New mpn_div_qr_1n_pi1 variants, DIV_QR_1N_METHOD 3 and 4.

Not enabled, but hooked into the speed and tuneup programs.

8 files changed, 103 insertions, 5 deletions

diff --git a/tune/Makefile.am b/tune/Makefile.am
index 9e382567c..0f564ed49 100644
--- a/tune/Makefile.am
+++ b/tune/Makefile.am
@@ -53,7 +53,8 @@ EXTRA_LTLIBRARIES = libspeed.la
 
 libspeed_la_SOURCES =							\
   common.c divrem1div.c divrem1inv.c divrem2div.c divrem2inv.c		\
-  div_qr_1n_pi1_1.c div_qr_1n_pi1_2.c div_qr_1_tune.c			\
+  div_qr_1n_pi1_1.c div_qr_1n_pi1_2.c div_qr_1n_pi1_3.c 		\
+  div_qr_1n_pi1_4.c div_qr_1_tune.c					\
   freq.c								\
   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	\
diff --git a/tune/common.c b/tune/common.c
index 68bcfa3ad..a62e0a91a 100644
--- a/tune/common.c
+++ b/tune/common.c
@@ -718,6 +718,16 @@ speed_mpn_div_qr_1n_pi1_2 (struct speed_params *s)
 {
   SPEED_ROUTINE_MPN_DIV_QR_1N_PI1 (mpn_div_qr_1n_pi1_2);
 }
+double
+speed_mpn_div_qr_1n_pi1_3 (struct speed_params *s)
+{
+  SPEED_ROUTINE_MPN_DIV_QR_1N_PI1 (mpn_div_qr_1n_pi1_3);
+}
+double
+speed_mpn_div_qr_1n_pi1_4 (struct speed_params *s)
+{
+  SPEED_ROUTINE_MPN_DIV_QR_1N_PI1 (mpn_div_qr_1n_pi1_4);
+}
 
 double
 speed_mpn_div_qr_1 (struct speed_params *s)
diff --git a/tune/div_qr_1_tune.c b/tune/div_qr_1_tune.c
index ef680eecc..3c30558a3 100644
--- a/tune/div_qr_1_tune.c
+++ b/tune/div_qr_1_tune.c
@@ -34,10 +34,13 @@ see https://www.gnu.org/licenses/.  */
 
 mp_limb_t mpn_div_qr_1n_pi1_1 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
 mp_limb_t mpn_div_qr_1n_pi1_2 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
+mp_limb_t mpn_div_qr_1n_pi1_3 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
 
 #if !HAVE_NATIVE_mpn_div_qr_1n_pi1
-#define __gmpn_div_qr_1n_pi1 \
-  (div_qr_1n_pi1_method == 1 ? mpn_div_qr_1n_pi1_1 : mpn_div_qr_1n_pi1_2)
+#define __gmpn_div_qr_1n_pi1						\
+  (div_qr_1n_pi1_method <= 2						\
+   ? (div_qr_1n_pi1_method == 1 ? mpn_div_qr_1n_pi1_1 : mpn_div_qr_1n_pi1_2) \
+   : (div_qr_1n_pi1_method == 3 ? mpn_div_qr_1n_pi1_3 : mpn_div_qr_1n_pi1_4))
 #endif
 
 #undef mpn_div_qr_1
diff --git a/tune/div_qr_1n_pi1_3.c b/tune/div_qr_1n_pi1_3.c
new file mode 100644
index 000000000..826244c44
--- /dev/null
+++ b/tune/div_qr_1n_pi1_3.c
@@ -0,0 +1,38 @@
+/* mpn/generic/div_qr_1n_pi1.c method 3.
+
+Copyright 2013 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 DIV_QR_1N_METHOD
+#define DIV_QR_1N_METHOD 3
+#undef mpn_div_qr_1n_pi1
+#define mpn_div_qr_1n_pi1 mpn_div_qr_1n_pi1_3
+
+#include "mpn/generic/div_qr_1n_pi1.c"
diff --git a/tune/div_qr_1n_pi1_4.c b/tune/div_qr_1n_pi1_4.c
new file mode 100644
index 000000000..0f69ea0a1
--- /dev/null
+++ b/tune/div_qr_1n_pi1_4.c
@@ -0,0 +1,38 @@
+/* mpn/generic/div_qr_1n_pi1.c method 4.
+
+Copyright 2013 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 DIV_QR_1N_METHOD
+#define DIV_QR_1N_METHOD 4
+#undef mpn_div_qr_1n_pi1
+#define mpn_div_qr_1n_pi1 mpn_div_qr_1n_pi1_4
+
+#include "mpn/generic/div_qr_1n_pi1.c"
diff --git a/tune/speed.c b/tune/speed.c
index a833dab57..eb793ee51 100644
--- a/tune/speed.c
+++ b/tune/speed.c
@@ -244,6 +244,8 @@ const struct routine_t {
   { "mpn_div_qr_1n_pi1", speed_mpn_div_qr_1n_pi1, FLAG_R  },
   { "mpn_div_qr_1n_pi1_1",speed_mpn_div_qr_1n_pi1_1, FLAG_R  },
   { "mpn_div_qr_1n_pi1_2",speed_mpn_div_qr_1n_pi1_2, FLAG_R  },
+  { "mpn_div_qr_1n_pi1_3",speed_mpn_div_qr_1n_pi1_3, FLAG_R  },
+  { "mpn_div_qr_1n_pi1_4",speed_mpn_div_qr_1n_pi1_4, FLAG_R  },
   { "mpn_div_qr_1",      speed_mpn_div_qr_1,      FLAG_R },
 
   { "mpn_div_qr_2n",     speed_mpn_div_qr_2n,       },
diff --git a/tune/speed.h b/tune/speed.h
index 6f65dada2..636b7671e 100644
--- a/tune/speed.h
+++ b/tune/speed.h
@@ -210,6 +210,8 @@ double speed_mpn_divrem_2_inv (struct speed_params *);
 double speed_mpn_div_qr_1n_pi1 (struct speed_params *);
 double speed_mpn_div_qr_1n_pi1_1 (struct speed_params *);
 double speed_mpn_div_qr_1n_pi1_2 (struct speed_params *);
+double speed_mpn_div_qr_1n_pi1_3 (struct speed_params *);
+double speed_mpn_div_qr_1n_pi1_4 (struct speed_params *);
 double speed_mpn_div_qr_1 (struct speed_params *);
 double speed_mpn_div_qr_2n (struct speed_params *);
 double speed_mpn_div_qr_2u (struct speed_params *);
@@ -482,6 +484,8 @@ void speed_option_set (const char *);
 
 mp_limb_t mpn_div_qr_1n_pi1_1 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
 mp_limb_t mpn_div_qr_1n_pi1_2 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
+mp_limb_t mpn_div_qr_1n_pi1_3 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
+mp_limb_t mpn_div_qr_1n_pi1_4 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t, mp_limb_t, mp_limb_t);
 
 mp_limb_t mpn_divrem_1_div (mp_ptr, mp_size_t, mp_srcptr, mp_size_t, mp_limb_t);
 mp_limb_t mpn_divrem_1_inv (mp_ptr, mp_size_t, mp_srcptr, mp_size_t, mp_limb_t);
diff --git a/tune/tuneup.c b/tune/tuneup.c
index 86ba5edf6..8ae211eef 100644
--- a/tune/tuneup.c
+++ b/tune/tuneup.c
@@ -2253,16 +2253,18 @@ tune_div_qr_1 (void)
   if (!HAVE_NATIVE_mpn_div_qr_1n_pi1)
     {
       static struct param_t  param;
-      speed_function_t f[2] =
+      speed_function_t f[] =
 	{
 	 speed_mpn_div_qr_1n_pi1_1,
 	 speed_mpn_div_qr_1n_pi1_2,
+	 speed_mpn_div_qr_1n_pi1_3,
+	 speed_mpn_div_qr_1n_pi1_4,
 	};
 
       s.size = 10;
       s.r = randlimb_norm ();
 
-      one_method (2, f, "mpn_div_qr_1n_pi1", "DIV_QR_1N_PI1_METHOD", &param);
+      one_method (numberof(f), f, "mpn_div_qr_1n_pi1", "DIV_QR_1N_PI1_METHOD", &param);
     }
 
   {