gonum.org/v1/gonum@v0.14.0/internal/asm/f64/scalinc_amd64.s (about)

     1  // Copyright ©2016 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 LEN CX
    43  #define TAIL BX
    44  #define INC_X R8
    45  #define INCx3_X R9
    46  #define ALPHA X0
    47  #define ALPHA_2 X1
    48  
    49  // func ScalInc(alpha float64, x []float64, n, incX uintptr)
    50  TEXT ·ScalInc(SB), NOSPLIT, $0
    51  	MOVSD alpha+0(FP), ALPHA  // ALPHA = alpha
    52  	MOVQ  x_base+8(FP), X_PTR // X_PTR = &x
    53  	MOVQ  incX+40(FP), INC_X  // INC_X = incX
    54  	SHLQ  $3, INC_X           // INC_X *= sizeof(float64)
    55  	MOVQ  n+32(FP), LEN       // LEN = n
    56  	CMPQ  LEN, $0
    57  	JE    end                 // if LEN == 0 { return }
    58  
    59  	MOVQ LEN, TAIL
    60  	ANDQ $3, TAIL   // TAIL = LEN % 4
    61  	SHRQ $2, LEN    // LEN = floor( LEN / 4 )
    62  	JZ   tail_start // if LEN == 0 { goto tail_start }
    63  
    64  	MOVUPS ALPHA, ALPHA_2            // ALPHA_2 = ALPHA for pipelining
    65  	LEAQ   (INC_X)(INC_X*2), INCx3_X // INCx3_X = INC_X * 3
    66  
    67  loop:  // do { // x[i] *= alpha unrolled 4x.
    68  	MOVSD (X_PTR), X2            // X_i = x[i]
    69  	MOVSD (X_PTR)(INC_X*1), X3
    70  	MOVSD (X_PTR)(INC_X*2), X4
    71  	MOVSD (X_PTR)(INCx3_X*1), X5
    72  
    73  	MULSD ALPHA, X2   // X_i *= a
    74  	MULSD ALPHA_2, X3
    75  	MULSD ALPHA, X4
    76  	MULSD ALPHA_2, X5
    77  
    78  	MOVSD X2, (X_PTR)            // x[i] = X_i
    79  	MOVSD X3, (X_PTR)(INC_X*1)
    80  	MOVSD X4, (X_PTR)(INC_X*2)
    81  	MOVSD X5, (X_PTR)(INCx3_X*1)
    82  
    83  	LEAQ (X_PTR)(INC_X*4), X_PTR // X_PTR = &(X_PTR[incX*4])
    84  	DECQ LEN
    85  	JNZ  loop                    // } while --LEN > 0
    86  	CMPQ TAIL, $0
    87  	JE   end                     // if TAIL == 0 { return }
    88  
    89  tail_start: // Reset loop registers
    90  	MOVQ TAIL, LEN // Loop counter: LEN = TAIL
    91  	SHRQ $1, LEN   // LEN = floor( LEN / 2 )
    92  	JZ   tail_one
    93  
    94  tail_two: // do {
    95  	MOVSD (X_PTR), X2          // X_i = x[i]
    96  	MOVSD (X_PTR)(INC_X*1), X3
    97  	MULSD ALPHA, X2            // X_i *= a
    98  	MULSD ALPHA, X3
    99  	MOVSD X2, (X_PTR)          // x[i] = X_i
   100  	MOVSD X3, (X_PTR)(INC_X*1)
   101  
   102  	LEAQ (X_PTR)(INC_X*2), X_PTR // X_PTR = &(X_PTR[incX*2])
   103  
   104  	ANDQ $1, TAIL
   105  	JZ   end
   106  
   107  tail_one:
   108  	MOVSD (X_PTR), X2 // X_i = x[i]
   109  	MULSD ALPHA, X2   // X_i *= ALPHA
   110  	MOVSD X2, (X_PTR) // x[i] = X_i
   111  
   112  end:
   113  	RET