[Gmp-commit] /var/hg/gmp: New bulldozer mul_1, addmul_1, submul_1.
mercurial at gmplib.org
mercurial at gmplib.org
Sun Feb 26 19:44:24 CET 2012
details: /var/hg/gmp/rev/828fda9ee236
changeset: 14686:828fda9ee236
user: Torbjorn Granlund <tege at gmplib.org>
date: Sun Feb 26 19:44:20 2012 +0100
description:
New bulldozer mul_1, addmul_1, submul_1.
diffstat:
ChangeLog | 5 +
mpn/x86_64/bd1/aorsmul_1.asm | 171 ++++++++++++++++++++++++++++++++++++++++++
mpn/x86_64/bd1/mul_1.asm | 174 +++++++++++++++++++++++++++++++++++++++++++
3 files changed, 350 insertions(+), 0 deletions(-)
diffs (truncated from 365 to 300 lines):
diff -r bd1a61e87b58 -r 828fda9ee236 ChangeLog
--- a/ChangeLog Sun Feb 26 18:07:00 2012 +0100
+++ b/ChangeLog Sun Feb 26 19:44:20 2012 +0100
@@ -1,3 +1,8 @@
+2012-02-26 Torbjorn Granlund <tege at gmplib.org>
+
+ * mpn/x86_64/bd1/mul_1.asm: New file.
+ * mpn/x86_64/bd1/aorsmul_1.asm: New file.
+
2012-02-26 Marco Bodrato <bodrato at mail.dm.unipi.it>
* mpz/2fac_ui.c: New file: implements n!!.
diff -r bd1a61e87b58 -r 828fda9ee236 mpn/x86_64/bd1/aorsmul_1.asm
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/mpn/x86_64/bd1/aorsmul_1.asm Sun Feb 26 19:44:20 2012 +0100
@@ -0,0 +1,171 @@
+dnl AMD64 mpn_addmul_1 and mpn_submul_1 optimised for AMD Bulldozer.
+
+dnl Copyright 2003, 2004, 2005, 2007, 2008, 2011, 2012 Free Software
+dnl Foundation, Inc.
+
+dnl This file is part of the GNU MP Library.
+
+dnl The GNU MP Library is free software; you can redistribute it and/or modify
+dnl it under the terms of the GNU Lesser General Public License as published
+dnl by the Free Software Foundation; either version 3 of the License, or (at
+dnl your option) any later version.
+
+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 Lesser General Public
+dnl License for more details.
+
+dnl You should have received a copy of the GNU Lesser General Public License
+dnl along with the GNU MP Library. If not, see http://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C cycles/limb
+C AMD K8,K9
+C AMD K10
+C AMD bd1 4.5-4.7
+C AMD bobcat
+C Intel P4
+C Intel core2
+C Intel NHM
+C Intel SBR
+C Intel atom
+C VIA nano
+
+C The loop of this code is the result of running a code generation and
+C optimisation tool suite written by David Harvey and Torbjorn Granlund.
+
+C TODO
+C * Try to make loop run closer to 4 c/l.
+
+define(`rp', `%rdi') C rcx
+define(`up', `%rsi') C rdx
+define(`n_param', `%rdx') C r8
+define(`v0', `%rcx') C r9
+
+define(`n', `%r11')
+
+ifdef(`OPERATION_addmul_1',`
+ define(`ADDSUB', `add')
+ define(`func', `mpn_addmul_1')
+')
+ifdef(`OPERATION_submul_1',`
+ define(`ADDSUB', `sub')
+ define(`func', `mpn_submul_1')
+')
+
+ABI_SUPPORT(DOS64)
+ABI_SUPPORT(STD64)
+
+MULFUNC_PROLOGUE(mpn_addmul_1 mpn_submul_1)
+
+IFDOS(` define(`up', ``%rsi'') ') dnl
+IFDOS(` define(`rp', ``%rcx'') ') dnl
+IFDOS(` define(`v0', ``%r9'') ') dnl
+IFDOS(` define(`r9', ``rdi'') ') dnl
+IFDOS(` define(`n', ``%r8'') ') dnl
+IFDOS(` define(`r8', ``r11'') ') dnl
+
+ASM_START()
+ TEXT
+ ALIGN(16)
+PROLOGUE(func)
+IFDOS(``push %rsi '')
+IFDOS(``push %rdi '')
+IFDOS(``mov %rdx, %rsi '')
+
+ mov (up), %rax C read first u limb early
+ push %rbx
+IFSTD(` mov n_param, %rbx ') C move away n from rdx, mul uses it
+IFDOS(` mov n, %rbx ')
+ mul v0
+
+IFSTD(` mov %rbx, n ')
+
+ and $3, R32(%rbx)
+ lea -16(rp,n,8), rp
+ jz L(b0)
+ cmp $2, R32(%rbx)
+ jb L(b1)
+ jz L(b2)
+
+L(b3): mov $0, R32(%r8)
+ mov %rax, %rbx
+ mov $0, R32(%r9)
+ mov 8(up), %rax
+ mov %rdx, %r10
+ lea (up,n,8), up
+ not n
+ jmp L(L3)
+
+L(b0): mov $0, R32(%r10)
+ mov %rax, %r8
+ mov %rdx, %rbx
+ mov 8(up), %rax
+ lea (up,n,8), up
+ neg n
+ jmp L(L0)
+
+L(b1): cmp $1, n
+ jz L(n1)
+ mov %rax, %r9
+ mov 8(up), %rax
+ mov %rdx, %r8
+ mov $0, R32(%rbx)
+ lea (up,n,8), up
+ neg n
+ inc n
+ jmp L(L1)
+
+L(b2): mov $0, R32(%rbx)
+ mov %rax, %r10
+ mov %rdx, %r9
+ mov 8(up), %rax
+ mov $0, R32(%r8)
+ lea (up,n,8), up
+ neg n
+ add $2, n
+ jns L(end)
+
+ ALIGN(32)
+L(top): mul v0
+ ADDSUB %r10, (rp,n,8)
+ adc %rax, %r9
+ mov (up,n,8), %rax
+ adc %rdx, %r8
+L(L1): mul v0
+ mov $0, R32(%r10)
+ ADDSUB %r9, 8(rp,n,8)
+ adc %rax, %r8
+ adc %rdx, %rbx
+ mov 8(up,n,8), %rax
+L(L0): mul v0
+ ADDSUB %r8, 16(rp,n,8)
+ mov $0, R32(%r8)
+ adc %rax, %rbx
+ mov $0, R32(%r9)
+ mov 16(up,n,8), %rax
+ adc %rdx, %r10
+L(L3): mul v0
+ ADDSUB %rbx, 24(rp,n,8)
+ mov $0, R32(%rbx)
+ adc %rax, %r10
+ adc %rdx, %r9
+ mov 24(up,n,8), %rax
+ add $4, n
+ js L(top)
+
+L(end): mul v0
+ ADDSUB %r10, (rp)
+ adc %r9, %rax
+ adc %r8, %rdx
+L(n1): ADDSUB %rax, 8(rp)
+ adc $0, %rdx
+ mov %rdx, %rax
+
+ pop %rbx
+IFDOS(``pop %rdi '')
+IFDOS(``pop %rsi '')
+ ret
+EPILOGUE()
+ASM_END()
diff -r bd1a61e87b58 -r 828fda9ee236 mpn/x86_64/bd1/mul_1.asm
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/mpn/x86_64/bd1/mul_1.asm Sun Feb 26 19:44:20 2012 +0100
@@ -0,0 +1,174 @@
+dnl AMD64 mpn_mul_1 optimised for AMD Bulldozer.
+
+dnl Copyright 2003, 2004, 2005, 2007, 2008, 2011, 2012 Free Software
+dnl Foundation, Inc.
+
+dnl This file is part of the GNU MP Library.
+
+dnl The GNU MP Library is free software; you can redistribute it and/or modify
+dnl it under the terms of the GNU Lesser General Public License as published
+dnl by the Free Software Foundation; either version 3 of the License, or (at
+dnl your option) any later version.
+
+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 Lesser General Public
+dnl License for more details.
+
+dnl You should have received a copy of the GNU Lesser General Public License
+dnl along with the GNU MP Library. If not, see http://www.gnu.org/licenses/.
+
+include(`../config.m4')
+
+C cycles/limb
+C AMD K8,K9
+C AMD K10
+C AMD bd1 4
+C AMD bobcat
+C Intel P4
+C Intel core2
+C Intel NHM
+C Intel SBR
+C Intel atom
+C VIA nano
+
+C The loop of this code is the result of running a code generation and
+C optimisation tool suite written by David Harvey and Torbjorn Granlund.
+
+C TODO
+C * Move loop code into feed-in blocks, to save insn for zeroing regs.
+
+define(`rp', `%rdi') C rcx
+define(`up', `%rsi') C rdx
+define(`n_param', `%rdx') C r8
+define(`v0', `%rcx') C r9
+
+define(`n', `%rbx')
+
+ABI_SUPPORT(DOS64)
+ABI_SUPPORT(STD64)
+
+IFDOS(` define(`up', ``%rsi'') ') dnl
+IFDOS(` define(`rp', ``%rcx'') ') dnl
+IFDOS(` define(`v0', ``%r9'') ') dnl
+IFDOS(` define(`r9', ``rdi'') ') dnl
+IFDOS(` define(`n', ``%r8'') ') dnl
+IFDOS(` define(`r8', ``r11'') ') dnl
+
+ASM_START()
+ TEXT
+ ALIGN(16)
+PROLOGUE(mpn_mul_1c)
+IFDOS(``push %rsi '')
+IFDOS(``push %rdi '')
+IFDOS(``mov %rdx, %rsi '')
+
+ mov (up), %rax C read first u limb early
+ push %rbx
+IFSTD(` mov n_param, %r11 ') C move away n from rdx, mul uses it
+IFDOS(` mov n, %r11 ')
+ mul v0
+
+IFSTD(` add %r8, %rax ')
+IFDOS(` add 64(%rsp), %rax ') C 40 + 3*8 (3 push insns)
+ adc $0, %rdx
+ jmp L(common)
+
+EPILOGUE()
+
+ ALIGN(16)
+PROLOGUE(mpn_mul_1)
+IFDOS(``push %rsi '')
+IFDOS(``push %rdi '')
+IFDOS(``mov %rdx, %rsi '')
+
+ mov (up), %rax C read first u limb early
+ push %rbx
+IFSTD(` mov n_param, %r11 ') C move away n from rdx, mul uses it
+IFDOS(` mov n, %r11 ')
+ mul v0
+
+L(common):
+IFSTD(` mov %r11, n ')
+
+ and $3, R32(%r11)
+ lea -16(rp,n,8), rp
+ jz L(b0)
+ cmp $2, R32(%r11)
+ jb L(b1)
+ jz L(b2)
+
+L(b3): mov %rax, %r10
+ mov %rdx, %r11
+ mov 8(up), %rax
+ mul v0
+ lea (up,n,8), up
+ not n
+ jmp L(L3)
+
+L(b0): mov %rax, %r9
More information about the gmp-commit
mailing list