1 # Alpha 21064 __mpn_lshift --
3 # Copyright (C) 1994-2016 Free Software Foundation, Inc.
5 # This file is part of the GNU MP Library.
7 # The GNU MP Library is free software; you can redistribute it and/or modify
8 # it under the terms of the GNU Lesser General Public License as published by
9 # the Free Software Foundation; either version 2.1 of the License, or (at your
10 # option) any later version.
12 # The GNU MP Library is distributed in the hope that it will be useful, but
13 # WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
14 # or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
15 # License for more details.
17 # You should have received a copy of the GNU Lesser General Public License
18 # along with the GNU MP Library. If not, see <http://www.gnu.org/licenses/>.
27 # This code runs at 4.8 cycles/limb on the 21064. With infinite unrolling,
28 # it would take 4 cycles/limb. It should be possible to get down to 3
29 # cycles/limb since both ldq and stq can be paired with the other used
30 # instructions. But there are many restrictions in the 21064 pipeline that
31 # makes it hard, if not impossible, to get down to 3 cycles/limb:
33 # 1. ldq has a 3 cycle delay, srl and sll have a 2 cycle delay.
34 # 2. Only aligned instruction pairs can be paired.
35 # 3. The store buffer or silo might not be able to deal with the bandwidth.
46 s8addq $18,$17,$17 # make r17 point at end of s1
47 ldq $4,-8($17) # load first limb
50 s8addq $18,$16,$16 # make r16 point at end of RES
52 and $18,4-1,$20 # number of limbs in first loop
53 srl $4,$7,$0 # compute function result