github.com/gopherd/gonum@v0.0.4/internal/asm/f64/scalincto_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 DST_PTR DI
    43  #define LEN CX
    44  #define TAIL BX
    45  #define INC_X R8
    46  #define INCx3_X R9
    47  #define INC_DST R10
    48  #define INCx3_DST R11
    49  #define ALPHA X0
    50  #define ALPHA_2 X1
    51  
    52  // func ScalIncTo(dst []float64, incDst uintptr, alpha float64, x []float64, n, incX uintptr)
    53  TEXT ·ScalIncTo(SB), NOSPLIT, $0
    54  	MOVQ  dst_base+0(FP), DST_PTR // DST_PTR = &dst
    55  	MOVQ  incDst+24(FP), INC_DST  // INC_DST = incDst
    56  	SHLQ  $3, INC_DST             // INC_DST *= sizeof(float64)
    57  	MOVSD alpha+32(FP), ALPHA     // ALPHA = alpha
    58  	MOVQ  x_base+40(FP), X_PTR    // X_PTR = &x
    59  	MOVQ  n+64(FP), LEN           // LEN = n
    60  	MOVQ  incX+72(FP), INC_X      // INC_X = incX
    61  	SHLQ  $3, INC_X               // INC_X *= sizeof(float64)
    62  	CMPQ  LEN, $0
    63  	JE    end                     // if LEN == 0 { return }
    64  
    65  	MOVQ LEN, TAIL
    66  	ANDQ $3, TAIL   // TAIL = LEN % 4
    67  	SHRQ $2, LEN    // LEN = floor( LEN / 4 )
    68  	JZ   tail_start // if LEN == 0 { goto tail_start }
    69  
    70  	MOVUPS ALPHA, ALPHA_2                  // ALPHA_2 = ALPHA for pipelining
    71  	LEAQ   (INC_X)(INC_X*2), INCx3_X       // INCx3_X = INC_X * 3
    72  	LEAQ   (INC_DST)(INC_DST*2), INCx3_DST // INCx3_DST = INC_DST * 3
    73  
    74  loop:  // do { // x[i] *= alpha unrolled 4x.
    75  	MOVSD (X_PTR), X2            // X_i = x[i]
    76  	MOVSD (X_PTR)(INC_X*1), X3
    77  	MOVSD (X_PTR)(INC_X*2), X4
    78  	MOVSD (X_PTR)(INCx3_X*1), X5
    79  
    80  	MULSD ALPHA, X2   // X_i *= a
    81  	MULSD ALPHA_2, X3
    82  	MULSD ALPHA, X4
    83  	MULSD ALPHA_2, X5
    84  
    85  	MOVSD X2, (DST_PTR)              // dst[i] = X_i
    86  	MOVSD X3, (DST_PTR)(INC_DST*1)
    87  	MOVSD X4, (DST_PTR)(INC_DST*2)
    88  	MOVSD X5, (DST_PTR)(INCx3_DST*1)
    89  
    90  	LEAQ (X_PTR)(INC_X*4), X_PTR       // X_PTR = &(X_PTR[incX*4])
    91  	LEAQ (DST_PTR)(INC_DST*4), DST_PTR // DST_PTR = &(DST_PTR[incDst*4])
    92  	DECQ LEN
    93  	JNZ  loop                          // } while --LEN > 0
    94  	CMPQ TAIL, $0
    95  	JE   end                           // if TAIL == 0 { return }
    96  
    97  tail_start: // Reset loop registers
    98  	MOVQ TAIL, LEN // Loop counter: LEN = TAIL
    99  	SHRQ $1, LEN   // LEN = floor( LEN / 2 )
   100  	JZ   tail_one
   101  
   102  tail_two:
   103  	MOVSD (X_PTR), X2              // X_i = x[i]
   104  	MOVSD (X_PTR)(INC_X*1), X3
   105  	MULSD ALPHA, X2                // X_i *= a
   106  	MULSD ALPHA, X3
   107  	MOVSD X2, (DST_PTR)            // dst[i] = X_i
   108  	MOVSD X3, (DST_PTR)(INC_DST*1)
   109  
   110  	LEAQ (X_PTR)(INC_X*2), X_PTR       // X_PTR = &(X_PTR[incX*2])
   111  	LEAQ (DST_PTR)(INC_DST*2), DST_PTR // DST_PTR = &(DST_PTR[incDst*2])
   112  
   113  	ANDQ $1, TAIL
   114  	JZ   end
   115  
   116  tail_one:
   117  	MOVSD (X_PTR), X2   // X_i = x[i]
   118  	MULSD ALPHA, X2     // X_i *= ALPHA
   119  	MOVSD X2, (DST_PTR) // x[i] = X_i
   120  
   121  end:
   122  	RET