gonum.org/v1/gonum@v0.14.0/internal/asm/f64/dot_amd64.s (about) 1 // Copyright ©2015 The Gonum Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style 3 // license that can be found in the LICENSE file. 4 // 5 // Some of the loop unrolling code is copied from: 6 // http://golang.org/src/math/big/arith_amd64.s 7 // which is distributed under these terms: 8 // 9 // Copyright (c) 2012 The Go Authors. All rights reserved. 10 // 11 // Redistribution and use in source and binary forms, with or without 12 // modification, are permitted provided that the following conditions are 13 // met: 14 // 15 // * Redistributions of source code must retain the above copyright 16 // notice, this list of conditions and the following disclaimer. 17 // * Redistributions in binary form must reproduce the above 18 // copyright notice, this list of conditions and the following disclaimer 19 // in the documentation and/or other materials provided with the 20 // distribution. 21 // * Neither the name of Google Inc. nor the names of its 22 // contributors may be used to endorse or promote products derived from 23 // this software without specific prior written permission. 24 // 25 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 26 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 27 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 28 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 29 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 30 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 31 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 32 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 33 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 34 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 35 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 36 37 // +build !noasm,!gccgo,!safe 38 39 #include "textflag.h" 40 41 // func DdotUnitary(x, y []float64) (sum float64) 42 // This function assumes len(y) >= len(x). 43 TEXT ·DotUnitary(SB), NOSPLIT, $0 44 MOVQ x+0(FP), R8 45 MOVQ x_len+8(FP), DI // n = len(x) 46 MOVQ y+24(FP), R9 47 48 MOVSD $(0.0), X7 // sum = 0 49 MOVSD $(0.0), X8 // sum = 0 50 51 MOVQ $0, SI // i = 0 52 SUBQ $4, DI // n -= 4 53 JL tail_uni // if n < 0 goto tail_uni 54 55 loop_uni: 56 // sum += x[i] * y[i] unrolled 4x. 57 MOVUPD 0(R8)(SI*8), X0 58 MOVUPD 0(R9)(SI*8), X1 59 MOVUPD 16(R8)(SI*8), X2 60 MOVUPD 16(R9)(SI*8), X3 61 MULPD X1, X0 62 MULPD X3, X2 63 ADDPD X0, X7 64 ADDPD X2, X8 65 66 ADDQ $4, SI // i += 4 67 SUBQ $4, DI // n -= 4 68 JGE loop_uni // if n >= 0 goto loop_uni 69 70 tail_uni: 71 ADDQ $4, DI // n += 4 72 JLE end_uni // if n <= 0 goto end_uni 73 74 onemore_uni: 75 // sum += x[i] * y[i] for the remaining 1-3 elements. 76 MOVSD 0(R8)(SI*8), X0 77 MOVSD 0(R9)(SI*8), X1 78 MULSD X1, X0 79 ADDSD X0, X7 80 81 ADDQ $1, SI // i++ 82 SUBQ $1, DI // n-- 83 JNZ onemore_uni // if n != 0 goto onemore_uni 84 85 end_uni: 86 // Add the four sums together. 87 ADDPD X8, X7 88 MOVSD X7, X0 89 UNPCKHPD X7, X7 90 ADDSD X0, X7 91 MOVSD X7, sum+48(FP) // Return final sum. 92 RET 93 94 // func DdotInc(x, y []float64, n, incX, incY, ix, iy uintptr) (sum float64) 95 TEXT ·DotInc(SB), NOSPLIT, $0 96 MOVQ x+0(FP), R8 97 MOVQ y+24(FP), R9 98 MOVQ n+48(FP), CX 99 MOVQ incX+56(FP), R11 100 MOVQ incY+64(FP), R12 101 MOVQ ix+72(FP), R13 102 MOVQ iy+80(FP), R14 103 104 MOVSD $(0.0), X7 // sum = 0 105 LEAQ (R8)(R13*8), SI // p = &x[ix] 106 LEAQ (R9)(R14*8), DI // q = &y[ix] 107 SHLQ $3, R11 // incX *= sizeof(float64) 108 SHLQ $3, R12 // indY *= sizeof(float64) 109 110 SUBQ $2, CX // n -= 2 111 JL tail_inc // if n < 0 goto tail_inc 112 113 loop_inc: 114 // sum += *p * *q unrolled 2x. 115 MOVHPD (SI), X0 116 MOVHPD (DI), X1 117 ADDQ R11, SI // p += incX 118 ADDQ R12, DI // q += incY 119 MOVLPD (SI), X0 120 MOVLPD (DI), X1 121 ADDQ R11, SI // p += incX 122 ADDQ R12, DI // q += incY 123 124 MULPD X1, X0 125 ADDPD X0, X7 126 127 SUBQ $2, CX // n -= 2 128 JGE loop_inc // if n >= 0 goto loop_inc 129 130 tail_inc: 131 ADDQ $2, CX // n += 2 132 JLE end_inc // if n <= 0 goto end_inc 133 134 // sum += *p * *q for the last iteration if n is odd. 135 MOVSD (SI), X0 136 MULSD (DI), X0 137 ADDSD X0, X7 138 139 end_inc: 140 // Add the two sums together. 141 MOVSD X7, X0 142 UNPCKHPD X7, X7 143 ADDSD X0, X7 144 MOVSD X7, sum+88(FP) // Return final sum. 145 RET