/*********************************************************************/

/* Copyright 2009, 2010 The University of Texas at Austin.           */

/* All rights reserved.                                              */

/*                                                                   */

/* Redistribution and use in source and binary forms, with or        */

/* without modification, are permitted provided that the following   */

/* conditions are met:                                               */

/*                                                                   */

/*   1. Redistributions of source code must retain the above         */

/*      copyright notice, this list of conditions and the following  */

/*      disclaimer.                                                  */

/*                                                                   */

/*   2. Redistributions in binary form must reproduce the above      */

/*      copyright notice, this list of conditions and the following  */

/*      disclaimer in the documentation and/or other materials       */

/*      provided with the distribution.                              */

/*                                                                   */

/*    THIS  SOFTWARE IS PROVIDED  BY THE  UNIVERSITY OF  TEXAS AT    */

/*    AUSTIN  ``AS IS''  AND ANY  EXPRESS OR  IMPLIED WARRANTIES,    */

/*    INCLUDING, BUT  NOT LIMITED  TO, THE IMPLIED  WARRANTIES OF    */

/*    MERCHANTABILITY  AND FITNESS FOR  A PARTICULAR  PURPOSE ARE    */

/*    DISCLAIMED.  IN  NO EVENT SHALL THE UNIVERSITY  OF TEXAS AT    */

/*    AUSTIN OR CONTRIBUTORS BE  LIABLE FOR ANY DIRECT, INDIRECT,    */

/*    INCIDENTAL,  SPECIAL, EXEMPLARY,  OR  CONSEQUENTIAL DAMAGES    */

/*    (INCLUDING, BUT  NOT LIMITED TO,  PROCUREMENT OF SUBSTITUTE    */

/*    GOODS  OR  SERVICES; LOSS  OF  USE,  DATA,  OR PROFITS;  OR    */

/*    BUSINESS INTERRUPTION) HOWEVER CAUSED  AND ON ANY THEORY OF    */

/*    LIABILITY, WHETHER  IN CONTRACT, STRICT  LIABILITY, OR TORT    */

/*    (INCLUDING NEGLIGENCE OR OTHERWISE)  ARISING IN ANY WAY OUT    */

/*    OF  THE  USE OF  THIS  SOFTWARE,  EVEN  IF ADVISED  OF  THE    */

/*    POSSIBILITY OF SUCH DAMAGE.                                    */

/*                                                                   */

/* The views and conclusions contained in the software and           */

/* documentation are those of the authors and should not be          */

/* interpreted as representing official policies, either expressed   */

/* or implied, of The University of Texas at Austin.                 */

/*********************************************************************/

#define ASSEMBLER

#include "common.h"

#ifndef WINDOWS_ABI

#define N	ARG1	/* rdi */

#define X	ARG2	/* rsi */

#define INCX	ARG3	/* rdx */

#define Y	ARG4	/* rcx */

#define INCY	ARG5	/* r8  */

#else

#define RESULT_ADDRESS	ARG1    /*rcx*/

#define N	ARG2	/* rdx */

#define X	ARG3	/* r8 */

#define INCX	ARG4	/* r9*/

#define Y	%r10	

#define INCY	%r11

#endif

#include "l1param.h"

#undef movsd

#ifndef OPTERON

#define MOVLPS	movsd

#else

#define MOVLPS	movlps

#endif

	PROLOGUE

	PROFCODE

#ifdef WINDOWS_ABI

	movq	40(%rsp), Y

	movq	48(%rsp), INCY

#endif

	SAVEREGISTERS

	salq	$ZBASE_SHIFT, INCX

	salq	$ZBASE_SHIFT, INCY

	xorps	%xmm0, %xmm0

	xorps	%xmm1, %xmm1

	xorps	%xmm2, %xmm2

	xorps	%xmm3, %xmm3

	cmpq	$0, N

	jle	.L999

	cmpq	$2 * SIZE, INCX

	jne	.L50

	cmpq	$2 * SIZE, INCY

	jne	.L50

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	testq	$SIZE, Y

	jne	.L30

	testq	$SIZE, X

	jne	.L20

	movq	N,  %rax

	sarq	$3, %rax

	jle	.L15

	movaps	-16 * SIZE(X), %xmm4

	movaps	-14 * SIZE(X), %xmm5

	movaps	-16 * SIZE(Y), %xmm8

	movaps	-14 * SIZE(Y), %xmm9

	movaps	-12 * SIZE(X), %xmm6

	movaps	-10 * SIZE(X), %xmm7

	movaps	-12 * SIZE(Y), %xmm10

	movaps	-10 * SIZE(Y), %xmm11

	decq	%rax

	jle	.L12

	ALIGN_3

.L11:

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	movaps	 -8 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	movaps	 -6 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	movaps	 -4 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	movaps	 -2 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	  0 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	movaps	  0 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	  2 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	movaps	  2 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	  4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	movaps	  4 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	  6 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	movaps	  6 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	decq	%rax

	jg	.L11

	ALIGN_3

.L12:

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	movaps	 -8 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	movaps	 -6 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	movaps	 -4 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	movaps	 -2 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	ALIGN_3

.L15:

	testq	$4, N

	jle	.L16

	movaps	-16 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	movaps	-14 * SIZE(X), %xmm5

	movaps	-14 * SIZE(Y), %xmm9

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	movaps	-12 * SIZE(X), %xmm6

	movaps	-12 * SIZE(Y), %xmm10

	movaps	-10 * SIZE(X), %xmm7

	movaps	-10 * SIZE(Y), %xmm11

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	addq	$8 * SIZE, X

	addq	$8 * SIZE, Y

	ALIGN_3

.L16:

	testq	$2, N

	jle	.L17

	movaps	-16 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	movaps	-14 * SIZE(X), %xmm5

	movaps	-14 * SIZE(Y), %xmm9

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	addq	$4 * SIZE, X

	addq	$4 * SIZE, Y

	ALIGN_3

.L17:

	testq	$1, N

	jle	.L98

	movaps	-16 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	jmp	.L98

	ALIGN_3

.L20:

	movq	N,  %rax

	sarq	$3, %rax

	jle	.L25

	MOVLPS	-16 * SIZE(X), %xmm4

	movhps	-15 * SIZE(X), %xmm4

	MOVLPS	-14 * SIZE(X), %xmm5

	movhps	-13 * SIZE(X), %xmm5

	movaps	-16 * SIZE(Y), %xmm8

	movaps	-14 * SIZE(Y), %xmm9

	MOVLPS	-12 * SIZE(X), %xmm6

	movhps	-11 * SIZE(X), %xmm6

	MOVLPS	-10 * SIZE(X), %xmm7

	movhps	 -9 * SIZE(X), %xmm7

	movaps	-12 * SIZE(Y), %xmm10

	movaps	-10 * SIZE(Y), %xmm11

	decq	%rax

	jle	.L22

	ALIGN_3

.L21:

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	 -8 * SIZE(X), %xmm4

	movhps	 -7 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	 -6 * SIZE(X), %xmm5

	movhps	 -5 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	 -4 * SIZE(X), %xmm6

	movhps	 -3 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	 -2 * SIZE(X), %xmm7

	movhps	 -1 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	  0 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	  0 * SIZE(X), %xmm4

	movhps	  1 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	  2 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	  2 * SIZE(X), %xmm5

	movhps	  3 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	  4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	  4 * SIZE(X), %xmm6

	movhps	  5 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	  6 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	  6 * SIZE(X), %xmm7

	movhps	  7 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	decq	%rax

	jg	.L21

	ALIGN_3

.L22:

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(Y), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	 -8 * SIZE(X), %xmm4

	movhps	 -7 * SIZE(X), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(Y), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	 -6 * SIZE(X), %xmm5

	movhps	 -5 * SIZE(X), %xmm5

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(Y), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	 -4 * SIZE(X), %xmm6

	movhps	 -3 * SIZE(X), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(Y), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	 -2 * SIZE(X), %xmm7

	movhps	 -1 * SIZE(X), %xmm7

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	ALIGN_3

.L25:

	testq	$4, N

	jle	.L26

	MOVLPS	-16 * SIZE(X), %xmm4

	movhps	-15 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-14 * SIZE(X), %xmm5

	movhps	-13 * SIZE(X), %xmm5

	movaps	-14 * SIZE(Y), %xmm9

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	MOVLPS	-12 * SIZE(X), %xmm6

	movhps	-11 * SIZE(X), %xmm6

	movaps	-12 * SIZE(Y), %xmm10

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-10 * SIZE(X), %xmm7

	movhps	 -9 * SIZE(X), %xmm7

	movaps	-10 * SIZE(Y), %xmm11

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	addq	$8 * SIZE, X

	addq	$8 * SIZE, Y

	ALIGN_3

.L26:

	testq	$2, N

	jle	.L27

	MOVLPS	-16 * SIZE(X), %xmm4

	movhps	-15 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-14 * SIZE(X), %xmm5

	movhps	-13 * SIZE(X), %xmm5

	movaps	-14 * SIZE(Y), %xmm9

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	addq	$4 * SIZE, X

	addq	$4 * SIZE, Y

	ALIGN_3

.L27:

	testq	$1, N

	jle	.L98

	MOVLPS	-16 * SIZE(X), %xmm4

	movhps	-15 * SIZE(X), %xmm4

	movaps	-16 * SIZE(Y), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	jmp	.L98

	ALIGN_3

.L30:

	testq	$SIZE, X

	jne	.L40

	movq	N,  %rax

	sarq	$3, %rax

	jle	.L35

	MOVLPS	-16 * SIZE(Y), %xmm4

	movhps	-15 * SIZE(Y), %xmm4

	MOVLPS	-14 * SIZE(Y), %xmm5

	movhps	-13 * SIZE(Y), %xmm5

	movaps	-16 * SIZE(X), %xmm8

	movaps	-14 * SIZE(X), %xmm9

	MOVLPS	-12 * SIZE(Y), %xmm6

	movhps	-11 * SIZE(Y), %xmm6

	MOVLPS	-10 * SIZE(Y), %xmm7

	movhps	 -9 * SIZE(Y), %xmm7

	movaps	-12 * SIZE(X), %xmm10

	movaps	-10 * SIZE(X), %xmm11

	decq	%rax

	jle	.L32

	ALIGN_3

.L31:

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(X), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	 -8 * SIZE(Y), %xmm4

	movhps	 -7 * SIZE(Y), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(X), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	 -6 * SIZE(Y), %xmm5

	movhps	 -5 * SIZE(Y), %xmm5

	addpd	%xmm12, %xmm3

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(X), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	 -4 * SIZE(Y), %xmm6

	movhps	 -3 * SIZE(Y), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(X), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	 -2 * SIZE(Y), %xmm7

	movhps	 -1 * SIZE(Y), %xmm7

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	  0 * SIZE(X), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	  0 * SIZE(Y), %xmm4

	movhps	  1 * SIZE(Y), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	  2 * SIZE(X), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	  2 * SIZE(Y), %xmm5

	movhps	  3 * SIZE(Y), %xmm5

	addpd	%xmm12, %xmm3

#if defined(PREFETCH) && !defined(FETCH128)

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	  4 * SIZE(X), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	  4 * SIZE(Y), %xmm6

	movhps	  5 * SIZE(Y), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	  6 * SIZE(X), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	  6 * SIZE(Y), %xmm7

	movhps	  7 * SIZE(Y), %xmm7

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	decq	%rax

	jg	.L31

	ALIGN_3

.L32:

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	movaps	 -8 * SIZE(X), %xmm8

	mulpd	%xmm4,  %xmm12

	MOVLPS	 -8 * SIZE(Y), %xmm4

	movhps	 -7 * SIZE(Y), %xmm4

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	movaps	 -6 * SIZE(X), %xmm9

	mulpd	%xmm5,  %xmm12

	MOVLPS	 -6 * SIZE(Y), %xmm5

	movhps	 -5 * SIZE(Y), %xmm5

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	movaps	 -4 * SIZE(X), %xmm10

	mulpd	%xmm6,  %xmm12

	MOVLPS	 -4 * SIZE(Y), %xmm6

	movhps	 -3 * SIZE(Y), %xmm6

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	movaps	 -2 * SIZE(X), %xmm11

	mulpd	%xmm7,  %xmm12

	MOVLPS	 -2 * SIZE(Y), %xmm7

	movhps	 -1 * SIZE(Y), %xmm7

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	subq	$-16 * SIZE, X

	subq	$-16 * SIZE, Y

	ALIGN_3

.L35:

	testq	$4, N

	jle	.L36

	MOVLPS	-16 * SIZE(Y), %xmm4

	movhps	-15 * SIZE(Y), %xmm4

	movaps	-16 * SIZE(X), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-14 * SIZE(Y), %xmm5

	movhps	-13 * SIZE(Y), %xmm5

	movaps	-14 * SIZE(X), %xmm9

	pshufd	$0x4e,  %xmm9, %xmm12 

	mulpd	%xmm5,  %xmm9

	addpd	%xmm9,  %xmm2

	mulpd	%xmm5,  %xmm12

	addpd	%xmm12, %xmm3

	MOVLPS	-12 * SIZE(Y), %xmm6

	movhps	-11 * SIZE(Y), %xmm6

	movaps	-12 * SIZE(X), %xmm10

	pshufd	$0x4e,  %xmm10, %xmm12 

	mulpd	%xmm6,  %xmm10

	addpd	%xmm10, %xmm0

	mulpd	%xmm6,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-10 * SIZE(Y), %xmm7

	movhps	 -9 * SIZE(Y), %xmm7

	movaps	-10 * SIZE(X), %xmm11

	pshufd	$0x4e,  %xmm11, %xmm12 

	mulpd	%xmm7,  %xmm11

	addpd	%xmm11, %xmm2

	mulpd	%xmm7,  %xmm12

	addpd	%xmm12, %xmm3

	addq	$8 * SIZE, X

	addq	$8 * SIZE, Y

	ALIGN_3

.L36:

	testq	$2, N

	jle	.L37

	MOVLPS	-16 * SIZE(Y), %xmm4

	movhps	-15 * SIZE(Y), %xmm4

	movaps	-16 * SIZE(X), %xmm8

	pshufd	$0x4e,  %xmm8, %xmm12 

	mulpd	%xmm4,  %xmm8

	addpd	%xmm8,  %xmm0

	mulpd	%xmm4,  %xmm12

	addpd	%xmm12, %xmm1

	MOVLPS	-14 * SIZE(Y), %xmm5

	movhps	-13 * SIZE(Y), %xmm5

	movaps	-14 * SIZE(X), %xmm9