gonum.org/v1/gonum@v0.14.0/internal/asm/f64/scalunitary_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 MOVDDUP_ALPHA    LONG $0x44120FF2; WORD $0x0824 // @ MOVDDUP XMM0, 8[RSP]
    42  
    43  #define X_PTR SI
    44  #define DST_PTR DI
    45  #define IDX AX
    46  #define LEN CX
    47  #define TAIL BX
    48  #define ALPHA X0
    49  #define ALPHA_2 X1
    50  
    51  // func ScalUnitary(alpha float64, x []float64)
    52  TEXT ·ScalUnitary(SB), NOSPLIT, $0
    53  	MOVDDUP_ALPHA            // ALPHA = { alpha, alpha }
    54  	MOVQ x_base+8(FP), X_PTR // X_PTR = &x
    55  	MOVQ x_len+16(FP), LEN   // LEN = len(x)
    56  	CMPQ LEN, $0
    57  	JE   end                 // if LEN == 0 { return }
    58  	XORQ IDX, IDX            // IDX = 0
    59  
    60  	MOVQ LEN, TAIL
    61  	ANDQ $7, TAIL   // TAIL = LEN % 8
    62  	SHRQ $3, LEN    // LEN = floor( LEN / 8 )
    63  	JZ   tail_start // if LEN == 0 { goto tail_start }
    64  
    65  	MOVUPS ALPHA, ALPHA_2
    66  
    67  loop:  // do {  // x[i] *= alpha unrolled 8x.
    68  	MOVUPS (X_PTR)(IDX*8), X2   // X_i = x[i]
    69  	MOVUPS 16(X_PTR)(IDX*8), X3
    70  	MOVUPS 32(X_PTR)(IDX*8), X4
    71  	MOVUPS 48(X_PTR)(IDX*8), X5
    72  
    73  	MULPD ALPHA, X2   // X_i *= ALPHA
    74  	MULPD ALPHA_2, X3
    75  	MULPD ALPHA, X4
    76  	MULPD ALPHA_2, X5
    77  
    78  	MOVUPS X2, (X_PTR)(IDX*8)   // x[i] = X_i
    79  	MOVUPS X3, 16(X_PTR)(IDX*8)
    80  	MOVUPS X4, 32(X_PTR)(IDX*8)
    81  	MOVUPS X5, 48(X_PTR)(IDX*8)
    82  
    83  	ADDQ $8, IDX  // i += 8
    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( TAIL / 2 )
    92  	JZ   tail_one  // if n == 0 goto end
    93  
    94  tail_two: // do {
    95  	MOVUPS (X_PTR)(IDX*8), X2 // X_i = x[i]
    96  	MULPD  ALPHA, X2          // X_i *= ALPHA
    97  	MOVUPS X2, (X_PTR)(IDX*8) // x[i] = X_i
    98  	ADDQ   $2, IDX            // i += 2
    99  	DECQ   LEN
   100  	JNZ    tail_two           // while --LEN > 0
   101  
   102  	ANDQ $1, TAIL
   103  	JZ   end      // if TAIL == 0 { return }
   104  
   105  tail_one:
   106  	// x[i] *= alpha for the remaining element.
   107  	MOVSD (X_PTR)(IDX*8), X2
   108  	MULSD ALPHA, X2
   109  	MOVSD X2, (X_PTR)(IDX*8)
   110  
   111  end:
   112  	RET