rpms/gmp
7
0
Fork 0
Code Issues Pull Requests Releases Activity
27b0fbc019
gmp/ibm_z13_simd_part4.patch
Jakub Martisko 27b0fbc019 Add: optimizations for s390x 
Resolves: RHEL-10549
2024-02-05 17:37:08 +01:00

152 lines
6.5 KiB
Diff
Raw Blame History

From: Marius Hillenbrand <mhillen at linux.ibm.com>
---
mpn/s390_64/z13/gmp-mparam.h | 129 ++++++++++++++++++++++++++++++++++-
1 file changed, 127 insertions(+), 2 deletions(-)
diff --git a/mpn/s390_64/z13/gmp-mparam.h b/mpn/s390_64/z13/gmp-mparam.h
index a17503fd0..50e7f39d1 100644
--- a/mpn/s390_64/z13/gmp-mparam.h
+++ b/mpn/s390_64/z13/gmp-mparam.h
@@ -1,6 +1,6 @@
/* S/390-64 for IBM z13 gmp-mparam.h -- Compiler/machine parameter header file.
-Copyright 1991, 1993, 1994, 2000-2011 Free Software Foundation, Inc.
+Copyright 2021 Free Software Foundation, Inc.
This file is part of the GNU MP Library.
@@ -34,4 +34,129 @@ see https://www.gnu.org/licenses/. */
#define HAVE_NATIVE_mpn_addmul_2 1
#define HAVE_NATIVE_mpn_mul_2 1
-#include "mpn/s390_64/gmp-mparam.h"
+/* Generated by tuneup.c, 2021-07-30, gcc 10.2 */
+
+#define DIVREM_1_NORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define DIVREM_1_UNNORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define MOD_1_1P_METHOD 2
+#define MOD_1_NORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define MOD_1_UNNORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define MOD_1N_TO_MOD_1_1_THRESHOLD 17
+#define MOD_1U_TO_MOD_1_1_THRESHOLD 15
+#define MOD_1_1_TO_MOD_1_2_THRESHOLD 0 /* never mpn_mod_1_1p */
+#define MOD_1_2_TO_MOD_1_4_THRESHOLD 0 /* never mpn_mod_1s_2p */
+#define PREINV_MOD_1_TO_MOD_1_THRESHOLD 5
+#define USE_PREINV_DIVREM_1 1
+#define DIV_QR_1N_PI1_METHOD 3
+#define DIV_QR_1_NORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define DIV_QR_1_UNNORM_THRESHOLD MP_SIZE_T_MAX /* never */
+#define DIV_QR_2_PI2_THRESHOLD 996
+#define DIVEXACT_1_THRESHOLD 4
+#define BMOD_1_TO_MOD_1_THRESHOLD 0 /* always */
+
+#define DIV_1_VS_MUL_1_PERCENT 404
+
+#define MUL_TOOM22_THRESHOLD 23
+#define MUL_TOOM33_THRESHOLD 94
+#define MUL_TOOM44_THRESHOLD 166
+#define MUL_TOOM6H_THRESHOLD 286
+#define MUL_TOOM8H_THRESHOLD 626
+
+#define MUL_TOOM32_TO_TOOM43_THRESHOLD 113
+#define MUL_TOOM32_TO_TOOM53_THRESHOLD 138
+#define MUL_TOOM42_TO_TOOM53_THRESHOLD 143
+#define MUL_TOOM42_TO_TOOM63_THRESHOLD 145
+#define MUL_TOOM43_TO_TOOM54_THRESHOLD 130
+
+#define SQR_BASECASE_THRESHOLD 0 /* always (native) */
+#define SQR_TOOM2_THRESHOLD 12
+#define SQR_TOOM3_THRESHOLD 84
+#define SQR_TOOM4_THRESHOLD 234
+#define SQR_TOOM6_THRESHOLD 318
+#define SQR_TOOM8_THRESHOLD 478
+
+#define MULMID_TOOM42_THRESHOLD 42
+
+#define MULMOD_BNM1_THRESHOLD 13
+#define SQRMOD_BNM1_THRESHOLD 7
+
+#define MUL_FFT_MODF_THRESHOLD 332 /* k = 5 */
+#define MUL_FFT_TABLE3 \
+ { { 332, 5}, { 19, 6}, { 10, 5}, { 21, 6}, \
+ { 21, 7}, { 21, 8}, { 11, 7}, { 24, 8}, \
+ { 13, 7}, { 27, 8}, { 15, 7}, { 31, 8}, \
+ { 17, 7}, { 35, 8}, { 19, 7}, { 39, 8}, \
+ { 21, 9}, { 11, 8}, { 27, 9}, { 15, 8}, \
+ { 35, 9}, { 19, 8}, { 41, 9}, { 23, 8}, \
+ { 47, 9}, { 27,10}, { 15, 9}, { 39,10}, \
+ { 23, 9}, { 47,11}, { 15,10}, { 31, 9}, \
+ { 67,10}, { 47,11}, { 2048,12}, { 4096,13}, \
+ { 8192,14}, { 16384,15}, { 32768,16}, { 65536,17}, \
+ { 131072,18}, { 262144,19}, { 524288,20}, {1048576,21}, \
+ {2097152,22}, {4194304,23}, {8388608,24} }
+#define MUL_FFT_TABLE3_SIZE 47
+#define MUL_FFT_THRESHOLD 2752
+
+#define SQR_FFT_MODF_THRESHOLD 240 /* k = 5 */
+#define SQR_FFT_TABLE3 \
+ { { 240, 5}, { 8, 4}, { 17, 5}, { 13, 6}, \
+ { 7, 5}, { 15, 6}, { 8, 5}, { 17, 6}, \
+ { 9, 5}, { 19, 6}, { 15, 7}, { 8, 6}, \
+ { 17, 7}, { 9, 6}, { 19, 7}, { 10, 6}, \
+ { 21, 7}, { 17, 8}, { 9, 7}, { 20, 8}, \
+ { 11, 7}, { 23, 8}, { 13, 9}, { 7, 8}, \
+ { 21, 9}, { 11, 8}, { 23, 9}, { 15, 8}, \
+ { 31, 9}, { 19, 8}, { 39, 9}, { 23,10}, \
+ { 15, 9}, { 39,10}, { 23,11}, { 15,10}, \
+ { 31, 9}, { 63,10}, { 47,11}, { 2048,12}, \
+ { 4096,13}, { 8192,14}, { 16384,15}, { 32768,16}, \
+ { 65536,17}, { 131072,18}, { 262144,19}, { 524288,20}, \
+ {1048576,21}, {2097152,22}, {4194304,23}, {8388608,24} }
+#define SQR_FFT_TABLE3_SIZE 52
+#define SQR_FFT_THRESHOLD 1856
+
+#define MULLO_BASECASE_THRESHOLD 0 /* always */
+#define MULLO_DC_THRESHOLD 25
+#define MULLO_MUL_N_THRESHOLD 5397
+#define SQRLO_BASECASE_THRESHOLD 0 /* always */
+#define SQRLO_DC_THRESHOLD 396
+#define SQRLO_SQR_THRESHOLD 3704
+
+#define DC_DIV_QR_THRESHOLD 15
+#define DC_DIVAPPR_Q_THRESHOLD 50
+#define DC_BDIV_QR_THRESHOLD 66
+#define DC_BDIV_Q_THRESHOLD 202
+
+#define INV_MULMOD_BNM1_THRESHOLD 46
+#define INV_NEWTON_THRESHOLD 29
+#define INV_APPR_THRESHOLD 13
+
+#define BINV_NEWTON_THRESHOLD 312
+#define REDC_1_TO_REDC_2_THRESHOLD 79
+#define REDC_2_TO_REDC_N_THRESHOLD 0 /* always */
+
+#define MU_DIV_QR_THRESHOLD 979
+#define MU_DIVAPPR_Q_THRESHOLD 979
+#define MUPI_DIV_QR_THRESHOLD 13
+#define MU_BDIV_QR_THRESHOLD 942
+#define MU_BDIV_Q_THRESHOLD 1367
+
+#define POWM_SEC_TABLE 3,19,215,1730
+
+#define GET_STR_DC_THRESHOLD 10
+#define GET_STR_PRECOMPUTE_THRESHOLD 15
+#define SET_STR_DC_THRESHOLD 882
+#define SET_STR_PRECOMPUTE_THRESHOLD 2520
+
+#define FAC_DSC_THRESHOLD 228
+#define FAC_ODD_THRESHOLD 24
+
+#define MATRIX22_STRASSEN_THRESHOLD 19
+#define HGCD2_DIV1_METHOD 1
+#define HGCD_THRESHOLD 61
+#define HGCD_APPR_THRESHOLD 51
+#define HGCD_REDUCE_THRESHOLD 1962
+#define GCD_DC_THRESHOLD 217
+#define GCDEXT_DC_THRESHOLD 263
+#define JACOBI_BASE_METHOD 4
+
--
2.40.1
Reference in New Issue View Git Blame Copy Permalink