Measure variant 4 and 5 of HGCD2_DIV1_METHOD.

6 files changed, 109 insertions, 12 deletions

diff --git a/tune/Makefile.am b/tune/Makefile.am
index 7a8f04ac7..9e382567c 100644
--- a/tune/Makefile.am
+++ b/tune/Makefile.am
@@ -58,7 +58,7 @@ libspeed_la_SOURCES =							\
   gcdext_single.c gcdext_double.c gcdextod.c gcdextos.c			\
   hgcd_lehmer.c hgcd_appr_lehmer.c hgcd_reduce_1.c hgcd_reduce_2.c	\
   jacbase1.c jacbase2.c jacbase3.c jacbase4.c				\
-  hgcd2-1.c hgcd2-2.c hgcd2-3.c						\
+  hgcd2-1.c hgcd2-2.c hgcd2-3.c hgcd2-4.c hgcd2-5.c			\
   mod_1_div.c mod_1_inv.c mod_1_1-1.c mod_1_1-2.c modlinv.c		\
   noop.c powm_mod.c powm_redc.c pre_divrem_1.c				\
   set_strb.c set_strs.c set_strp.c time.c
diff --git a/tune/common.c b/tune/common.c
index ec7168175..6f4332700 100644
--- a/tune/common.c
+++ b/tune/common.c
@@ -1653,6 +1653,16 @@ speed_mpn_hgcd2_3 (struct speed_params *s)
 {
   SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2_3);
 }
+double
+speed_mpn_hgcd2_4 (struct speed_params *s)
+{
+  SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2_4);
+}
+double
+speed_mpn_hgcd2_5 (struct speed_params *s)
+{
+  SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2_5);
+}
 
 double
 speed_mpn_hgcd (struct speed_params *s)
diff --git a/tune/hgcd2-4.c b/tune/hgcd2-4.c
new file mode 100644
index 000000000..ec7f92753
--- /dev/null
+++ b/tune/hgcd2-4.c
@@ -0,0 +1,39 @@
+/* mpn/generic/hgcd2.c method 4.
+
+Copyright 2019 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of either:
+
+  * the GNU Lesser General Public License as published by the Free
+    Software Foundation; either version 3 of the License, or (at your
+    option) any later version.
+
+or
+
+  * the GNU General Public License as published by the Free Software
+    Foundation; either version 2 of the License, or (at your option) any
+    later version.
+
+or both in parallel, as here.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received copies of the GNU General Public License and the
+GNU Lesser General Public License along with the GNU MP Library.  If not,
+see https://www.gnu.org/licenses/.  */
+
+#include "gmp-impl.h"
+
+#undef HGCD2_DIV1_METHOD
+#define HGCD2_DIV1_METHOD 4
+#define __gmpn_hgcd2 mpn_hgcd2_4
+/* Not used, but renamed to not get duplicate definitions */
+#define __gmpn_hgcd_mul_matrix1_vector mpn_hgcd_mul_matrix1_vector_4
+
+#include "mpn/generic/hgcd2.c"
diff --git a/tune/hgcd2-5.c b/tune/hgcd2-5.c
new file mode 100644
index 000000000..ed661716e
--- /dev/null
+++ b/tune/hgcd2-5.c
@@ -0,0 +1,39 @@
+/* mpn/generic/hgcd2.c method 5.
+
+Copyright 2019 Free Software Foundation, Inc.
+
+This file is part of the GNU MP Library.
+
+The GNU MP Library is free software; you can redistribute it and/or modify
+it under the terms of either:
+
+  * the GNU Lesser General Public License as published by the Free
+    Software Foundation; either version 3 of the License, or (at your
+    option) any later version.
+
+or
+
+  * the GNU General Public License as published by the Free Software
+    Foundation; either version 2 of the License, or (at your option) any
+    later version.
+
+or both in parallel, as here.
+
+The GNU MP Library is distributed in the hope that it will be useful, but
+WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received copies of the GNU General Public License and the
+GNU Lesser General Public License along with the GNU MP Library.  If not,
+see https://www.gnu.org/licenses/.  */
+
+#include "gmp-impl.h"
+
+#undef HGCD2_DIV1_METHOD
+#define HGCD2_DIV1_METHOD 5
+#define __gmpn_hgcd2 mpn_hgcd2_5
+/* Not used, but renamed to not get duplicate definitions */
+#define __gmpn_hgcd_mul_matrix1_vector mpn_hgcd_mul_matrix1_vector_5
+
+#include "mpn/generic/hgcd2.c"
diff --git a/tune/speed.h b/tune/speed.h
index 5df155841..18f54600a 100644
--- a/tune/speed.h
+++ b/tune/speed.h
@@ -218,6 +218,8 @@ double speed_mpn_hgcd2 (struct speed_params *);
 double speed_mpn_hgcd2_1 (struct speed_params *);
 double speed_mpn_hgcd2_2 (struct speed_params *);
 double speed_mpn_hgcd2_3 (struct speed_params *);
+double speed_mpn_hgcd2_4 (struct speed_params *);
+double speed_mpn_hgcd2_5 (struct speed_params *);
 double speed_mpn_hgcd (struct speed_params *);
 double speed_mpn_hgcd_lehmer (struct speed_params *);
 double speed_mpn_hgcd_appr (struct speed_params *);
@@ -484,12 +486,11 @@ int mpn_jacobi_base_2 (mp_limb_t, mp_limb_t, int);
 int mpn_jacobi_base_3 (mp_limb_t, mp_limb_t, int);
 int mpn_jacobi_base_4 (mp_limb_t, mp_limb_t, int);
 
-int mpn_hgcd2_1 (mp_limb_t ah, mp_limb_t al, mp_limb_t bh, mp_limb_t bl,
-		 struct hgcd_matrix1 *M);
-int mpn_hgcd2_2 (mp_limb_t ah, mp_limb_t al, mp_limb_t bh, mp_limb_t bl,
-		 struct hgcd_matrix1 *M);
-int mpn_hgcd2_3 (mp_limb_t ah, mp_limb_t al, mp_limb_t bh, mp_limb_t bl,
-		 struct hgcd_matrix1 *M);
+int mpn_hgcd2_1 (mp_limb_t, mp_limb_t, mp_limb_t, mp_limb_t, struct hgcd_matrix1*);
+int mpn_hgcd2_2 (mp_limb_t, mp_limb_t, mp_limb_t, mp_limb_t, struct hgcd_matrix1*);
+int mpn_hgcd2_3 (mp_limb_t, mp_limb_t, mp_limb_t, mp_limb_t, struct hgcd_matrix1*);
+int mpn_hgcd2_4 (mp_limb_t, mp_limb_t, mp_limb_t, mp_limb_t, struct hgcd_matrix1*);
+int mpn_hgcd2_5 (mp_limb_t, mp_limb_t, mp_limb_t, mp_limb_t, struct hgcd_matrix1*);
 
 mp_limb_t mpn_mod_1_div (mp_srcptr, mp_size_t, mp_limb_t);
 mp_limb_t mpn_mod_1_inv (mp_srcptr, mp_size_t, mp_limb_t);
diff --git a/tune/tuneup.c b/tune/tuneup.c
index 94926d653..86ba5edf6 100644
--- a/tune/tuneup.c
+++ b/tune/tuneup.c
@@ -1962,14 +1962,22 @@ void
 tune_hgcd2 (void)
 {
   static struct param_t  param;
-  hgcd2_func_t *f[3] =
-    { mpn_hgcd2_1, mpn_hgcd2_2, mpn_hgcd2_3 };
-  speed_function_t speed_f[3] =
-    { speed_mpn_hgcd2_1, speed_mpn_hgcd2_2, speed_mpn_hgcd2_3 };
+  hgcd2_func_t *f[5] =
+    { mpn_hgcd2_1,
+      mpn_hgcd2_2,
+      mpn_hgcd2_3,
+      mpn_hgcd2_4,
+      mpn_hgcd2_5 };
+  speed_function_t speed_f[5] =
+    { speed_mpn_hgcd2_1,
+      speed_mpn_hgcd2_2,
+      speed_mpn_hgcd2_3,
+      speed_mpn_hgcd2_4,
+      speed_mpn_hgcd2_5 };
   int best;
 
   s.size = 1;
-  best = one_method (3, speed_f, "mpn_hgcd2", "HGCD2_DIV1_METHOD", &param);
+  best = one_method (5, speed_f, "mpn_hgcd2", "HGCD2_DIV1_METHOD", &param);
 
   /* Use selected function when tuning hgcd and gcd */
   hgcd2_func = f[best];