From 11da511c784eca003deb90c23570f0873954e0de Mon Sep 17 00:00:00 2001 From: Duncan Wilkie Date: Sat, 18 Nov 2023 06:11:09 -0600 Subject: Initial commit. --- gmp-6.3.0/mpn/s390_32/esame/aors_n.asm | 137 +++++++++++++++++++++++++++++++++ 1 file changed, 137 insertions(+) create mode 100644 gmp-6.3.0/mpn/s390_32/esame/aors_n.asm (limited to 'gmp-6.3.0/mpn/s390_32/esame/aors_n.asm') diff --git a/gmp-6.3.0/mpn/s390_32/esame/aors_n.asm b/gmp-6.3.0/mpn/s390_32/esame/aors_n.asm new file mode 100644 index 0000000..98b0dbc --- /dev/null +++ b/gmp-6.3.0/mpn/s390_32/esame/aors_n.asm @@ -0,0 +1,137 @@ +dnl S/390-32 mpn_add_n and mpn_sub_n. + +dnl Copyright 2011 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 z900 ? +C z990 2.75-3 (fast for even n, slow for odd n) +C z9 ? +C z10 ? +C z196 ? + +C TODO +C * Optimise for small n +C * Use r0 and save/restore one less register +C * Using logops_n's v1 inner loop operand order make the loop about 20% +C faster, at the expense of highly alignment-dependent performance. + +C INPUT PARAMETERS +define(`rp', `%r2') +define(`up', `%r3') +define(`vp', `%r4') +define(`n', `%r5') + +ifdef(`OPERATION_add_n', ` + define(ADSB, al) + define(ADSBCR, alcr) + define(ADSBC, alc) + define(RETVAL,`dnl + lhi %r2, 0 + alcr %r2, %r2') + define(func, mpn_add_n) + define(func_nc, mpn_add_nc)') +ifdef(`OPERATION_sub_n', ` + define(ADSB, sl) + define(ADSBCR, slbr) + define(ADSBC, slb) + define(RETVAL,`dnl + slbr %r2, %r2 + lcr %r2, %r2') + define(func, mpn_sub_n) + define(func_nc, mpn_sub_nc)') + +MULFUNC_PROLOGUE(mpn_add_n mpn_sub_n) + +ASM_START() +PROLOGUE(func) + stm %r6, %r8, 24(%r15) + + ahi n, 3 + lhi %r7, 3 + lr %r1, n + srl %r1, 2 + nr %r7, n C n mod 4 + je L(b1) + chi %r7, 2 + jl L(b2) + jne L(b0) + +L(b3): lm %r5, %r7, 0(up) + la up, 12(up) + ADSB %r5, 0(vp) + ADSBC %r6, 4(vp) + ADSBC %r7, 8(vp) + la vp, 12(vp) + stm %r5, %r7, 0(rp) + la rp, 12(rp) + brct %r1, L(top) + j L(end) + +L(b0): lm %r5, %r8, 0(up) C This redundant insns is no mistake, + la up, 16(up) C it is needed to make main loop run + ADSB %r5, 0(vp) C fast for n = 0 (mod 4). + ADSBC %r6, 4(vp) + j L(m0) + +L(b1): l %r5, 0(up) + la up, 4(up) + ADSB %r5, 0(vp) + la vp, 4(vp) + st %r5, 0(rp) + la rp, 4(rp) + brct %r1, L(top) + j L(end) + +L(b2): lm %r5, %r6, 0(up) + la up, 8(up) + ADSB %r5, 0(vp) + ADSBC %r6, 4(vp) + la vp, 8(vp) + stm %r5, %r6, 0(rp) + la rp, 8(rp) + brct %r1, L(top) + j L(end) + +L(top): lm %r5, %r8, 0(up) + la up, 16(up) + ADSBC %r5, 0(vp) + ADSBC %r6, 4(vp) +L(m0): ADSBC %r7, 8(vp) + ADSBC %r8, 12(vp) + la vp, 16(vp) + stm %r5, %r8, 0(rp) + la rp, 16(rp) + brct %r1, L(top) + +L(end): RETVAL + lm %r6, %r8, 24(%r15) + br %r14 +EPILOGUE() -- cgit v1.2.3