gonum.org/v1/gonum@v0.14.0/internal/asm/f64/axpyincto_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 #define X_PTR SI 42 #define Y_PTR DI 43 #define DST_PTR DX 44 #define IDX AX 45 #define LEN CX 46 #define TAIL BX 47 #define INC_X R8 48 #define INCx3_X R11 49 #define INC_Y R9 50 #define INCx3_Y R12 51 #define INC_DST R10 52 #define INCx3_DST R13 53 #define ALPHA X0 54 #define ALPHA_2 X1 55 56 // func AxpyIncTo(dst []float64, incDst, idst uintptr, alpha float64, x, y []float64, n, incX, incY, ix, iy uintptr) 57 TEXT ·AxpyIncTo(SB), NOSPLIT, $0 58 MOVQ dst_base+0(FP), DST_PTR // DST_PTR := &dst 59 MOVQ x_base+48(FP), X_PTR // X_PTR := &x 60 MOVQ y_base+72(FP), Y_PTR // Y_PTR := &y 61 MOVQ n+96(FP), LEN // LEN := n 62 CMPQ LEN, $0 // if LEN == 0 { return } 63 JE end 64 65 MOVQ ix+120(FP), INC_X 66 LEAQ (X_PTR)(INC_X*8), X_PTR // X_PTR = &(x[ix]) 67 MOVQ iy+128(FP), INC_Y 68 LEAQ (Y_PTR)(INC_Y*8), Y_PTR // Y_PTR = &(dst[idst]) 69 MOVQ idst+32(FP), INC_DST 70 LEAQ (DST_PTR)(INC_DST*8), DST_PTR // DST_PTR = &(y[iy]) 71 72 MOVQ incX+104(FP), INC_X // INC_X = incX * sizeof(float64) 73 SHLQ $3, INC_X 74 MOVQ incY+112(FP), INC_Y // INC_Y = incY * sizeof(float64) 75 SHLQ $3, INC_Y 76 MOVQ incDst+24(FP), INC_DST // INC_DST = incDst * sizeof(float64) 77 SHLQ $3, INC_DST 78 MOVSD alpha+40(FP), ALPHA 79 80 MOVQ LEN, TAIL 81 ANDQ $3, TAIL // TAIL = n % 4 82 SHRQ $2, LEN // LEN = floor( n / 4 ) 83 JZ tail_start // if LEN == 0 { goto tail_start } 84 85 MOVSD ALPHA, ALPHA_2 // ALPHA_2 = ALPHA for pipelining 86 LEAQ (INC_X)(INC_X*2), INCx3_X // INCx3_X = INC_X * 3 87 LEAQ (INC_Y)(INC_Y*2), INCx3_Y // INCx3_Y = INC_Y * 3 88 LEAQ (INC_DST)(INC_DST*2), INCx3_DST // INCx3_DST = INC_DST * 3 89 90 loop: // do { // y[i] += alpha * x[i] unrolled 2x. 91 MOVSD (X_PTR), X2 // X_i = x[i] 92 MOVSD (X_PTR)(INC_X*1), X3 93 MOVSD (X_PTR)(INC_X*2), X4 94 MOVSD (X_PTR)(INCx3_X*1), X5 95 96 MULSD ALPHA, X2 // X_i *= a 97 MULSD ALPHA_2, X3 98 MULSD ALPHA, X4 99 MULSD ALPHA_2, X5 100 101 ADDSD (Y_PTR), X2 // X_i += y[i] 102 ADDSD (Y_PTR)(INC_Y*1), X3 103 ADDSD (Y_PTR)(INC_Y*2), X4 104 ADDSD (Y_PTR)(INCx3_Y*1), X5 105 106 MOVSD X2, (DST_PTR) // y[i] = X_i 107 MOVSD X3, (DST_PTR)(INC_DST*1) 108 MOVSD X4, (DST_PTR)(INC_DST*2) 109 MOVSD X5, (DST_PTR)(INCx3_DST*1) 110 111 LEAQ (X_PTR)(INC_X*4), X_PTR // X_PTR = &(X_PTR[incX*4]) 112 LEAQ (Y_PTR)(INC_Y*4), Y_PTR // Y_PTR = &(Y_PTR[incY*4]) 113 LEAQ (DST_PTR)(INC_DST*4), DST_PTR // DST_PTR = &(DST_PTR[incDst*4] 114 DECQ LEN 115 JNZ loop // } while --LEN > 0 116 CMPQ TAIL, $0 // if TAIL == 0 { return } 117 JE end 118 119 tail_start: // Reset Loop registers 120 MOVQ TAIL, LEN // Loop counter: LEN = TAIL 121 SHRQ $1, LEN // LEN = floor( LEN / 2 ) 122 JZ tail_one 123 124 tail_two: 125 MOVSD (X_PTR), X2 // X_i = x[i] 126 MOVSD (X_PTR)(INC_X*1), X3 127 MULSD ALPHA, X2 // X_i *= a 128 MULSD ALPHA, X3 129 ADDSD (Y_PTR), X2 // X_i += y[i] 130 ADDSD (Y_PTR)(INC_Y*1), X3 131 MOVSD X2, (DST_PTR) // y[i] = X_i 132 MOVSD X3, (DST_PTR)(INC_DST*1) 133 134 LEAQ (X_PTR)(INC_X*2), X_PTR // X_PTR = &(X_PTR[incX*2]) 135 LEAQ (Y_PTR)(INC_Y*2), Y_PTR // Y_PTR = &(Y_PTR[incY*2]) 136 LEAQ (DST_PTR)(INC_DST*2), DST_PTR // DST_PTR = &(DST_PTR[incY*2] 137 138 ANDQ $1, TAIL 139 JZ end // if TAIL == 0 { goto end } 140 141 tail_one: 142 MOVSD (X_PTR), X2 // X2 = x[i] 143 MULSD ALPHA, X2 // X2 *= a 144 ADDSD (Y_PTR), X2 // X2 += y[i] 145 MOVSD X2, (DST_PTR) // y[i] = X2 146 147 end: 148 RET