diff options
Diffstat (limited to 'vendor/gmp-6.3.0/mpn/x86/pentium/rshift.asm')
-rw-r--r-- | vendor/gmp-6.3.0/mpn/x86/pentium/rshift.asm | 243 |
1 files changed, 243 insertions, 0 deletions
diff --git a/vendor/gmp-6.3.0/mpn/x86/pentium/rshift.asm b/vendor/gmp-6.3.0/mpn/x86/pentium/rshift.asm new file mode 100644 index 0000000..2105c4c --- /dev/null +++ b/vendor/gmp-6.3.0/mpn/x86/pentium/rshift.asm @@ -0,0 +1,243 @@ +dnl Intel Pentium mpn_rshift -- mpn right shift. + +dnl Copyright 1992, 1994-1996, 1999, 2000, 2002 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 P5,P54: 6.0 +C P55: 5.375 + + +C mp_limb_t mpn_rshift (mp_ptr dst, mp_srcptr src, mp_size_t size, +C unsigned shift); +C +C The main shift-by-N loop should run at 5.375 c/l and that's what P55 does, +C but P5 and P54 run only at 6.0 c/l, which is 4 cycles lost somewhere. + +defframe(PARAM_SHIFT,16) +defframe(PARAM_SIZE, 12) +defframe(PARAM_SRC, 8) +defframe(PARAM_DST, 4) + + TEXT + ALIGN(8) +PROLOGUE(mpn_rshift) + + pushl %edi + pushl %esi + pushl %ebx + pushl %ebp +deflit(`FRAME',16) + + movl PARAM_DST,%edi + movl PARAM_SRC,%esi + movl PARAM_SIZE,%ebp + movl PARAM_SHIFT,%ecx + +C We can use faster code for shift-by-1 under certain conditions. + cmp $1,%ecx + jne L(normal) + leal 4(%edi),%eax + cmpl %esi,%eax + jnc L(special) C jump if res_ptr + 1 >= s_ptr + leal (%edi,%ebp,4),%eax + cmpl %eax,%esi + jnc L(special) C jump if s_ptr >= res_ptr + size + +L(normal): + movl (%esi),%edx + addl $4,%esi + xorl %eax,%eax + shrdl( %cl, %edx, %eax) C compute carry limb + pushl %eax C push carry limb onto stack + + decl %ebp + pushl %ebp + shrl $3,%ebp + jz L(end) + + movl (%edi),%eax C fetch destination cache line + + ALIGN(4) +L(oop): movl 28(%edi),%eax C fetch destination cache line + movl %edx,%ebx + + movl (%esi),%eax + movl 4(%esi),%edx + shrdl( %cl, %eax, %ebx) + shrdl( %cl, %edx, %eax) + movl %ebx,(%edi) + movl %eax,4(%edi) + + movl 8(%esi),%ebx + movl 12(%esi),%eax + shrdl( %cl, %ebx, %edx) + shrdl( %cl, %eax, %ebx) + movl %edx,8(%edi) + movl %ebx,12(%edi) + + movl 16(%esi),%edx + movl 20(%esi),%ebx + shrdl( %cl, %edx, %eax) + shrdl( %cl, %ebx, %edx) + movl %eax,16(%edi) + movl %edx,20(%edi) + + movl 24(%esi),%eax + movl 28(%esi),%edx + shrdl( %cl, %eax, %ebx) + shrdl( %cl, %edx, %eax) + movl %ebx,24(%edi) + movl %eax,28(%edi) + + addl $32,%esi + addl $32,%edi + decl %ebp + jnz L(oop) + +L(end): popl %ebp + andl $7,%ebp + jz L(end2) +L(oop2): + movl (%esi),%eax + shrdl( %cl,%eax,%edx) C compute result limb + movl %edx,(%edi) + movl %eax,%edx + addl $4,%esi + addl $4,%edi + decl %ebp + jnz L(oop2) + +L(end2): + shrl %cl,%edx C compute most significant limb + movl %edx,(%edi) C store it + + popl %eax C pop carry limb + + popl %ebp + popl %ebx + popl %esi + popl %edi + ret + + +C We loop from least significant end of the arrays, which is only +C permissable if the source and destination don't overlap, since the +C function is documented to work for overlapping source and destination. + +L(special): + leal -4(%edi,%ebp,4),%edi + leal -4(%esi,%ebp,4),%esi + + movl (%esi),%edx + subl $4,%esi + + decl %ebp + pushl %ebp + shrl $3,%ebp + + shrl %edx + incl %ebp + decl %ebp + jz L(Lend) + + movl (%edi),%eax C fetch destination cache line + + ALIGN(4) +L(Loop): + movl -28(%edi),%eax C fetch destination cache line + movl %edx,%ebx + + movl (%esi),%eax + movl -4(%esi),%edx + rcrl %eax + movl %ebx,(%edi) + rcrl %edx + movl %eax,-4(%edi) + + movl -8(%esi),%ebx + movl -12(%esi),%eax + rcrl %ebx + movl %edx,-8(%edi) + rcrl %eax + movl %ebx,-12(%edi) + + movl -16(%esi),%edx + movl -20(%esi),%ebx + rcrl %edx + movl %eax,-16(%edi) + rcrl %ebx + movl %edx,-20(%edi) + + movl -24(%esi),%eax + movl -28(%esi),%edx + rcrl %eax + movl %ebx,-24(%edi) + rcrl %edx + movl %eax,-28(%edi) + + leal -32(%esi),%esi C use leal not to clobber carry + leal -32(%edi),%edi + decl %ebp + jnz L(Loop) + +L(Lend): + popl %ebp + sbbl %eax,%eax C save carry in %eax + andl $7,%ebp + jz L(Lend2) + addl %eax,%eax C restore carry from eax +L(Loop2): + movl %edx,%ebx + movl (%esi),%edx + rcrl %edx + movl %ebx,(%edi) + + leal -4(%esi),%esi C use leal not to clobber carry + leal -4(%edi),%edi + decl %ebp + jnz L(Loop2) + + jmp L(L1) +L(Lend2): + addl %eax,%eax C restore carry from eax +L(L1): movl %edx,(%edi) C store last limb + + movl $0,%eax + rcrl %eax + + popl %ebp + popl %ebx + popl %esi + popl %edi + ret + +EPILOGUE() |