Make vectorized compute_inverse_size4 compile with AVX.

2 files changed, 25 insertions, 24 deletions

diff --git a/Eigen/LU b/Eigen/LU
index 0fb184bcb..1236ceb04 100644
--- a/Eigen/LU
+++ b/Eigen/LU
@@ -38,9 +38,7 @@
 #include "src/LU/Determinant.h"
 #include "src/LU/InverseImpl.h"
 
-// Use the SSE optimized version whenever possible. At the moment the
-// SSE version doesn't compile when AVX is enabled
-#if (defined EIGEN_VECTORIZE_SSE && !defined EIGEN_VECTORIZE_AVX) || defined EIGEN_VECTORIZE_NEON
+#if defined EIGEN_VECTORIZE_SSE || defined EIGEN_VECTORIZE_NEON
   #include "src/LU/arch/InverseSize4.h"
 #endif
 
diff --git a/Eigen/src/LU/arch/InverseSize4.h b/Eigen/src/LU/arch/InverseSize4.h
index 5a8d0c114..ee5548aed 100644
--- a/Eigen/src/LU/arch/InverseSize4.h
+++ b/Eigen/src/LU/arch/InverseSize4.h
@@ -54,10 +54,12 @@ struct compute_inverse_size4<Architecture::Target, float, MatrixType, ResultType
   {
     ActualMatrixType matrix(mat);
 
-    Packet4f _L1 = matrix.template packet<MatrixAlignment>(0);
-    Packet4f _L2 = matrix.template packet<MatrixAlignment>(4);
-    Packet4f _L3 = matrix.template packet<MatrixAlignment>(8);
-    Packet4f _L4 = matrix.template packet<MatrixAlignment>(12);
+    const float* data = matrix.data();
+    const Index stride = matrix.innerStride();
+    Packet4f _L1 = ploadt<Packet4f,MatrixAlignment>(data);
+    Packet4f _L2 = ploadt<Packet4f,MatrixAlignment>(data + stride*4);
+    Packet4f _L3 = ploadt<Packet4f,MatrixAlignment>(data + stride*8);
+    Packet4f _L4 = ploadt<Packet4f,MatrixAlignment>(data + stride*12);
 
     // Four 2x2 sub-matrices of the input matrix
     // input = [[A, B],
@@ -189,25 +191,26 @@ struct compute_inverse_size4<Architecture::Target, double, MatrixType, ResultTyp
 
     Packet2d A1, A2, B1, B2, C1, C2, D1, D2;
 
+    const double* data = matrix.data();
+    const Index stride = matrix.innerStride();
     if (StorageOrdersMatch)
     {
-      A1 = matrix.template packet<MatrixAlignment>(0);
-      B1 = matrix.template packet<MatrixAlignment>(2);
-      A2 = matrix.template packet<MatrixAlignment>(4);
-      B2 = matrix.template packet<MatrixAlignment>(6);
-      C1 = matrix.template packet<MatrixAlignment>(8);
-      D1 = matrix.template packet<MatrixAlignment>(10);
-      C2 = matrix.template packet<MatrixAlignment>(12);
-      D2 = matrix.template packet<MatrixAlignment>(14);
+      A1 = ploadt<Packet2d,MatrixAlignment>(data + stride*0);
+      B1 = ploadt<Packet2d,MatrixAlignment>(data + stride*2);
+      A2 = ploadt<Packet2d,MatrixAlignment>(data + stride*4);
+      B2 = ploadt<Packet2d,MatrixAlignment>(data + stride*6);
+      C1 = ploadt<Packet2d,MatrixAlignment>(data + stride*8);
+      D1 = ploadt<Packet2d,MatrixAlignment>(data + stride*10);
+      C2 = ploadt<Packet2d,MatrixAlignment>(data + stride*12);
+      D2 = ploadt<Packet2d,MatrixAlignment>(data + stride*14);
     }
     else
     {
       Packet2d temp;
-      A1 = matrix.template packet<MatrixAlignment>(0);
-      C1 = matrix.template packet<MatrixAlignment>(2);
-      A2 = matrix.template packet<MatrixAlignment>(4);
-      C2 = matrix.template packet<MatrixAlignment>(6);
-
+      A1 = ploadt<Packet2d,MatrixAlignment>(data + stride*0);
+      C1 = ploadt<Packet2d,MatrixAlignment>(data + stride*2);
+      A2 = ploadt<Packet2d,MatrixAlignment>(data + stride*4);
+      C2 = ploadt<Packet2d,MatrixAlignment>(data + stride*6);
       temp = A1;
       A1 = vec2d_unpacklo(A1, A2);
       A2 = vec2d_unpackhi(temp, A2);
@@ -216,10 +219,10 @@ struct compute_inverse_size4<Architecture::Target, double, MatrixType, ResultTyp
       C1 = vec2d_unpacklo(C1, C2);
       C2 = vec2d_unpackhi(temp, C2);
 
-      B1 = matrix.template packet<MatrixAlignment>(8);
-      D1 = matrix.template packet<MatrixAlignment>(10);
-      B2 = matrix.template packet<MatrixAlignment>(12);
-      D2 = matrix.template packet<MatrixAlignment>(14);
+      B1 = ploadt<Packet2d,MatrixAlignment>(data + stride*8);
+      D1 = ploadt<Packet2d,MatrixAlignment>(data + stride*10);
+      B2 = ploadt<Packet2d,MatrixAlignment>(data + stride*12);
+      D2 = ploadt<Packet2d,MatrixAlignment>(data + stride*14);
 
       temp = B1;
       B1 = vec2d_unpacklo(B1, B2);