[Gmp-commit] /var/hg/gmp: (relspeed_div_1_vs_mul_1): New function.

[mercurial at gmplib.org]

details:   /var/hg/gmp/rev/9a5d42a0d435
changeset: 17195:9a5d42a0d435
user:      Torbjorn Granlund <tg at gmplib.org>
date:      Sat Dec 31 22:09:11 2016 +0100
description:
(relspeed_div_1_vs_mul_1): New function.

diffstat:

 tune/tuneup.c |  46 +++++++++++++++++++++++++++++++++++++++++++++-
 1 files changed, 45 insertions(+), 1 deletions(-)

diffs (69 lines):

diff -r 0ce63e6ff725 -r 9a5d42a0d435 tune/tuneup.c
--- a/tune/tuneup.c	Fri Dec 30 07:56:52 2016 +0100
+++ b/tune/tuneup.c	Sat Dec 31 22:09:11 2016 +0100
@@ -1,6 +1,6 @@
 /* Create tuned thresholds for various algorithms.
 
-Copyright 1999-2003, 2005, 2006, 2008-2012 Free Software Foundation, Inc.
+Copyright 1999-2003, 2005, 2006, 2008-2016 Free Software Foundation, Inc.
 
 This file is part of the GNU MP Library.
 
@@ -1194,6 +1194,47 @@
     }
 }
 
+/* Compare mpn_mul_1 to whatever fast exact single-limb division we have.  This
+   is currently mpn_divexact_1, but will become mpn_bdiv_1_qr_pi2 or somesuch.
+   This is used in get_str and set_str.  */
+void
+relspeed_div_1_vs_mul_1 (void)
+{
+  const size_t max_opsize = 100;
+  const mp_limb_t fake_big_base = (~CNST_LIMB(0)) / 3;
+  mp_size_t n;
+  long j;
+  mp_limb_t rp[max_opsize];
+  mp_limb_t ap[max_opsize];
+  double multime, divtime;
+
+  mpn_random (ap, max_opsize);
+
+  multime = 0;
+  for (n = max_opsize; n > 1; n--)
+    {
+      mpn_mul_1 (rp, ap, n, fake_big_base);
+      speed_starttime ();
+      for (j = speed_precision; j != 0 ; j--)
+	mpn_mul_1 (rp, ap, n, fake_big_base);
+      multime += speed_endtime () / n;
+    }
+
+  divtime = 0;
+  for (n = max_opsize; n > 1; n--)
+    {
+      /* Make input divisible for good measure.  */
+      ap[n - 1] = mpn_mul_1 (ap, ap, n - 1, fake_big_base);
+
+      mpn_divexact_1 (rp, ap, n, fake_big_base);
+      speed_starttime ();
+      for (j = speed_precision; j != 0 ; j--)
+	mpn_divexact_1 (rp, ap, n, fake_big_base);
+      divtime += speed_endtime () / n;
+    }
+
+  print_define ("DIV_1_VS_MUL_1_PERCENT", (int) (100 * divtime/multime));
+}
 
 
 /* Start karatsuba from 4, since the Cray t90 ieee code is much faster at 2,
@@ -2862,6 +2903,9 @@
   tune_modexact_1_odd ();
   printf("\n");
 
+  relspeed_div_1_vs_mul_1 ();
+  printf("\n");
+
   tune_mul_n ();
   printf("\n");