Basic constant folding implementation

7 files changed, 612 insertions, 0 deletions

diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/add_n.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/add_n.asm
new file mode 100644
index 0000000..c781596
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/add_n.asm
@@ -0,0 +1,70 @@
+dnl  SPARC T1 32-bit mpn_add_n.
+
+dnl  Copyright 2010 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C INPUT PARAMETERS
+define(`rp',  %o0)
+define(`ap',  %o1)
+define(`bp',  %o2)
+define(`n',   %o3)
+define(`cy',  %o4)
+
+define(`i',   %o3)
+
+MULFUNC_PROLOGUE(mpn_add_n mpn_add_nc)
+
+ASM_START()
+PROLOGUE(mpn_add_nc)
+	b	L(ent)
+	srl	cy, 0, cy	C strip any bogus high bits
+EPILOGUE()
+
+PROLOGUE(mpn_add_n)
+	mov	0, cy
+L(ent):	srl	n, 0, n		C strip any bogus high bits
+	sll	n, 2, n
+	add	ap, n, ap
+	add	bp, n, bp
+	add	rp, n, rp
+	neg	n, i
+
+L(top):	lduw	[ap+i], %g1
+	lduw	[bp+i], %g2
+	add	%g1, %g2, %g3
+	add	%g3, cy, %g3
+	stw	%g3, [rp+i]
+	add	i, 4, i
+	brnz	i, L(top)
+	srlx	%g3, 32, cy
+
+	retl
+	mov	cy, %o0		C return value
+EPILOGUE()
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/addmul_1.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/addmul_1.asm
new file mode 100644
index 0000000..89da186
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/addmul_1.asm
@@ -0,0 +1,90 @@
+dnl  SPARC T1 32-bit mpn_addmul_1.
+
+dnl  Contributed to the GNU project by David Miller.
+
+dnl  Copyright 2010, 2013 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C		   cycles/limb
+C UltraSPARC T1:       24
+C UltraSPARC T2:       19
+C UltraSPARC T3:       19
+C UltraSPARC T4:       5
+
+C INPUT PARAMETERS
+define(`rp',	`%i0')
+define(`up',	`%i1')
+define(`n',	`%i2')
+define(`v0',	`%i3')
+
+ASM_START()
+PROLOGUE(mpn_addmul_1)
+	save	%sp, -96, %sp
+	srl	n, 0, %o4
+	srl	v0, 0, %g1
+	subcc	%o4, 1, %o4
+	be	L(final_one)
+	 clr	%o5
+
+L(top):	lduw	[up+0], %l0
+	lduw	[rp+0], %l2
+	lduw	[up+4], %l1
+	lduw	[rp+4], %l3
+	mulx	%l0, %g1, %g3
+	add	up, 8, up
+	mulx	%l1, %g1, %o3
+	sub	%o4, 2, %o4
+	add	rp, 8, rp
+	add	%l2, %g3, %g3
+	add	%o5, %g3, %g3
+	stw	%g3, [rp-8]
+	srlx	%g3, 32, %o5
+	add	%l3, %o3, %o3
+	add	%o5, %o3, %o3
+	stw	%o3, [rp-4]
+	brgz	%o4, L(top)
+	 srlx	%o3, 32, %o5
+
+	brlz,pt	%o4, L(done)
+	 nop
+
+L(final_one):
+	lduw	[up+0], %l0
+	lduw	[rp+0], %l2
+	mulx	%l0, %g1, %g3
+	add	%l2, %g3, %g3
+	add	%o5, %g3, %g3
+	stw	%g3, [rp+0]
+	srlx	%g3, 32, %o5
+
+L(done):
+	ret
+	 restore %o5, 0, %o0
+EPILOGUE()
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/gmp-mparam.h b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/gmp-mparam.h
new file mode 100644
index 0000000..6f9d5a4
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/gmp-mparam.h
@@ -0,0 +1,153 @@
+/* UltraSPARC T 32-bit gmp-mparam.h -- Compiler/machine parameter header file.
+
+Copyright 1991, 1993, 1994, 2000-2011 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/.  */
+
+#define DIVREM_1_NORM_THRESHOLD              0  /* always */
+#define DIVREM_1_UNNORM_THRESHOLD            3
+#define MOD_1_1P_METHOD                      2
+#define MOD_1_NORM_THRESHOLD                 0  /* always */
+#define MOD_1_UNNORM_THRESHOLD               0  /* always */
+#define MOD_1N_TO_MOD_1_1_THRESHOLD          9
+#define MOD_1U_TO_MOD_1_1_THRESHOLD         10
+#define MOD_1_1_TO_MOD_1_2_THRESHOLD         0  /* never mpn_mod_1_1p */
+#define MOD_1_2_TO_MOD_1_4_THRESHOLD        21
+#define PREINV_MOD_1_TO_MOD_1_THRESHOLD     22
+#define USE_PREINV_DIVREM_1                  1
+#define DIV_QR_2_PI2_THRESHOLD           MP_SIZE_T_MAX  /* never */
+#define DIVEXACT_1_THRESHOLD                 0  /* always */
+#define BMOD_1_TO_MOD_1_THRESHOLD           35
+
+#define MUL_TOOM22_THRESHOLD                14
+#define MUL_TOOM33_THRESHOLD                98
+#define MUL_TOOM44_THRESHOLD               166
+#define MUL_TOOM6H_THRESHOLD               226
+#define MUL_TOOM8H_THRESHOLD               333
+
+#define MUL_TOOM32_TO_TOOM43_THRESHOLD      97
+#define MUL_TOOM32_TO_TOOM53_THRESHOLD     139
+#define MUL_TOOM42_TO_TOOM53_THRESHOLD      97
+#define MUL_TOOM42_TO_TOOM63_THRESHOLD      98
+#define MUL_TOOM43_TO_TOOM54_THRESHOLD     120
+
+#define SQR_BASECASE_THRESHOLD               6
+#define SQR_TOOM2_THRESHOLD                 34
+#define SQR_TOOM3_THRESHOLD                110
+#define SQR_TOOM4_THRESHOLD                178
+#define SQR_TOOM6_THRESHOLD                240
+#define SQR_TOOM8_THRESHOLD                333
+
+#define MULMID_TOOM42_THRESHOLD             22
+
+#define MULMOD_BNM1_THRESHOLD                9
+#define SQRMOD_BNM1_THRESHOLD               13
+
+#define MUL_FFT_MODF_THRESHOLD             280  /* k = 5 */
+#define MUL_FFT_TABLE3                                      \
+  { {    280, 5}, {     15, 6}, {      8, 5}, {     17, 6}, \
+    {      9, 5}, {     19, 6}, {     13, 7}, {      7, 6}, \
+    {     17, 7}, {      9, 6}, {     20, 7}, {     11, 6}, \
+    {     23, 7}, {     13, 8}, {      7, 7}, {     21, 8}, \
+    {     11, 7}, {     25, 9}, {      7, 8}, {     15, 7}, \
+    {     33, 8}, {     19, 7}, {     41, 8}, {     23, 7}, \
+    {     49, 8}, {     27, 9}, {     15, 8}, {     31, 7}, \
+    {     63, 8}, {     39, 9}, {     23, 8}, {     47,10}, \
+    {     15, 9}, {     31, 8}, {     67, 9}, {     39, 8}, \
+    {     79, 9}, {     47,10}, {     31, 9}, {     79,10}, \
+    {     47,11}, {     31,10}, {     63, 9}, {    127, 8}, \
+    {    255,10}, {     79, 9}, {    159, 8}, {    319,10}, \
+    {     95, 9}, {    191, 8}, {    383,11}, {     63,10}, \
+    {    127, 9}, {    255,10}, {    143, 9}, {    287,10}, \
+    {    159, 9}, {    319,10}, {    175,11}, {     95,10}, \
+    {    191, 9}, {    383,12}, {   4096,13}, {   8192,14}, \
+    {  16384,15}, {  32768,16} }
+#define MUL_FFT_TABLE3_SIZE 66
+#define MUL_FFT_THRESHOLD                 3712
+
+#define SQR_FFT_MODF_THRESHOLD             240  /* k = 5 */
+#define SQR_FFT_TABLE3                                      \
+  { {    240, 5}, {     15, 6}, {      8, 5}, {     17, 6}, \
+    {     13, 7}, {      7, 6}, {     17, 7}, {      9, 6}, \
+    {     20, 7}, {     11, 6}, {     23, 7}, {     13, 8}, \
+    {      7, 7}, {     19, 8}, {     11, 7}, {     25, 9}, \
+    {      7, 8}, {     15, 7}, {     33, 8}, {     19, 7}, \
+    {     39, 8}, {     23, 7}, {     47, 8}, {     27, 9}, \
+    {     15, 8}, {     39, 9}, {     23, 8}, {     47,10}, \
+    {     15, 9}, {     31, 8}, {     63, 9}, {     39, 8}, \
+    {     79, 9}, {     47,10}, {     31, 9}, {     63, 8}, \
+    {    127, 9}, {     71, 8}, {    143, 9}, {     79,10}, \
+    {     47,11}, {     31,10}, {     63, 9}, {    127, 8}, \
+    {    255, 9}, {    143,10}, {     79, 9}, {    159, 8}, \
+    {    319, 9}, {    175,10}, {     95, 9}, {    191, 8}, \
+    {    383, 9}, {    207,11}, {     63,10}, {    127, 9}, \
+    {    255,10}, {    143, 9}, {    287,10}, {    159, 9}, \
+    {    319,10}, {    175,11}, {     95,10}, {    191, 9}, \
+    {    383,10}, {    207,12}, {   4096,13}, {   8192,14}, \
+    {  16384,15}, {  32768,16} }
+#define SQR_FFT_TABLE3_SIZE 70
+#define SQR_FFT_THRESHOLD                 2624
+
+#define MULLO_BASECASE_THRESHOLD             0  /* always */
+#define MULLO_DC_THRESHOLD                  51
+#define MULLO_MUL_N_THRESHOLD             6633
+
+#define DC_DIV_QR_THRESHOLD                 51
+#define DC_DIVAPPR_Q_THRESHOLD             202
+#define DC_BDIV_QR_THRESHOLD                47
+#define DC_BDIV_Q_THRESHOLD                124
+
+#define INV_MULMOD_BNM1_THRESHOLD           26
+#define INV_NEWTON_THRESHOLD               266
+#define INV_APPR_THRESHOLD                 222
+
+#define BINV_NEWTON_THRESHOLD              296
+#define REDC_1_TO_REDC_N_THRESHOLD          59
+
+#define MU_DIV_QR_THRESHOLD               1334
+#define MU_DIVAPPR_Q_THRESHOLD            1499
+#define MUPI_DIV_QR_THRESHOLD              116
+#define MU_BDIV_QR_THRESHOLD              1057
+#define MU_BDIV_Q_THRESHOLD               1334
+
+#define POWM_SEC_TABLE  6,35,213,724,2618
+
+#define MATRIX22_STRASSEN_THRESHOLD         15
+#define HGCD_THRESHOLD                      84
+#define HGCD_APPR_THRESHOLD                101
+#define HGCD_REDUCE_THRESHOLD             1437
+#define GCD_DC_THRESHOLD                   372
+#define GCDEXT_DC_THRESHOLD                253
+#define JACOBI_BASE_METHOD                   2
+
+#define GET_STR_DC_THRESHOLD                12
+#define GET_STR_PRECOMPUTE_THRESHOLD        27
+#define SET_STR_DC_THRESHOLD               399
+#define SET_STR_PRECOMPUTE_THRESHOLD       885
+
+#define FAC_DSC_THRESHOLD                  179
+#define FAC_ODD_THRESHOLD                   29
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/mul_1.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/mul_1.asm
new file mode 100644
index 0000000..0239cd2
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/mul_1.asm
@@ -0,0 +1,83 @@
+dnl  SPARC T1 32-bit mpn_mul_1.
+
+dnl  Contributed to the GNU project by David Miller.
+
+dnl  Copyright 2010, 2013 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C		   cycles/limb
+C UltraSPARC T1:       20
+C UltraSPARC T2:       18
+C UltraSPARC T3:       18
+C UltraSPARC T4:       4
+
+C INPUT PARAMETERS
+define(`rp',	`%o0')
+define(`up',	`%o1')
+define(`n',	`%o2')
+define(`v0',	`%o3')
+
+ASM_START()
+PROLOGUE(mpn_mul_1)
+	srl	n, 0, n
+	srl	v0, 0, v0
+	subcc	n, 1, n
+	be	L(final_one)
+	 clr	%o5
+
+L(top):	lduw	[up+0], %g1
+	lduw	[up+4], %g2
+	mulx	%g1, v0, %g3
+	add	up, 8, up
+	mulx	%g2, v0, %o4
+	sub	n, 2, n
+	add	rp, 8, rp
+	add	%o5, %g3, %g3
+	stw	%g3, [rp-8]
+	srlx	%g3, 32, %o5
+	add	%o5, %o4, %o4
+	stw	%o4, [rp-4]
+	brgz	n, L(top)
+	 srlx	%o4, 32, %o5
+
+	brlz,pt	n, L(done)
+	 nop
+
+L(final_one):
+	lduw	[up+0], %g1
+	mulx	%g1, v0, %g3
+	add	%o5, %g3, %g3
+	stw	%g3, [rp+0]
+	srlx	%g3, 32, %o5
+
+L(done):
+	retl
+	 mov	%o5, %o0
+EPILOGUE()
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sqr_diagonal.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sqr_diagonal.asm
new file mode 100644
index 0000000..3b906ef
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sqr_diagonal.asm
@@ -0,0 +1,55 @@
+dnl  SPARC T1 32-bit mpn_sqr_diagonal.
+
+dnl  Copyright 2010 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C INPUT PARAMETERS
+define(`rp',	`%o0')
+define(`up',	`%o1')
+define(`n',	`%o2')
+
+ASM_START()
+PROLOGUE(mpn_sqr_diagonal)
+	deccc	n			C n--
+	nop
+
+L(top):	lduw	[up+0], %g1
+	add	up, 4, up		C up++
+	mulx	%g1, %g1, %g3
+	stw	%g3, [rp+0]
+	srlx	%g3, 32, %g4
+	stw	%g4, [rp+4]
+	add	rp, 8, rp		C rp += 2
+	bnz	%icc, L(top)
+	deccc	n			C n--
+
+	retl
+	nop
+EPILOGUE()
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sub_n.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sub_n.asm
new file mode 100644
index 0000000..946bc3f
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/sub_n.asm
@@ -0,0 +1,70 @@
+dnl  SPARC T1 32-bit mpn_sub_n.
+
+dnl  Copyright 2010 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C INPUT PARAMETERS
+define(`rp',  %o0)
+define(`ap',  %o1)
+define(`bp',  %o2)
+define(`n',   %o3)
+define(`cy',  %o4)
+
+define(`i',   %o3)
+
+MULFUNC_PROLOGUE(mpn_sub_n mpn_sub_nc)
+
+ASM_START()
+PROLOGUE(mpn_sub_nc)
+	b	L(ent)
+	srl	cy, 0, cy	C strip any bogus high bits
+EPILOGUE()
+
+PROLOGUE(mpn_sub_n)
+	mov	0, cy
+L(ent):	srl	n, 0, n		C strip any bogus high bits
+	sll	n, 2, n
+	add	ap, n, ap
+	add	bp, n, bp
+	add	rp, n, rp
+	neg	n, i
+
+L(top):	lduw	[ap+i], %g1
+	lduw	[bp+i], %g2
+	sub	%g1, %g2, %g3
+	sub	%g3, cy, %g3
+	stw	%g3, [rp+i]
+	add	i, 4, i
+	brnz	i, L(top)
+	srlx	%g3, 63, cy
+
+	retl
+	mov	cy, %o0		C return value
+EPILOGUE()
diff --git a/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/submul_1.asm b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/submul_1.asm
new file mode 100644
index 0000000..8920070
--- /dev/null
+++ b/vendor/gmp-6.3.0/mpn/sparc32/ultrasparct1/submul_1.asm
@@ -0,0 +1,91 @@
+dnl  SPARC T1 32-bit mpn_submul_1.
+
+dnl  Contributed to the GNU project by David Miller.
+
+dnl  Copyright 2010, 2013 Free Software Foundation, Inc.
+
+dnl  This file is part of the GNU MP Library.
+dnl
+dnl  The GNU MP Library is free software; you can redistribute it and/or modify
+dnl  it under the terms of either:
+dnl
+dnl    * the GNU Lesser General Public License as published by the Free
+dnl      Software Foundation; either version 3 of the License, or (at your
+dnl      option) any later version.
+dnl
+dnl  or
+dnl
+dnl    * the GNU General Public License as published by the Free Software
+dnl      Foundation; either version 2 of the License, or (at your option) any
+dnl      later version.
+dnl
+dnl  or both in parallel, as here.
+dnl
+dnl  The GNU MP Library is distributed in the hope that it will be useful, but
+dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
+dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+dnl  for more details.
+dnl
+dnl  You should have received copies of the GNU General Public License and the
+dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
+dnl  see https://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C		   cycles/limb
+C UltraSPARC T1:       24
+C UltraSPARC T2:       19
+C UltraSPARC T3:       19
+C UltraSPARC T4:       5
+
+C INPUT PARAMETERS
+define(`rp',	`%i0')
+define(`up',	`%i1')
+define(`n',	`%i2')
+define(`v0',	`%i3')
+
+ASM_START()
+PROLOGUE(mpn_submul_1)
+	save	%sp, -96, %sp
+	srl	n, 0, %o4
+	srl	v0, 0, %g1
+	subcc	%o4, 1, %o4
+	be	L(final_one)
+	 subcc	%g0, 0, %o5
+
+L(top):	lduw	[up+0], %l0
+	lduw	[rp+0], %l2
+	lduw	[up+4], %l1
+	lduw	[rp+4], %l3
+	mulx	%l0, %g1, %g3
+	add	up, 8, up
+	mulx	%l1, %g1, %o3
+	sub	%o4, 2, %o4
+	add	rp, 8, rp
+	addx	%o5, %g3, %g3
+	srlx	%g3, 32, %o5
+	subcc	%l2, %g3, %g3
+	stw	%g3, [rp-8]
+	addx	%o5, %o3, %o3
+	srlx	%o3, 32, %o5
+	subcc	%l3, %o3, %o3
+	brgz	%o4, L(top)
+	 stw	%o3, [rp-4]
+
+	brlz,pt	%o4, L(done)
+	 nop
+
+L(final_one):
+	lduw	[up+0], %l0
+	lduw	[rp+0], %l2
+	mulx	%l0, %g1, %g3
+	addx	%o5, %g3, %g3
+	srlx	%g3, 32, %o5
+	subcc	%l2, %g3, %g3
+	stw	%g3, [rp+0]
+
+L(done):
+	addx	%o5, 0, %o5
+	ret
+	 restore %o5, 0, %o0
+EPILOGUE()