1dnl SPARC v9 mpn_addmul_1 for T3/T4/T5. 2 3dnl Contributed to the GNU project by David Miller and Torbjörn Granlund. 4 5dnl Copyright 2013 Free Software Foundation, Inc. 6 7dnl This file is part of the GNU MP Library. 8dnl 9dnl The GNU MP Library is free software; you can redistribute it and/or modify 10dnl it under the terms of either: 11dnl 12dnl * the GNU Lesser General Public License as published by the Free 13dnl Software Foundation; either version 3 of the License, or (at your 14dnl option) any later version. 15dnl 16dnl or 17dnl 18dnl * the GNU General Public License as published by the Free Software 19dnl Foundation; either version 2 of the License, or (at your option) any 20dnl later version. 21dnl 22dnl or both in parallel, as here. 23dnl 24dnl The GNU MP Library is distributed in the hope that it will be useful, but 25dnl WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 26dnl or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 27dnl for more details. 28dnl 29dnl You should have received copies of the GNU General Public License and the 30dnl GNU Lesser General Public License along with the GNU MP Library. If not, 31dnl see https://www.gnu.org/licenses/. 32 33include(`../config.m4') 34 35C cycles/limb 36C UltraSPARC T3: 26 37C UltraSPARC T4: 4.5 38 39C INPUT PARAMETERS 40define(`rp', `%i0') 41define(`up', `%i1') 42define(`n', `%i2') 43define(`v0', `%i3') 44 45define(`u0', `%l0') 46define(`u1', `%l1') 47define(`u2', `%l2') 48define(`u3', `%l3') 49define(`r0', `%l4') 50define(`r1', `%l5') 51define(`r2', `%l6') 52define(`r3', `%l7') 53 54ASM_START() 55 REGISTER(%g2,#scratch) 56 REGISTER(%g3,#scratch) 57PROLOGUE(mpn_addmul_1) 58 save %sp, -176, %sp 59 ldx [up+0], %g1 60 61 and n, 3, %g3 62 brz %g3, L(b0) 63 addcc %g0, %g0, %g5 C clear carry limb, flag 64 cmp %g3, 2 65 bcs %xcc, L(b01) 66 nop 67 be %xcc, L(b10) 68 ldx [up+8], %g5 69 70L(b11): ldx [up+16], u3 71 mulx %g1, v0, %o2 72 umulxhi(%g1, v0, %o3) 73 ldx [rp+0], r1 74 mulx %g5, v0, %o4 75 ldx [rp+8], r2 76 umulxhi(%g5, v0, %o5) 77 ldx [rp+16], r3 78 mulx u3, v0, %g4 79 umulxhi(u3, v0, %g5) 80 addcc %o3, %o4, %o4 81 addxccc(%o5, %g4, %g4) 82 addxc( %g0, %g5, %g5) 83 addcc r1, %o2, r1 84 stx r1, [rp+0] 85 addxccc(r2, %o4, r2) 86 stx r2, [rp+8] 87 addxccc(r3, %g4, r3) 88 stx r3, [rp+16] 89 add n, -3, n 90 add up, 24, up 91 brz n, L(xit) 92 add rp, 24, rp 93 b L(com) 94 nop 95 96L(b10): mulx %g1, v0, %o4 97 ldx [rp+0], r2 98 umulxhi(%g1, v0, %o5) 99 ldx [rp+8], r3 100 mulx %g5, v0, %g4 101 umulxhi(%g5, v0, %g5) 102 addcc %o5, %g4, %g4 103 addxc( %g0, %g5, %g5) 104 addcc r2, %o4, r2 105 stx r2, [rp+0] 106 addxccc(r3, %g4, r3) 107 stx r3, [rp+8] 108 add n, -2, n 109 add up, 16, up 110 brz n, L(xit) 111 add rp, 16, rp 112 b L(com) 113 nop 114 115L(b01): ldx [rp+0], r3 116 mulx %g1, v0, %g4 117 umulxhi(%g1, v0, %g5) 118 addcc r3, %g4, r3 119 stx r3, [rp+0] 120 add n, -1, n 121 add up, 8, up 122 brz n, L(xit) 123 add rp, 8, rp 124 125L(com): ldx [up+0], %g1 126L(b0): ldx [up+8], u1 127 ldx [up+16], u2 128 ldx [up+24], u3 129 mulx %g1, v0, %o0 130 umulxhi(%g1, v0, %o1) 131 b L(lo0) 132 nop 133 134 ALIGN(16) 135L(top): ldx [up+0], u0 136 addxc( %g0, %g5, %g5) C propagate carry into carry limb 137 ldx [up+8], u1 138 addcc r0, %o0, r0 139 ldx [up+16], u2 140 addxccc(r1, %o2, r1) 141 ldx [up+24], u3 142 addxccc(r2, %o4, r2) 143 stx r0, [rp-32] 144 addxccc(r3, %g4, r3) 145 stx r1, [rp-24] 146 mulx u0, v0, %o0 147 stx r2, [rp-16] 148 umulxhi(u0, v0, %o1) 149 stx r3, [rp-8] 150L(lo0): mulx u1, v0, %o2 151 ldx [rp+0], r0 152 umulxhi(u1, v0, %o3) 153 ldx [rp+8], r1 154 mulx u2, v0, %o4 155 ldx [rp+16], r2 156 umulxhi(u2, v0, %o5) 157 ldx [rp+24], r3 158 mulx u3, v0, %g4 159 addxccc(%g5, %o0, %o0) 160 umulxhi(u3, v0, %g5) 161 add up, 32, up 162 addxccc(%o1, %o2, %o2) 163 add rp, 32, rp 164 addxccc(%o3, %o4, %o4) 165 add n, -4, n 166 addxccc(%o5, %g4, %g4) 167 brgz n, L(top) 168 nop 169 170 addxc( %g0, %g5, %g5) 171 addcc r0, %o0, r0 172 stx r0, [rp-32] 173 addxccc(r1, %o2, r1) 174 stx r1, [rp-24] 175 addxccc(r2, %o4, r2) 176 stx r2, [rp-16] 177 addxccc(r3, %g4, r3) 178 stx r3, [rp-8] 179L(xit): addxc( %g0, %g5, %i0) 180 ret 181 restore 182EPILOGUE() 183