github.com/aergoio/aergo@v1.3.1/libtool/src/gmp-6.1.2/mpn/s390_64/README (about) 1 Copyright 2011 Free Software Foundation, Inc. 2 3 This file is part of the GNU MP Library. 4 5 The GNU MP Library is free software; you can redistribute it and/or modify 6 it under the terms of either: 7 8 * the GNU Lesser General Public License as published by the Free 9 Software Foundation; either version 3 of the License, or (at your 10 option) any later version. 11 12 or 13 14 * the GNU General Public License as published by the Free Software 15 Foundation; either version 2 of the License, or (at your option) any 16 later version. 17 18 or both in parallel, as here. 19 20 The GNU MP Library is distributed in the hope that it will be useful, but 21 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 22 or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 23 for more details. 24 25 You should have received copies of the GNU General Public License and the 26 GNU Lesser General Public License along with the GNU MP Library. If not, 27 see https://www.gnu.org/licenses/. 28 29 30 31 There are 5 generations of 64-but s390 processors, z900, z990, z9, 32 z10, and z196. The current GMP code was optimised for the two oldest, 33 z900 and z990. 34 35 36 mpn_copyi 37 38 This code makes use of a loop around MVC. It almost surely runs very 39 close to optimally. A small improvement could be done by using one 40 MVC for size 256 bytes, now we use two (we use an extra MVC when 41 copying any multiple of 256 bytes). 42 43 44 mpn_copyd 45 46 We have tried several feed-in variants here, branch tree, jump table 47 and computed goto. The fastest (on z990) turned out to be computed 48 goto. 49 50 An approach not tried is EX of LMG and STMG, modifying the register set 51 on-the-fly. Using that trick, we could completely avoid using 52 separate feed-in paths. 53 54 55 mpn_lshift, mpn_rshift 56 57 The current code runs at pipeline decode bandwidth on z990. 58 59 60 mpn_add_n, mpn_sub_n 61 62 The current code is 4-way unrolled. It should be unrolled more, at 63 least 8x, in order to reach 2.5 c/l. 64 65 66 mpn_mul_1, mpn_addmul_1, mpn_submul_1 67 68 The current code is very naive, but due to the non-pipelined nature of 69 MLGR on z900 and z990, more sophisticated code would not gain much. 70 71 On z10 one would need to cluster at least 4 MLGR together, in order to 72 reduce stalling. 73 74 On z196, one surely want to use unrolling and pipelining, to perhaps 75 reach around 12 c/l. A major issue here and on z10 is ALCGR's 3 cycle 76 stalling. 77 78 79 mpn_mul_2, mpn_addmul_2 80 81 At least for older machines (z900, z990) with very slow MLGR, we 82 should use Karatsuba's algorithm on 2-limb units, making mul_2 and 83 addmul_2 the main multiplication primitives. The newer machines might 84 benefit less from this approach, perhaps in particular z10, where MLGR 85 clustering is more important. 86 87 With Karatsuba, one could hope for around 16 cycles per accumulated 88 128 cross product, on z990.