sub_n.s revision 1.1
1 1.1 mrg ; mc88100 mpn_sub_n -- Subtract two limb vectors of the same length > 0 and 2 1.1 mrg ; store difference in a third limb vector. 3 1.1 mrg 4 1.1 mrg ; Copyright 1992, 1994, 1996, 2000 Free Software Foundation, Inc. 5 1.1 mrg 6 1.1 mrg ; This file is part of the GNU MP Library. 7 1.1 mrg 8 1.1 mrg ; The GNU MP Library is free software; you can redistribute it and/or modify 9 1.1 mrg ; it under the terms of the GNU Lesser General Public License as published by 10 1.1 mrg ; the Free Software Foundation; either version 3 of the License, or (at your 11 1.1 mrg ; option) any later version. 12 1.1 mrg 13 1.1 mrg ; The GNU MP Library is distributed in the hope that it will be useful, but 14 1.1 mrg ; WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 15 1.1 mrg ; or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public 16 1.1 mrg ; License for more details. 17 1.1 mrg 18 1.1 mrg ; You should have received a copy of the GNU Lesser General Public License 19 1.1 mrg ; along with the GNU MP Library. If not, see http://www.gnu.org/licenses/. 20 1.1 mrg 21 1.1 mrg 22 1.1 mrg ; INPUT PARAMETERS 23 1.1 mrg ; res_ptr r2 24 1.1 mrg ; s1_ptr r3 25 1.1 mrg ; s2_ptr r4 26 1.1 mrg ; size r5 27 1.1 mrg 28 1.1 mrg ; This code has been optimized to run one instruction per clock, avoiding 29 1.1 mrg ; load stalls and writeback contention. As a result, the instruction 30 1.1 mrg ; order is not always natural. 31 1.1 mrg 32 1.1 mrg ; The speed is about 4.6 clocks/limb + 18 clocks/limb-vector on an 88100, 33 1.1 mrg ; but on the 88110, it seems to run much slower, 6.6 clocks/limb. 34 1.1 mrg 35 1.1 mrg text 36 1.1 mrg align 16 37 1.1 mrg global ___gmpn_sub_n 38 1.1 mrg ___gmpn_sub_n: 39 1.1 mrg ld r6,r3,0 ; read first limb from s1_ptr 40 1.1 mrg extu r10,r5,3 41 1.1 mrg ld r7,r4,0 ; read first limb from s2_ptr 42 1.1 mrg 43 1.1 mrg subu r5,r0,r5 44 1.1 mrg mak r5,r5,3<4> 45 1.1 mrg bcnd.n eq0,r5,Lzero 46 1.1 mrg subu.co r0,r0,r0 ; initialize carry 47 1.1 mrg 48 1.1 mrg or r12,r0,lo16(Lbase) 49 1.1 mrg or.u r12,r12,hi16(Lbase) 50 1.1 mrg addu r12,r12,r5 ; r12 is address for entering in loop 51 1.1 mrg 52 1.1 mrg extu r5,r5,2 ; divide by 4 53 1.1 mrg subu r2,r2,r5 ; adjust res_ptr 54 1.1 mrg subu r3,r3,r5 ; adjust s1_ptr 55 1.1 mrg subu r4,r4,r5 ; adjust s2_ptr 56 1.1 mrg 57 1.1 mrg or r8,r6,r0 58 1.1 mrg 59 1.1 mrg jmp.n r12 60 1.1 mrg or r9,r7,r0 61 1.1 mrg 62 1.1 mrg Loop: addu r3,r3,32 63 1.1 mrg st r8,r2,28 64 1.1 mrg addu r4,r4,32 65 1.1 mrg ld r6,r3,0 66 1.1 mrg addu r2,r2,32 67 1.1 mrg ld r7,r4,0 68 1.1 mrg Lzero: subu r10,r10,1 ; subtract 0 + 8r limbs (adj loop cnt) 69 1.1 mrg Lbase: ld r8,r3,4 70 1.1 mrg subu.cio r6,r6,r7 71 1.1 mrg ld r9,r4,4 72 1.1 mrg st r6,r2,0 73 1.1 mrg ld r6,r3,8 ; subtract 7 + 8r limbs 74 1.1 mrg subu.cio r8,r8,r9 75 1.1 mrg ld r7,r4,8 76 1.1 mrg st r8,r2,4 77 1.1 mrg ld r8,r3,12 ; subtract 6 + 8r limbs 78 1.1 mrg subu.cio r6,r6,r7 79 1.1 mrg ld r9,r4,12 80 1.1 mrg st r6,r2,8 81 1.1 mrg ld r6,r3,16 ; subtract 5 + 8r limbs 82 1.1 mrg subu.cio r8,r8,r9 83 1.1 mrg ld r7,r4,16 84 1.1 mrg st r8,r2,12 85 1.1 mrg ld r8,r3,20 ; subtract 4 + 8r limbs 86 1.1 mrg subu.cio r6,r6,r7 87 1.1 mrg ld r9,r4,20 88 1.1 mrg st r6,r2,16 89 1.1 mrg ld r6,r3,24 ; subtract 3 + 8r limbs 90 1.1 mrg subu.cio r8,r8,r9 91 1.1 mrg ld r7,r4,24 92 1.1 mrg st r8,r2,20 93 1.1 mrg ld r8,r3,28 ; subtract 2 + 8r limbs 94 1.1 mrg subu.cio r6,r6,r7 95 1.1 mrg ld r9,r4,28 96 1.1 mrg st r6,r2,24 97 1.1 mrg bcnd.n ne0,r10,Loop ; subtract 1 + 8r limbs 98 1.1 mrg subu.cio r8,r8,r9 99 1.1 mrg 100 1.1 mrg st r8,r2,28 ; store most significant limb 101 1.1 mrg 102 1.1 mrg addu.ci r2,r0,r0 ; return carry-out from most sign. limb 103 1.1 mrg jmp.n r1 104 1.1 mrg xor r2,r2,1 105
Indexes created Tue Oct 27 00:25:24 UTC 2026