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

/* 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"

#define STACK	16

#define ARGS	 0

#define STACK_M		 4 + STACK + ARGS(%esp)

#define STACK_ALPHA_R	16 + STACK + ARGS(%esp)

#define STACK_ALPHA_I	24 + STACK + ARGS(%esp)

#define STACK_X		32 + STACK + ARGS(%esp)

#define STACK_INCX	36 + STACK + ARGS(%esp)

#define STACK_Y		40 + STACK + ARGS(%esp)

#define STACK_INCY	44 + STACK + ARGS(%esp)

#define M	%ebx

#define X	%esi

#define	INCX	%ecx

#define Y	%edi

#define INCY	%edx

#define YY	%ebp

#define ALPHA_R	%xmm6

#define ALPHA_I	%xmm7

#if defined(HAVE_SSE3) && !defined(CORE_OPTERON)

#define MOVDDUP(a, b, c)	movddup	a(b), c

#define MOVDDUP2(a, b, c)	movddup	a##b, c

#else

#define MOVDDUP(a, b, c)	movlpd	a(b), c;movhpd	a(b), c

#define MOVDDUP2(a, b, c)	movlpd	a##b, c;movhpd	a##b, c

#endif

#include "l1param.h"

	PROLOGUE

	pushl	%ebp

	pushl	%edi

	pushl	%esi

	pushl	%ebx

	PROFCODE

	movl	STACK_M,    M

	movsd	STACK_ALPHA_R, %xmm0

	movsd	STACK_ALPHA_I, %xmm1

	movl	STACK_X,    X

	movl	STACK_INCX, INCX

	movl	STACK_Y,    Y

	movl	STACK_INCY, INCY

	sall	$ZBASE_SHIFT, INCX

	sall	$ZBASE_SHIFT, INCY

	testl	M, M

	jle	.L999

	cmpl	$2 * SIZE, INCX

	jne	.L50

	cmpl	$2 * SIZE, INCY

	jne	.L50

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	pcmpeqb	%xmm5, %xmm5

	psllq	$63,   %xmm5

#ifdef HAVE_SSE3

	movddup	%xmm0, ALPHA_R

	movddup	%xmm1, ALPHA_I

#else

	pshufd	$0x44, %xmm0, ALPHA_R

	pshufd	$0x44, %xmm1, ALPHA_I

#endif

#ifndef CONJ

	shufps	$0x0c, %xmm5, %xmm5	

	xorpd	%xmm5, ALPHA_I

#else

	shufps	$0xc0, %xmm5, %xmm5	

	xorpd	%xmm5, ALPHA_R

#endif

	testl	$SIZE, Y

	jne	.L30

	testl	$SIZE, X

	jne	.L20

	movl	M,  %eax

	sarl	$3, %eax

	jle	.L15

	movaps	-16 * SIZE(X), %xmm0

	movaps	-14 * SIZE(X), %xmm1

	movaps	-12 * SIZE(X), %xmm2

	movaps	-10 * SIZE(X), %xmm3

	decl	%eax

	jle	.L12

	ALIGN_3

.L11:

#ifdef PREFETCHW

	PREFETCHW (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movaps	 -8 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movaps	 -6 * SIZE(X), %xmm1

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movaps	 -4 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movaps	 -2 * SIZE(X), %xmm3

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

	PREFETCHW (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	 -8 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	movaps	  0 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	 -6 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	movaps	  2 * SIZE(X), %xmm1

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

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	 -4 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	movaps	  4 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	 -2 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	movaps	  6 * SIZE(X), %xmm3

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	decl	%eax

	jg	.L11

	ALIGN_3

.L12:

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movaps	 -8 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movaps	 -6 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movaps	 -4 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movaps	 -2 * SIZE(X), %xmm3

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	 -8 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	 -6 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	 -4 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	 -2 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	ALIGN_3

.L15:

	movl	M,  %eax

	andl	$4, %eax

	jle	.L16

	movaps	-16 * SIZE(X), %xmm0

	movaps	-14 * SIZE(X), %xmm1

	movaps	-12 * SIZE(X), %xmm2

	movaps	-10 * SIZE(X), %xmm3

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	addl	$8 * SIZE, X

	addl	$8 * SIZE, Y

	ALIGN_3

.L16:

	movl	M,  %eax

	andl	$2, %eax

	jle	.L17

	movaps	-16 * SIZE(X), %xmm0

	movaps	-14 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	addl	$4 * SIZE, X

	addl	$4 * SIZE, Y

	ALIGN_3

.L17:

	movl	M,  %eax

	andl	$1, %eax

	jle	.L999

	movaps	-16 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	jmp	.L999

	ALIGN_3

.L20:

	movl	M,  %eax

	sarl	$3, %eax

	jle	.L25

	movsd	-16 * SIZE(X), %xmm0

	movhps	-15 * SIZE(X), %xmm0

	movsd	-14 * SIZE(X), %xmm1

	movhps	-13 * SIZE(X), %xmm1

	movsd	-12 * SIZE(X), %xmm2

	movhps	-11 * SIZE(X), %xmm2

	movsd	-10 * SIZE(X), %xmm3

	movhps	 -9 * SIZE(X), %xmm3

	decl	%eax

	jle	.L22

	ALIGN_3

.L21:

#ifdef PREFETCHW

	PREFETCHW (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movsd	 -8 * SIZE(X), %xmm0

	movhps	 -7 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movsd	 -6 * SIZE(X), %xmm1

	movhps	 -5 * SIZE(X), %xmm1

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movsd	 -4 * SIZE(X), %xmm2

	movhps	 -3 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movsd	 -2 * SIZE(X), %xmm3

	movhps	 -1 * SIZE(X), %xmm3

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

	PREFETCHW (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	 -8 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	movsd	  0 * SIZE(X), %xmm0

	movhps	  1 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	 -6 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	movsd	  2 * SIZE(X), %xmm1

	movhps	  3 * SIZE(X), %xmm1

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

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	 -4 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	movsd	  4 * SIZE(X), %xmm2

	movhps	  5 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	 -2 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	movsd	  6 * SIZE(X), %xmm3

	movhps	  7 * SIZE(X), %xmm3

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	decl	%eax

	jg	.L21

	ALIGN_3

.L22:

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movsd	 -8 * SIZE(X), %xmm0

	movhps	 -7 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movsd	 -6 * SIZE(X), %xmm1

	movhps	 -5 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movsd	 -4 * SIZE(X), %xmm2

	movhps	 -3 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movsd	 -2 * SIZE(X), %xmm3

	movhps	 -1 * SIZE(X), %xmm3

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	 -8 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	 -6 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	 -4 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	 -2 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	ALIGN_3

.L25:

	movl	M,  %eax

	andl	$4, %eax

	jle	.L26

	movsd	-16 * SIZE(X), %xmm0

	movhps	-15 * SIZE(X), %xmm0

	movsd	-14 * SIZE(X), %xmm1

	movhps	-13 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movsd	-12 * SIZE(X), %xmm2

	movhps	-11 * SIZE(X), %xmm2

	movsd	-10 * SIZE(X), %xmm3

	movhps	 -9 * SIZE(X), %xmm3

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	-12 * SIZE(Y), %xmm2

	addpd	%xmm5,   %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	-10 * SIZE(Y), %xmm3

	addpd	%xmm5,   %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	addl	$8 * SIZE, X

	addl	$8 * SIZE, Y

	ALIGN_3

.L26:

	movl	M,  %eax

	andl	$2, %eax

	jle	.L27

	movsd	-16 * SIZE(X), %xmm0

	movhps	-15 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movsd	-14 * SIZE(X), %xmm1

	movhps	-13 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	-14 * SIZE(Y), %xmm1

	addpd	%xmm5,   %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	addl	$4 * SIZE, X

	addl	$4 * SIZE, Y

	ALIGN_3

.L27:

	movl	M,  %eax

	andl	$1, %eax

	jle	.L999

	movsd	-16 * SIZE(X), %xmm0

	movhps	-15 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	-16 * SIZE(Y), %xmm0

	addpd	%xmm5,   %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	jmp	.L999

	ALIGN_3

.L30:

	testl	$SIZE, X

	jne	.L40

	movaps	-16 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	xorps	%xmm0,   %xmm0

	SHUFPD_1 %xmm1,  %xmm0

	xorps	%xmm4,   %xmm4

	movhps	-16 * SIZE(Y), %xmm4

	addpd	%xmm0,   %xmm4

	movhps	%xmm4, -16 * SIZE(Y)

	movaps	%xmm1,   %xmm0

	addl	$2 * SIZE, X

	addl	$1 * SIZE, Y

	decl	M

	jle	.L39

	movl	M,  %eax

	sarl	$3, %eax

	jle	.L35

	movaps	-16 * SIZE(X), %xmm1

	movaps	-14 * SIZE(X), %xmm2

	movaps	-12 * SIZE(X), %xmm3

	decl	%eax

	jle	.L32

	ALIGN_3

.L31:

#ifdef PREFETCHW

	PREFETCHW (PREFETCHSIZE +  0) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	SHUFPD_1 %xmm1,  %xmm0

	addpd	-16 * SIZE(Y), %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movaps	-10 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm2

	SHUFPD_1 %xmm2,  %xmm1

	addpd	-14 * SIZE(Y), %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movaps	 -8 * SIZE(X), %xmm1

#ifdef PREFETCH

	PREFETCH (PREFETCHSIZE +  0) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm3

	SHUFPD_1 %xmm3,  %xmm2

	addpd	-12 * SIZE(Y), %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movaps	 -6 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm0

	SHUFPD_1 %xmm0,  %xmm3

	addpd	-10 * SIZE(Y), %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movaps	 -4 * SIZE(X), %xmm3

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

	PREFETCHW (PREFETCHSIZE +  64) - PREOFFSET(Y)

#endif

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	SHUFPD_1 %xmm1,  %xmm0

	addpd	 -8 * SIZE(Y), %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	movaps	 -2 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm2

	SHUFPD_1 %xmm2,  %xmm1

	addpd	 -6 * SIZE(Y), %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	movaps	  0 * SIZE(X), %xmm1

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

	PREFETCH (PREFETCHSIZE +  64) - PREOFFSET(X)

#endif

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm3

	SHUFPD_1 %xmm3,  %xmm2

	addpd	 -4 * SIZE(Y), %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	movaps	  2 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm0

	SHUFPD_1 %xmm0,  %xmm3

	addpd	 -2 * SIZE(Y), %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	movaps	  4 * SIZE(X), %xmm3

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	decl	%eax

	jg	.L31

	ALIGN_3

.L32:

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	SHUFPD_1 %xmm1,  %xmm0

	addpd	-16 * SIZE(Y), %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	movaps	-10 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm2

	SHUFPD_1 %xmm2,  %xmm1

	addpd	-14 * SIZE(Y), %xmm1

	movaps	%xmm1, -14 * SIZE(Y)

	movaps	 -8 * SIZE(X), %xmm1

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm3

	SHUFPD_1 %xmm3,  %xmm2

	addpd	-12 * SIZE(Y), %xmm2

	movaps	%xmm2, -12 * SIZE(Y)

	movaps	 -6 * SIZE(X), %xmm2

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm0

	SHUFPD_1 %xmm0,  %xmm3

	addpd	-10 * SIZE(Y), %xmm3

	movaps	%xmm3, -10 * SIZE(Y)

	movaps	 -4 * SIZE(X), %xmm3

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	SHUFPD_1 %xmm1,  %xmm0

	addpd	 -8 * SIZE(Y), %xmm0

	movaps	%xmm0,  -8 * SIZE(Y)

	movaps	 -2 * SIZE(X), %xmm0

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm2

	SHUFPD_1 %xmm2,  %xmm1

	addpd	 -6 * SIZE(Y), %xmm1

	movaps	%xmm1,  -6 * SIZE(Y)

	pshufd	$0x4e,   %xmm3, %xmm5

	mulpd	ALPHA_R, %xmm3

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm3

	SHUFPD_1 %xmm3,  %xmm2

	addpd	 -4 * SIZE(Y), %xmm2

	movaps	%xmm2,  -4 * SIZE(Y)

	pshufd	$0x4e,   %xmm0, %xmm5

	mulpd	ALPHA_R, %xmm0

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm0

	SHUFPD_1 %xmm0,  %xmm3

	addpd	 -2 * SIZE(Y), %xmm3

	movaps	%xmm3,  -2 * SIZE(Y)

	subl	$-16 * SIZE, X

	subl	$-16 * SIZE, Y

	ALIGN_3

.L35:

	movl	M,  %eax

	andl	$4, %eax

	jle	.L36

	movaps	-16 * SIZE(X), %xmm1

	movaps	-14 * SIZE(X), %xmm2

	movaps	-12 * SIZE(X), %xmm3

	movaps	-10 * SIZE(X), %xmm4

	pshufd	$0x4e,   %xmm1, %xmm5

	mulpd	ALPHA_R, %xmm1

	mulpd	ALPHA_I, %xmm5

	addpd	%xmm5,   %xmm1

	SHUFPD_1 %xmm1,  %xmm0

	addpd	-16 * SIZE(Y), %xmm0

	movaps	%xmm0, -16 * SIZE(Y)

	pshufd	$0x4e,   %xmm2, %xmm5

	mulpd	ALPHA_R, %xmm2