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

/* 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 PREFETCH_SIZE (8 * 16 + 4)

#define N	r32

#define X1	r33

#define INCX	r34

#define Y1	r35

#define INCY	r36

#define PREX	r2

#define PREY	r3

#define I	r14

#define J	r15

#define Y2	r16

#define X2	r17

#define INCX16	r18

#define INCY16	r19

#define INCX5	r20

#define INCY5	r21

#define YY	r22

#define XA	r23

#define YA	r24

#define XX	r25

#define PR	r30

#define ARLC	r31

	PROLOGUE

	.prologue

	PROFCODE

	{ .mfi

	nop.m 0

	mov	f8  = f0

	.save ar.lc, ARLC

	mov	ARLC = ar.lc

	}

	{ .mfi

	mov	r26 = 1

	mov	f9  = f0

	shr  XA = X1, 3

	}

	;;

	.body

#ifdef F_INTERFACE

	LDINT	N    = [N]

	LDINT	INCX = [INCX]

	LDINT	INCY = [INCY]

	;;

#ifndef USE64BITINT

	sxt4	N    = N

	sxt4	INCX = INCX

	sxt4	INCY = INCY

	;;

#endif

	cmp.le	p0, p6 = r0, INCX

	cmp.le	p0, p7 = r0, INCY

	sub	r26 = r26, N

	;;

	setf.sig f32 = r26

	setf.sig f33 = INCX

	setf.sig f34 = INCY

	;;

	xmpy.l f33 = f32, f33

	xmpy.l f34 = f32, f34

	;;

	getf.sig r26 = f33

	getf.sig r27 = f34

	;;

	(p6) shladd X1 = r26, BASE_SHIFT, X1

	(p7) shladd Y1 = r27, BASE_SHIFT, Y1

	;;

#endif

	{ .mfi

	shladd	INCX = INCX, BASE_SHIFT, r0

	mov	f32 = f0

	mov	PR = pr

	}

	{ .mfb

	cmp.lt	p0, p6 = r0, N

	mov	f80 = f0

	(p6) br.ret.sptk.many b0

	}

	;;

	{ .mfi

	shladd	INCY = INCY, BASE_SHIFT, r0

	mov	f10 = f0

	tbit.nz	p15, p0 = X1, BASE_SHIFT

	}

	{ .mfb

	cmp.ne	p6, p0 = SIZE, INCX

	mov	f11 = f0

	(p6) br.cond.dptk .L100

	}

	;;

	{ .mfi

	(p15) LDFD f32 = [X1], INCX

	mov	f12 = f0

	mov	pr.rot= 0

	}

	{ .mfi

	(p15) adds N = -1, N

	mov	f13 = f0

	shr  YA = Y1, 3

	}

	;;

	{ .mfi

	(p15) LDFD f80 = [Y1], INCY

	mov	f14 = f0

	shr	I =  N, 4

	}

	{ .mmi

	and	J =  15, N

	and  XA = 0x1f, XA

	and  YA = 0x1f, YA

	}

	;;

	{ .mmi

	shladd	INCX5 = INCX, 2, INCX

	shladd	INCY5 = INCY, 2, INCY

	sub	XA = YA, XA

	}

	{ .mmi

	shladd	INCX16 = INCX, 4, r0

	shladd	INCY16 = INCY, 4, r0

	tbit.z	p0, p12 = N, 3

	}

	;;

	{ .mmi

	shladd	Y2 = INCY, 2, Y1

	cmp.eq	p7, p0  =   r0, J

	mov	ar.ec= 3

	}

	{ .mmi

	adds	I = -1, I

	cmp.ge	p8, p0 = 4, XA

	cmp.eq	p16, p0 = r0, r0

	}

	;;

	{ .mbb

	cmp.le	p9, p0 = 24, XA

	(p8)  br.cond.dpnt  .L20

	(p9)  br.cond.dpnt  .L20

	}

	;;

	{ .mmi

	adds	PREX = PREFETCH_SIZE * SIZE, X1

	adds	PREY = (PREFETCH_SIZE + 6) * SIZE, Y1

	mov	ar.lc = I

	}

	{ .mfb

	cmp.eq	p6 ,p0  =   -1, I

	FMA  f15 = f32, f80, f0

	(p6) br.cond.dpnt  .L15

	}

	;;

	.align 32

/* INCX == 1 && X is aligned */

.L12:

	{ .mmf

	(p16) LDFPD	f32, f35  = [X1], 2 * SIZE

	(p16) lfetch.nt1 [PREX], INCX16

	(p18) FMA	f8  = f34, f82,  f8

	}

	{ .mmf

	(p16) LDFD	f80 = [Y1], INCY

	(p16) LDFD	f92 = [Y2], INCY

	(p18) FMA	f9  = f37, f85,  f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f38,  f41  = [X1], 2 * SIZE

	(p16) lfetch.nt1 [PREY], INCY16

	(p18) FMA	f10 = f40, f88,  f10

	}

	{ .mmf

	(p16) LDFD	f83  = [Y1], INCY

	(p16) LDFD	f95  = [Y2], INCY

	(p18) FMA	f11 = f43, f91,  f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f44,  f47  = [X1], 2 * SIZE

	(p18) FMA	f12 = f46, f94,  f12

	}

	{ .mmf

	(p16) LDFD	f86  = [Y1], INCY

	(p16) LDFD	f98  = [Y2], INCY

	(p18) FMA	f13 = f49, f97,  f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f50,  f53  = [X1], 2 * SIZE

	(p18) FMA	f14 = f52, f100,  f14

	}

	{ .mmf

	(p16) LDFD	f89  = [Y1], INCY5

	(p16) LDFD	f101 = [Y2], INCY5

	(p18) FMA	f15 = f55, f103, f15

	}

	;;

	{ .mmf

	(p16) LDFPD	f56,  f59  = [X1], 2 * SIZE

	(p18) FMA	f8  = f58, f106, f8

	}

	{ .mmf

	(p16) LDFD	f104 = [Y1], INCY

	(p16) LDFD	f116 = [Y2], INCY

	(p18) FMA	f9   = f61, f109, f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f62,  f65  = [X1], 2 * SIZE

	(p18) FMA	f10 = f64, f112, f10

	}

	{ .mmf

	(p16) LDFD	f107 = [Y1], INCY

	(p16) LDFD	f119 = [Y2], INCY

	(p18) FMA	f11 = f67, f115, f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f68,  f71  = [X1], 2 * SIZE

	(p18) FMA	f12 = f70, f118, f12

	}

	{ .mmf

	(p16) LDFD	f110 = [Y1], INCY

	(p16) LDFD	f122 = [Y2], INCY

	(p18) FMA	f13 = f73, f121, f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f74,  f77  = [X1], 2 * SIZE

	(p16) LDFD	f113 = [Y1], INCY5

	(p18) FMA	f14 = f76, f124, f14

	}

	{ .mfb

	(p16) LDFD	f125 = [Y2], INCY5

	(p18) FMA	f15 = f79, f127, f15

	br.ctop.sptk.few .L12

	}

	;;

	.align 32

.L15:

	{ .mmi

	(p12) LDFPD	f32, f33  = [X1], 2 * SIZE

	mov	YY = Y1

	tbit.z	p0, p13 = N, 2

	}

	{ .mmb

	(p12) LDFD	f34 = [Y1], INCY

	(p12) LDFD	f42 = [Y2], INCY

	(p7) br.cond.dptk .L999

	}

	;;

	{ .mmi

	(p12) LDFPD	f36, f37  = [X1], 2 * SIZE

	(p12) shladd YY = INCY, 3, YY

	tbit.z	p0, p14 = N, 1

	}

	{ .mmi

	(p12) LDFD	f35 = [Y1], INCY

	(p12) LDFD	f43 = [Y2], INCY

	tbit.z	p0, p15 = N, 0

	}

	;;

	{ .mmi

	(p12) LDFPD	f40, f41  = [X1], 2 * SIZE

	(p13) shladd YY = INCY, 2, YY

	}

	{ .mmi

	(p12) LDFD	f38 = [Y1], INCY

	(p12) LDFD	f46 = [Y2], INCY

	}

	;;

	(p12) LDFPD	f44, f45  = [X1], 2 * SIZE

	(p12) LDFD	f39 = [Y1], INCY5

	(p12) LDFD	f47 = [Y2], INCY5

	;;

	(p13) LDFPD	f48, f49  = [X1], 2 * SIZE

	(p13) LDFD	f50 = [Y1], INCY

	(p14) LDFD	f58 = [YY], INCY

	;;

	(p13) LDFPD	f52, f53  = [X1], 2 * SIZE

	(p13) LDFD	f51 = [Y1], INCY

	(p14) LDFD	f59 = [YY], INCY

	;;

	(p14) LDFPD	f56, f57  = [X1], 2 * SIZE

	(p13) LDFD	f54 = [Y1], INCY

	(p15) LDFD	f61  = [YY]

	;;

	(p13) LDFD	f55 = [Y1], INCY

	(p15) LDFD	f60  = [X1]

	;;

	(p12) FMA	f8  = f32, f34, f8

	(p12) FMA	f9  = f33, f35, f9

	(p12) FMA	f10 = f36, f38, f10

	(p12) FMA	f11 = f37, f39, f11

	(p12) FMA	f12 = f40, f42, f12

	(p12) FMA	f13 = f41, f43, f13

	(p12) FMA	f14 = f44, f46, f14

	(p12) FMA	f15 = f45, f47, f15

	;;		      

	(p13) FMA	f8  = f48, f50, f8

	(p13) FMA	f9  = f49, f51, f9

	(p13) FMA	f10 = f52, f54, f10

	(p13) FMA	f11 = f53, f55, f11

	(p14) FMA	f12 = f56, f58, f12

	(p14) FMA	f13 = f57, f59, f13

	(p15) FMA	f14 = f60, f61, f14

	br	.L999

	;;

	.align 32

.L20:

	{ .mmi

	adds	PREX = PREFETCH_SIZE * SIZE, X1

	adds	PREY = (PREFETCH_SIZE + 38) * SIZE, Y1

	mov	ar.lc = I

	}

	{ .mfb

	cmp.eq	p6 ,p0  =   -1, I

	FMA  f15 = f32, f80, f0

	(p6) br.cond.dpnt  .L25

	}

	;;

	.align 32

.L22:

	{ .mmf

	(p16) LDFPD	f32, f35  = [X1], 2 * SIZE

	(p16) lfetch.nt1 [PREX], INCX16

	(p18) FMA	f8  = f34, f82,  f8

	}

	{ .mmf

	(p17) LDFD	f81 = [Y1], INCY

	(p17) LDFD	f93 = [Y2], INCY

	(p18) FMA	f9  = f37, f85,  f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f38,  f41  = [X1], 2 * SIZE

	(p16) lfetch.nt1 [PREY], INCY16

	(p18) FMA	f10 = f40, f88,  f10

	}

	{ .mmf

	(p17) LDFD	f84  = [Y1], INCY

	(p17) LDFD	f96  = [Y2], INCY

	(p18) FMA	f11 = f43, f91,  f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f44,  f47  = [X1], 2 * SIZE

	(p18) FMA	f12 = f46, f94,  f12

	}

	{ .mmf

	(p17) LDFD	f87  = [Y1], INCY

	(p17) LDFD	f99  = [Y2], INCY

	(p18) FMA	f13 = f49, f97,  f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f50,  f53  = [X1], 2 * SIZE

	(p18) FMA	f14 = f52, f100,  f14

	}

	{ .mmf

	(p17) LDFD	f90  = [Y1], INCY5

	(p17) LDFD	f102 = [Y2], INCY5

	(p18) FMA	f15 = f55, f103, f15

	}

	;;

	{ .mmf

	(p16) LDFPD	f56,  f59  = [X1], 2 * SIZE

	(p18) FMA	f8  = f58, f106, f8

	}

	{ .mmf

	(p17) LDFD	f105 = [Y1], INCY

	(p17) LDFD	f117 = [Y2], INCY

	(p18) FMA	f9   = f61, f109, f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f62,  f65  = [X1], 2 * SIZE

	(p18) FMA	f10 = f64, f112, f10

	}

	{ .mmf

	(p17) LDFD	f108 = [Y1], INCY

	(p17) LDFD	f120 = [Y2], INCY

	(p18) FMA	f11 = f67, f115, f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f68,  f71  = [X1], 2 * SIZE

	(p18) FMA	f12 = f70, f118, f12

	}

	{ .mmf

	(p17) LDFD	f111 = [Y1], INCY

	(p17) LDFD	f123 = [Y2], INCY

	(p18) FMA	f13 = f73, f121, f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f74,  f77  = [X1], 2 * SIZE

	(p17) LDFD	f114 = [Y1], INCY5

	(p18) FMA	f14 = f76, f124, f14

	}

	{ .mfb

	(p17) LDFD	f126 = [Y2], INCY5

	(p18) FMA	f15 = f79, f127, f15

	br.ctop.sptk.few .L22

	}

	;;

	.align 32

.L25:

	{ .mmi

	(p12) LDFPD	f32, f33  = [X1], 2 * SIZE

	mov	YY = Y1

	tbit.z	p0, p13 = N, 2

	}

	{ .mmb

	(p12) LDFD	f34 = [Y1], INCY

	(p12) LDFD	f42 = [Y2], INCY

	(p7) br.cond.dptk .L999

	}

	;;

	{ .mmi

	(p12) LDFPD	f36, f37  = [X1], 2 * SIZE

	(p12) shladd YY = INCY, 3, YY

	tbit.z	p0, p14 = N, 1

	}

	{ .mmi

	(p12) LDFD	f35 = [Y1], INCY

	(p12) LDFD	f43 = [Y2], INCY

	tbit.z	p0, p15 = N, 0

	}

	;;

	{ .mmi

	(p12) LDFPD	f40, f41  = [X1], 2 * SIZE

	(p13) shladd YY = INCY, 2, YY

	}

	{ .mmi

	(p12) LDFD	f38 = [Y1], INCY

	(p12) LDFD	f46 = [Y2], INCY

	}

	;;

	(p12) LDFPD	f44, f45  = [X1], 2 * SIZE

	(p12) LDFD	f39 = [Y1], INCY5

	(p12) LDFD	f47 = [Y2], INCY5

	;;

	(p13) LDFPD	f48, f49  = [X1], 2 * SIZE

	(p13) LDFD	f50 = [Y1], INCY

	(p14) LDFD	f58 = [YY], INCY

	;;

	(p13) LDFPD	f52, f53  = [X1], 2 * SIZE

	(p13) LDFD	f51 = [Y1], INCY

	(p14) LDFD	f59 = [YY], INCY

	;;

	(p14) LDFPD	f56, f57  = [X1], 2 * SIZE

	(p13) LDFD	f54 = [Y1], INCY

	(p15) LDFD	f61  = [YY]

	;;

	(p13) LDFD	f55 = [Y1], INCY

	(p15) LDFD	f60  = [X1]

	;;

	(p12) FMA	f8  = f32, f34, f8

	(p12) FMA	f9  = f33, f35, f9

	(p12) FMA	f10 = f36, f38, f10

	(p12) FMA	f11 = f37, f39, f11

	(p12) FMA	f12 = f40, f42, f12

	(p12) FMA	f13 = f41, f43, f13

	(p12) FMA	f14 = f44, f46, f14

	(p12) FMA	f15 = f45, f47, f15

	;;		      

	(p13) FMA	f8  = f48, f50, f8

	(p13) FMA	f9  = f49, f51, f9

	(p13) FMA	f10 = f52, f54, f10

	(p13) FMA	f11 = f53, f55, f11

	(p14) FMA	f12 = f56, f58, f12

	(p14) FMA	f13 = f57, f59, f13

	(p15) FMA	f14 = f60, f61, f14

	br	.L999

	;;

	.align 32

.L100:

	{ .mmi

	shladd	X2 = INCX, 2, X1

	}

	{ .mib

	cmp.ne	p6, p0 = SIZE, INCY

	tbit.nz	p15, p0 = Y1, BASE_SHIFT

	(p6) br.cond.dptk .L200

	}

	;;

	{ .mfi

	(p15) LDFD f32 = [X1], INCX

	mov	f12 = f0

	mov	pr.rot= 0

	}

	{ .mfi

	(p15) adds N = -1, N

	mov	f13 = f0

	shr  YA = Y1, 3

	}

	;;

	{ .mfi

	(p15) LDFD f80 = [Y1], INCY

	mov	f14 = f0

	shr	I =  N, 4

	}

	{ .mmi

	and	J =  15, N

	and  XA = 0x1f, XA

	and  YA = 0x1f, YA

	}

	;;

	{ .mmi

	shladd	INCX5 = INCX, 2, INCX

	shladd	INCY5 = INCY, 2, INCY

	sub	XA = YA, XA

	}

	{ .mmi

	shladd	INCX16 = INCX, 4, r0

	shladd	INCY16 = INCY, 4, r0

	tbit.z	p0, p12 = N, 3

	}

	;;

	{ .mmi

	shladd	X2 = INCX, 2, X1

	cmp.eq	p7, p0  =   r0, J

	mov	ar.ec= 3

	}

	{ .mmi

	adds	I = -1, I

	cmp.ge	p8, p0 = 8, XA

	cmp.eq	p16, p0 = r0, r0

	}

	;;

	{ .mbb

	cmp.le	p9, p0 = 28, XA

	(p8)  br.cond.dpnt  .L120

	(p9)  br.cond.dpnt  .L120

	}

	;;

	{ .mmi

	adds	PREX = (PREFETCH_SIZE + 5) * SIZE, X1

	adds	PREY = (PREFETCH_SIZE + 3) * SIZE, Y1

	mov	ar.lc = I

	}

	{ .mfb

	cmp.eq	p6 ,p0  =   -1, I

	FMA  f15 = f32, f80, f0

	(p6) br.cond.dpnt  .L115

	}

	;;

	.align 32

/* INCY == 1 */

.L112:

	{ .mmf

	(p16) LDFPD	f32, f35  = [Y1], 2 * SIZE

	(p16) lfetch.nt1 [PREX], INCX16

	(p18) FMA	f8  = f34, f82,  f8

	}

	{ .mmf

	(p16) LDFD	f80 = [X1], INCX

	(p16) LDFD	f92 = [X2], INCX

	(p18) FMA	f9  = f37, f85,  f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f38,  f41  = [Y1], 2 * SIZE

	(p16) lfetch.nt1 [PREY], INCY16

	(p18) FMA	f10 = f40, f88,  f10

	}

	{ .mmf

	(p16) LDFD	f83  = [X1], INCX

	(p16) LDFD	f95  = [X2], INCX

	(p18) FMA	f11 = f43, f91,  f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f44,  f47  = [Y1], 2 * SIZE

	(p18) FMA	f12 = f46, f94,  f12

	}

	{ .mmf

	(p16) LDFD	f86  = [X1], INCX

	(p16) LDFD	f98  = [X2], INCX

	(p18) FMA	f13 = f49, f97,  f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f50,  f53  = [Y1], 2 * SIZE

	(p18) FMA	f14 = f52, f100,  f14

	}

	{ .mmf

	(p16) LDFD	f89  = [X1], INCX5

	(p16) LDFD	f101 = [X2], INCX5

	(p18) FMA	f15 = f55, f103, f15

	}

	;;

	{ .mmf

	(p16) LDFPD	f56,  f59  = [Y1], 2 * SIZE

	(p18) FMA	f8  = f58, f106, f8

	}

	{ .mmf

	(p16) LDFD	f104 = [X1], INCX

	(p16) LDFD	f116 = [X2], INCX

	(p18) FMA	f9   = f61, f109, f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f62,  f65  = [Y1], 2 * SIZE

	(p18) FMA	f10 = f64, f112, f10

	}

	{ .mmf

	(p16) LDFD	f107 = [X1], INCX

	(p16) LDFD	f119 = [X2], INCX

	(p18) FMA	f11 = f67, f115, f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f68,  f71  = [Y1], 2 * SIZE

	(p18) FMA	f12 = f70, f118, f12

	}

	{ .mmf

	(p16) LDFD	f110 = [X1], INCX

	(p16) LDFD	f122 = [X2], INCX

	(p18) FMA	f13 = f73, f121, f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f74,  f77  = [Y1], 2 * SIZE

	(p16) LDFD	f113 = [X1], INCX5

	(p18) FMA	f14 = f76, f124, f14

	}

	{ .mfb

	(p16) LDFD	f125 = [X2], INCX5

	(p18) FMA	f15 = f79, f127, f15

	br.ctop.sptk.few .L112

	}

	;;

	.align 32

.L115:

	{ .mmi

	(p12) LDFPD	f32, f33  = [Y1], 2 * SIZE

	mov	XX = X1

	tbit.z	p0, p13 = N, 2

	}

	{ .mmb

	(p12) LDFD	f34 = [X1], INCX

	(p12) LDFD	f42 = [X2], INCX

	(p7) br.cond.dptk .L999

	}

	;;

	{ .mmi

	(p12) LDFPD	f36, f37  = [Y1], 2 * SIZE

	(p12) shladd XX = INCX, 3, XX

	tbit.z	p0, p14 = N, 1

	}

	{ .mmi

	(p12) LDFD	f35 = [X1], INCX

	(p12) LDFD	f43 = [X2], INCX

	tbit.z	p0, p15 = N, 0

	}

	;;

	{ .mmi

	(p12) LDFPD	f40, f41  = [Y1], 2 * SIZE

	(p13) shladd XX = INCX, 2, XX

	}

	{ .mmi

	(p12) LDFD	f38 = [X1], INCX

	(p12) LDFD	f46 = [X2], INCX

	}

	;;

	(p12) LDFPD	f44, f45  = [Y1], 2 * SIZE

	(p12) LDFD	f39 = [X1], INCX5

	(p12) LDFD	f47 = [X2], INCX5

	;;

	(p13) LDFPD	f48, f49  = [Y1], 2 * SIZE

	(p13) LDFD	f50 = [X1], INCX

	(p14) LDFD	f58 = [XX], INCX

	;;

	(p13) LDFPD	f52, f53  = [Y1], 2 * SIZE

	(p13) LDFD	f51 = [X1], INCX

	(p14) LDFD	f59 = [XX], INCX

	;;

	(p14) LDFPD	f56, f57  = [Y1], 2 * SIZE

	(p13) LDFD	f54 = [X1], INCX

	(p15) LDFD	f61  = [XX]

	;;

	(p13) LDFD	f55 = [X1], INCX

	(p15) LDFD	f60  = [Y1]

	;;

	(p12) FMA	f8  = f32, f34, f8

	(p12) FMA	f9  = f33, f35, f9

	(p12) FMA	f10 = f36, f38, f10

	(p12) FMA	f11 = f37, f39, f11

	(p12) FMA	f12 = f40, f42, f12

	(p12) FMA	f13 = f41, f43, f13

	(p12) FMA	f14 = f44, f46, f14

	(p12) FMA	f15 = f45, f47, f15

	;;		      

	(p13) FMA	f8  = f48, f50, f8

	(p13) FMA	f9  = f49, f51, f9

	(p13) FMA	f10 = f52, f54, f10

	(p13) FMA	f11 = f53, f55, f11

	(p14) FMA	f12 = f56, f58, f12

	(p14) FMA	f13 = f57, f59, f13

	(p15) FMA	f14 = f60, f61, f14

	br	.L999

	;;

	.align 32

.L120:

	{ .mmi

	adds	PREX = (PREFETCH_SIZE + 17) * SIZE, X1

	adds	PREY = (PREFETCH_SIZE + 19) * SIZE, X1

	mov	ar.lc = I

	}

	{ .mfb

	cmp.eq	p6 ,p0  =   -1, I

	FMA  f15 = f32, f80, f0

	(p6) br.cond.dpnt  .L125

	}

	;;

	.align 32

.L122:

	{ .mmf

	(p16) LDFPD	f32, f35  = [Y1], 2 * SIZE

	(p16) lfetch.nt1 [PREX], INCX16

	(p18) FMA	f8  = f34, f82,  f8

	}

	{ .mmf

	(p17) LDFD	f81 = [X1], INCX

	(p17) LDFD	f93 = [X2], INCX

	(p18) FMA	f9  = f37, f85,  f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f38,  f41  = [Y1], 2 * SIZE

	(p16) lfetch.nt1 [PREY], INCX16

	(p18) FMA	f10 = f40, f88,  f10

	}

	{ .mmf

	(p17) LDFD	f84  = [X1], INCX

	(p17) LDFD	f96  = [X2], INCX

	(p18) FMA	f11 = f43, f91,  f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f44,  f47  = [Y1], 2 * SIZE

	(p18) FMA	f12 = f46, f94,  f12

	}

	{ .mmf

	(p17) LDFD	f87  = [X1], INCX

	(p17) LDFD	f99  = [X2], INCX

	(p18) FMA	f13 = f49, f97,  f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f50,  f53  = [Y1], 2 * SIZE

	(p18) FMA	f14 = f52, f100,  f14

	}

	{ .mmf

	(p17) LDFD	f90  = [X1], INCX5

	(p17) LDFD	f102 = [X2], INCX5

	(p18) FMA	f15 = f55, f103, f15

	}

	;;

	{ .mmf

	(p16) LDFPD	f56,  f59  = [Y1], 2 * SIZE

	(p18) FMA	f8  = f58, f106, f8

	}

	{ .mmf

	(p17) LDFD	f105 = [X1], INCX

	(p17) LDFD	f117 = [X2], INCX

	(p18) FMA	f9   = f61, f109, f9

	}

	;;

	{ .mmf

	(p16) LDFPD	f62,  f65  = [Y1], 2 * SIZE

	(p18) FMA	f10 = f64, f112, f10

	}

	{ .mmf

	(p17) LDFD	f108 = [X1], INCX

	(p17) LDFD	f120 = [X2], INCX

	(p18) FMA	f11 = f67, f115, f11

	}

	;;

	{ .mmf

	(p16) LDFPD	f68,  f71  = [Y1], 2 * SIZE

	(p18) FMA	f12 = f70, f118, f12

	}

	{ .mmf

	(p17) LDFD	f111 = [X1], INCX

	(p17) LDFD	f123 = [X2], INCX

	(p18) FMA	f13 = f73, f121, f13

	}

	;;

	{ .mmf

	(p16) LDFPD	f74,  f77  = [Y1], 2 * SIZE

	(p17) LDFD	f114 = [X1], INCX5

	(p18) FMA	f14 = f76, f124, f14

	}

	{ .mfb

	(p17) LDFD	f126 = [X2], INCX5

	(p18) FMA	f15 = f79, f127, f15

	br.ctop.sptk.few .L122

	}

	;;

	.align 32

.L125:

	{ .mmi

	(p12) LDFPD	f32, f33  = [Y1], 2 * SIZE

	mov	XX = X1

	tbit.z	p0, p13 = N, 2

	}

	{ .mmb

	(p12) LDFD	f34 = [X1], INCX

	(p12) LDFD	f42 = [X2], INCX

	(p7) br.cond.dptk .L999

	}

	;;

	{ .mmi

	(p12) LDFPD	f36, f37  = [Y1], 2 * SIZE

	(p12) shladd XX = INCX, 3, XX

	tbit.z	p0, p14 = N, 1

	}

	{ .mmi

	(p12) LDFD	f35 = [X1], INCX