diff mbox series

[libfortran] correct buffer size in library matmul

Message ID e0caea30-79f3-a3b2-36a6-ad856caacbb9@netcologne.de
State New
Headers show
Series [libfortran] correct buffer size in library matmul | expand

Commit Message

Thomas Koenig Aug. 25, 2018, 7:38 p.m. UTC 
Hello world,

the attached patch fixes a regression where the calculation of the size
of the buffer for matmul was too small for ine special case.

Regression-tested. OK for trunk?

Regards

	Thomas

2018-08-25  Thomas Koenig  <tkoenig@gcc.gnu.org>

	PR libfortran/86704
	* m4/matmul_internal.m4: Correct calculation of needed buffer size
	for arrays of shape (1,n).
	* generated/matmul_c10.c: Regenerated
	* generated/matmul_c16.c: Regenerated
	* generated/matmul_c4.c: Regenerated
	* generated/matmul_c8.c: Regenerated
	* generated/matmul_i1.c: Regenerated
	* generated/matmul_i16.c: Regenerated
	* generated/matmul_i2.c: Regenerated
	* generated/matmul_i4.c: Regenerated
	* generated/matmul_i8.c: Regenerated
	* generated/matmul_r10.c: Regenerated
	* generated/matmul_r16.c: Regenerated
	* generated/matmul_r4.c: Regenerated
	* generated/matmul_r8.c: Regenerated
	* generated/matmulavx128_c10.c: Regenerated
	* generated/matmulavx128_c16.c: Regenerated
	* generated/matmulavx128_c4.c: Regenerated
	* generated/matmulavx128_c8.c: Regenerated
	* generated/matmulavx128_i1.c: Regenerated
	* generated/matmulavx128_i16.c: Regenerated
	* generated/matmulavx128_i2.c: Regenerated
	* generated/matmulavx128_i4.c: Regenerated
	* generated/matmulavx128_i8.c: Regenerated
	* generated/matmulavx128_r10.c: Regenerated
	* generated/matmulavx128_r16.c: Regenerated
	* generated/matmulavx128_r4.c: Regenerated
	* generated/matmulavx128_r8.c: Regenerated

2018-08-25  Thomas Koenig  <tkoenig@gcc.gnu.org>

	PR libfortran/86704
	* gfortran.dg/matmul_19.f90: New test.

Comments

Jerry DeLisle Aug. 25, 2018, 9:23 p.m. UTC | #1 
On 08/25/2018 12:38 PM, Thomas Koenig wrote:
> Hello world,
> 
> the attached patch fixes a regression where the calculation of the size
> of the buffer for matmul was too small for ine special case.
> 
> Regression-tested. OK for trunk?
> 
> Regards
> 
>      Thomas

OK Thomas and thanks for the fix.

Jerry
diff mbox series

Patch

Index: m4/matmul_internal.m4
===================================================================
--- m4/matmul_internal.m4	(Revision 263752)
+++ m4/matmul_internal.m4	(Arbeitskopie)
@@ -233,8 +233,13 @@  sinclude(`matmul_asm_'rtype_code`.m4')dnl
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_c10.c
===================================================================
--- generated/matmul_c10.c	(Revision 263752)
+++ generated/matmul_c10.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_c10_avx (gfc_array_c10 * const restrict ret
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_c10_avx2 (gfc_array_c10 * const restrict re
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_c10_avx512f (gfc_array_c10 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_c10_vanilla (gfc_array_c10 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_c10 (gfc_array_c10 * const restrict retarra
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_c16.c
===================================================================
--- generated/matmul_c16.c	(Revision 263752)
+++ generated/matmul_c16.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_c16_avx (gfc_array_c16 * const restrict ret
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_c16_avx2 (gfc_array_c16 * const restrict re
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_c16_avx512f (gfc_array_c16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_c16_vanilla (gfc_array_c16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_c16 (gfc_array_c16 * const restrict retarra
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_c4.c
===================================================================
--- generated/matmul_c4.c	(Revision 263752)
+++ generated/matmul_c4.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_c4_avx (gfc_array_c4 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_c4_avx2 (gfc_array_c4 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_c4_avx512f (gfc_array_c4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_c4_vanilla (gfc_array_c4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_c4 (gfc_array_c4 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_c8.c
===================================================================
--- generated/matmul_c8.c	(Revision 263752)
+++ generated/matmul_c8.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_c8_avx (gfc_array_c8 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_c8_avx2 (gfc_array_c8 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_c8_avx512f (gfc_array_c8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_c8_vanilla (gfc_array_c8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_c8 (gfc_array_c8 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_i1.c
===================================================================
--- generated/matmul_i1.c	(Revision 263752)
+++ generated/matmul_i1.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_i1_avx (gfc_array_i1 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_i1_avx2 (gfc_array_i1 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_i1_avx512f (gfc_array_i1 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_i1_vanilla (gfc_array_i1 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_i1 (gfc_array_i1 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_i16.c
===================================================================
--- generated/matmul_i16.c	(Revision 263752)
+++ generated/matmul_i16.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_i16_avx (gfc_array_i16 * const restrict ret
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_i16_avx2 (gfc_array_i16 * const restrict re
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_i16_avx512f (gfc_array_i16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_i16_vanilla (gfc_array_i16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_i16 (gfc_array_i16 * const restrict retarra
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_i2.c
===================================================================
--- generated/matmul_i2.c	(Revision 263752)
+++ generated/matmul_i2.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_i2_avx (gfc_array_i2 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_i2_avx2 (gfc_array_i2 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_i2_avx512f (gfc_array_i2 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_i2_vanilla (gfc_array_i2 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_i2 (gfc_array_i2 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_i4.c
===================================================================
--- generated/matmul_i4.c	(Revision 263752)
+++ generated/matmul_i4.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_i4_avx (gfc_array_i4 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_i4_avx2 (gfc_array_i4 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_i4_avx512f (gfc_array_i4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_i4_vanilla (gfc_array_i4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_i4 (gfc_array_i4 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_i8.c
===================================================================
--- generated/matmul_i8.c	(Revision 263752)
+++ generated/matmul_i8.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_i8_avx (gfc_array_i8 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_i8_avx2 (gfc_array_i8 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_i8_avx512f (gfc_array_i8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_i8_vanilla (gfc_array_i8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_i8 (gfc_array_i8 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_r10.c
===================================================================
--- generated/matmul_r10.c	(Revision 263752)
+++ generated/matmul_r10.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_r10_avx (gfc_array_r10 * const restrict ret
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_r10_avx2 (gfc_array_r10 * const restrict re
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_r10_avx512f (gfc_array_r10 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_r10_vanilla (gfc_array_r10 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_r10 (gfc_array_r10 * const restrict retarra
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_r16.c
===================================================================
--- generated/matmul_r16.c	(Revision 263752)
+++ generated/matmul_r16.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_r16_avx (gfc_array_r16 * const restrict ret
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_r16_avx2 (gfc_array_r16 * const restrict re
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_r16_avx512f (gfc_array_r16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_r16_vanilla (gfc_array_r16 * const restrict
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_r16 (gfc_array_r16 * const restrict retarra
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_r4.c
===================================================================
--- generated/matmul_r4.c	(Revision 263752)
+++ generated/matmul_r4.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_r4_avx (gfc_array_r4 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_r4_avx2 (gfc_array_r4 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_r4_avx512f (gfc_array_r4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_r4_vanilla (gfc_array_r4 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_r4 (gfc_array_r4 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmul_r8.c
===================================================================
--- generated/matmul_r8.c	(Revision 263752)
+++ generated/matmul_r8.c	(Arbeitskopie)
@@ -317,8 +317,13 @@  matmul_r8_avx (gfc_array_r8 * const restrict retar
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -869,8 +874,13 @@  matmul_r8_avx2 (gfc_array_r8 * const restrict reta
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1421,8 +1431,13 @@  matmul_r8_avx512f (gfc_array_r8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -1987,8 +2002,13 @@  matmul_r8_vanilla (gfc_array_r8 * const restrict r
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -2613,8 +2633,13 @@  matmul_r8 (gfc_array_r8 * const restrict retarray,
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_c10.c
===================================================================
--- generated/matmulavx128_c10.c	(Revision 263752)
+++ generated/matmulavx128_c10.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_c10_avx128_fma3 (gfc_array_c10 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_c10_avx128_fma4 (gfc_array_c10 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_c16.c
===================================================================
--- generated/matmulavx128_c16.c	(Revision 263752)
+++ generated/matmulavx128_c16.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_c16_avx128_fma3 (gfc_array_c16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_c16_avx128_fma4 (gfc_array_c16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_c4.c
===================================================================
--- generated/matmulavx128_c4.c	(Revision 263752)
+++ generated/matmulavx128_c4.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_c4_avx128_fma3 (gfc_array_c4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_c4_avx128_fma4 (gfc_array_c4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_c8.c
===================================================================
--- generated/matmulavx128_c8.c	(Revision 263752)
+++ generated/matmulavx128_c8.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_c8_avx128_fma3 (gfc_array_c8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_c8_avx128_fma4 (gfc_array_c8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_i1.c
===================================================================
--- generated/matmulavx128_i1.c	(Revision 263752)
+++ generated/matmulavx128_i1.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_i1_avx128_fma3 (gfc_array_i1 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_i1_avx128_fma4 (gfc_array_i1 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_i16.c
===================================================================
--- generated/matmulavx128_i16.c	(Revision 263752)
+++ generated/matmulavx128_i16.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_i16_avx128_fma3 (gfc_array_i16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_i16_avx128_fma4 (gfc_array_i16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_i2.c
===================================================================
--- generated/matmulavx128_i2.c	(Revision 263752)
+++ generated/matmulavx128_i2.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_i2_avx128_fma3 (gfc_array_i2 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_i2_avx128_fma4 (gfc_array_i2 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_i4.c
===================================================================
--- generated/matmulavx128_i4.c	(Revision 263752)
+++ generated/matmulavx128_i4.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_i4_avx128_fma3 (gfc_array_i4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_i4_avx128_fma4 (gfc_array_i4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_i8.c
===================================================================
--- generated/matmulavx128_i8.c	(Revision 263752)
+++ generated/matmulavx128_i8.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_i8_avx128_fma3 (gfc_array_i8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_i8_avx128_fma4 (gfc_array_i8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_r10.c
===================================================================
--- generated/matmulavx128_r10.c	(Revision 263752)
+++ generated/matmulavx128_r10.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_r10_avx128_fma3 (gfc_array_r10 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_r10_avx128_fma4 (gfc_array_r10 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_r16.c
===================================================================
--- generated/matmulavx128_r16.c	(Revision 263752)
+++ generated/matmulavx128_r16.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_r16_avx128_fma3 (gfc_array_r16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_r16_avx128_fma4 (gfc_array_r16 * const rest
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_r4.c
===================================================================
--- generated/matmulavx128_r4.c	(Revision 263752)
+++ generated/matmulavx128_r4.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_r4_avx128_fma3 (gfc_array_r4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_r4_avx128_fma4 (gfc_array_r4 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
Index: generated/matmulavx128_r8.c
===================================================================
--- generated/matmulavx128_r8.c	(Revision 263752)
+++ generated/matmulavx128_r8.c	(Arbeitskopie)
@@ -282,8 +282,13 @@  matmul_r8_avx128_fma3 (gfc_array_r8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;
 
@@ -835,8 +840,13 @@  matmul_r8_avx128_fma4 (gfc_array_r8 * const restri
 	return;
 
       /* Adjust size of t1 to what is needed.  */
-      index_type t1_dim;
-      t1_dim = (a_dim1 - (ycount > 1)) * 256 + b_dim1;
+      index_type t1_dim, a_sz;
+      if (aystride == 1)
+        a_sz = rystride;
+      else
+        a_sz = a_dim1;
+
+      t1_dim = a_sz * 256 + b_dim1;
       if (t1_dim > 65536)
 	t1_dim = 65536;