1Copyright 2002, 2005 Free Software Foundation, Inc. 2 3This file is part of the GNU MP Library. 4 5The GNU MP Library is free software; you can redistribute it and/or modify it 6under the terms of the GNU Lesser General Public License as published by the 7Free Software Foundation; either version 3 of the License, or (at your 8option) any later version. 9 10The GNU MP Library is distributed in the hope that it will be useful, but 11WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or 12FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License 13for more details. 14 15You should have received a copy of the GNU Lesser General Public License along 16with the GNU MP Library. If not, see http://www.gnu.org/licenses/. 17 18 19 20 21 22This directory contains assembly code for nails-enabled 21264. The code is not 23very well optimized. 24 25For addmul_N, as N grows larger, we could make multiple loads together, then do 26about 3.3 i/c. 10 cycles after the last load, we can increase to 4 i/c. This 27would surely allow addmul_4 to run at 2 c/l, but the same should be possible 28also for addmul_3 and perhaps even addmul_2. 29 30 31 current fair best 32Routine c/l unroll c/l unroll c/l i/c 33mul_1 3.25 2.75 2.75 3.273 34addmul_1 4.0 4 3.5 4 14 3.25 3.385 35addmul_2 4.0 1 2.5 2 10 2.25 3.333 36addmul_3 3.0 1 2.33 2 14 2 3.333 37addmul_4 2.5 1 2.125 2 17 2 3.135 38 39addmul_5 2 1 10 40addmul_6 2 1 12 41addmul_7 2 1 14 42 43(The "best" column doesn't account for bookkeeping instructions and 44thereby assumes infinite unrolling.) 45 46Basecase usages: 47 481 addmul_1 492 addmul_2 503 addmul_3 514 addmul_4 525 addmul_3 + addmul_2 2.3998 536 addmul_4 + addmul_2 547 addmul_4 + addmul_3 55