v_a;
+        vy_9 += vx_9*v_a;
+        vy_10 += vx_10*v_a;
+        vy_11 += vx_11*v_a;
+        vy_12 += vx_12*v_a;
+        vy_13 += vx_13*v_a;
+        vy_14 += vx_14*v_a;
+        vy_15 += vx_15*v_a;
+
+    	vec_vsx_st( vy_0, offset_0 ,vptr_y ) ;
+        vec_vsx_st( vy_1, offset_1 ,vptr_y ) ;
+        vec_vsx_st( vy_2, offset_2 ,vptr_y ) ;
+        vec_vsx_st( vy_3, offset_3 ,vptr_y ) ;
+        vec_vsx_st( vy_4, offset_4 ,vptr_y ) ;
+        vec_vsx_st( vy_5, offset_5 ,vptr_y ) ;
+        vec_vsx_st( vy_6, offset_6 ,vptr_y ) ;
+	vec_vsx_st( vy_7, offset_7 ,vptr_y ) ; 
+    	vec_vsx_st( vy_8, offset_8 ,vptr_y ) ;
+        vec_vsx_st( vy_9, offset_9 ,vptr_y ) ;
+        vec_vsx_st( vy_10, offset_10 ,vptr_y ) ;
+        vec_vsx_st( vy_11, offset_11 ,vptr_y ) ;
+        vec_vsx_st( vy_12, offset_12 ,vptr_y ) ;
+        vec_vsx_st( vy_13, offset_13 ,vptr_y ) ;
+        vec_vsx_st( vy_14, offset_14 ,vptr_y ) ;
+	vec_vsx_st( vy_15, offset_15 ,vptr_y ) ; 
+
+        vptr_x+=16;
+	vptr_y+=16; 
+
+/*
+
         v_y[i]    += v_a * v_x[i];
         v_y[i+1]  += v_a * v_x[i+1];
         v_y[i+2]  += v_a * v_x[i+2];
@@ -60,6 +149,7 @@ static void saxpy_kernel_64(BLASLONG n, FLOAT *x, FLOAT *y, FLOAT alpha)
         v_y[i+13] += v_a * v_x[i+13];
         v_y[i+14] += v_a * v_x[i+14];
         v_y[i+15] += v_a * v_x[i+15];
+*/
     }
 }
 #endif