diff options
author | Niels M?ller <nisse@lysator.liu.se> | 2019-09-04 21:46:41 +0200 |
---|---|---|
committer | Niels M?ller <nisse@lysator.liu.se> | 2019-09-04 21:46:41 +0200 |
commit | 95d545683e958f411d896cc9ae55507c144a063a (patch) | |
tree | 394105f2853efc7b25dc90943651fb5a26644643 /tune | |
parent | 74469d71e6792fccdfb17f2913aaabafb7094a89 (diff) | |
download | gmp-95d545683e958f411d896cc9ae55507c144a063a.tar.gz |
Tuneup support for HGCD2_METHOD, controlling div1 usage in mpn_hgcd2.
* mpn/generic/hgcd2.c (HGCD2_METHOD): New parameter.
(DIV1): New macro, using either the div1 function or plain
division, depending on the value of HGCD2_METHOD.
(mpn_hgcd2): Use DIV1.
* tune/speed.c, tune/speed.h, tune/common.c, tune/Makefile.am: Add
measuring of mpn_hgcd2 methods.
* tune/hgcd2-1.c, tune/hgcd2-2.c: New files.
* tune/tuneup.c: Tune HGCD2_METHOD.
Diffstat (limited to 'tune')
-rw-r--r-- | tune/Makefile.am | 1 | ||||
-rw-r--r-- | tune/common.c | 10 | ||||
-rw-r--r-- | tune/hgcd2-1.c | 39 | ||||
-rw-r--r-- | tune/hgcd2-2.c | 39 | ||||
-rw-r--r-- | tune/speed.c | 2 | ||||
-rw-r--r-- | tune/speed.h | 7 | ||||
-rw-r--r-- | tune/tuneup.c | 27 |
7 files changed, 125 insertions, 0 deletions
diff --git a/tune/Makefile.am b/tune/Makefile.am index 5fb3b50d3..4fa28ed12 100644 --- a/tune/Makefile.am +++ b/tune/Makefile.am @@ -58,6 +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 \ 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 a918a67a4..da4fb1145 100644 --- a/tune/common.c +++ b/tune/common.c @@ -1638,6 +1638,16 @@ speed_mpn_hgcd2 (struct speed_params *s) { SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2); } +double +speed_mpn_hgcd2_1 (struct speed_params *s) +{ + SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2_1); +} +double +speed_mpn_hgcd2_2 (struct speed_params *s) +{ + SPEED_ROUTINE_MPN_HGCD2 (mpn_hgcd2_2); +} double speed_mpn_hgcd (struct speed_params *s) diff --git a/tune/hgcd2-1.c b/tune/hgcd2-1.c new file mode 100644 index 000000000..ecc033b50 --- /dev/null +++ b/tune/hgcd2-1.c @@ -0,0 +1,39 @@ +/* mpn/generic/hgcd2.c method 1. + +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_METHOD +#define HGCD2_METHOD 1 +#define __gmpn_hgcd2 mpn_hgcd2_1 +/* Not used, but renamed to not get duplicate definitions */ +#define __gmpn_hgcd_mul_matrix1_vector mpn_hgcd_mul_matrix1_vector_1 + +#include "mpn/generic/hgcd2.c" diff --git a/tune/hgcd2-2.c b/tune/hgcd2-2.c new file mode 100644 index 000000000..aa045dd7b --- /dev/null +++ b/tune/hgcd2-2.c @@ -0,0 +1,39 @@ +/* mpn/generic/hgcd2.c method 2. + +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_METHOD +#define HGCD2_METHOD 2 +#define __gmpn_hgcd2 mpn_hgcd2_2 +/* Not used, but renamed to not get duplicate definitions */ +#define __gmpn_hgcd_mul_matrix1_vector mpn_hgcd_mul_matrix1_vector_2 + +#include "mpn/generic/hgcd2.c" diff --git a/tune/speed.c b/tune/speed.c index 4220d6be5..b46d94476 100644 --- a/tune/speed.c +++ b/tune/speed.c @@ -286,6 +286,8 @@ const struct routine_t { { "mpn_matrix22_mul", speed_mpn_matrix22_mul }, { "mpn_hgcd2", speed_mpn_hgcd2, FLAG_NODATA }, + { "mpn_hgcd2_1", speed_mpn_hgcd2_1, FLAG_NODATA }, + { "mpn_hgcd2_2", speed_mpn_hgcd2_2, FLAG_NODATA }, { "mpn_hgcd", speed_mpn_hgcd }, { "mpn_hgcd_lehmer", speed_mpn_hgcd_lehmer }, { "mpn_hgcd_appr", speed_mpn_hgcd_appr }, diff --git a/tune/speed.h b/tune/speed.h index 044bc9c2b..968bccac7 100644 --- a/tune/speed.h +++ b/tune/speed.h @@ -215,6 +215,8 @@ double speed_mpn_div_qr_2u (struct speed_params *); double speed_mpn_fib2_ui (struct speed_params *); double speed_mpn_matrix22_mul (struct speed_params *); 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_hgcd (struct speed_params *); double speed_mpn_hgcd_lehmer (struct speed_params *); double speed_mpn_hgcd_appr (struct speed_params *); @@ -481,6 +483,11 @@ 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); + 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 600a4a03b..5642bf789 100644 --- a/tune/tuneup.c +++ b/tune/tuneup.c @@ -1899,6 +1899,32 @@ tune_matrix22_mul (void) } void +tune_hgcd2 (void) +{ + static struct param_t param; + double t1, t2; + int method; + + s.size = 1; + t1 = tuneup_measure (speed_mpn_hgcd2_1, ¶m, &s); + if (option_trace >= 1) + printf ("size=%ld, mpn_hgcd2_1 %.9f\n", (long) s.size, t1); + + t2 = tuneup_measure (speed_mpn_hgcd2_2, ¶m, &s); + if (option_trace >= 1) + printf ("size=%ld, mpn_hgcd2_2 %.9f\n", (long) s.size, t2); + + if (t1 == -1.0 || t2 == -1.0) + { + printf ("Oops, can't measure all mpn_hgcd2 methods\n"); + abort (); + } + + method = (t1 < t2) ? 1 : 2; + print_define ("HGCD2_METHOD", method); +} + +void tune_hgcd (void) { static struct param_t param; @@ -2973,6 +2999,7 @@ all (void) printf("\n"); tune_matrix22_mul (); + tune_hgcd2 (); tune_hgcd (); tune_hgcd_appr (); tune_hgcd_reduce(); |