gemv_t.s

Optimized GotoBLAS libraries

	mov	f13 = f0
	}
	{ .mmf
	(p8) LDFD f34 = [AO3], SIZE
	(p8) LDFD f35 = [AO4], SIZE
	mov	f15 = f0
	}
	;;
	{ .mmi
	adds	RPRE1  = RPREFETCH * SIZE, AO1
	adds	RPRE2  = (RPREFETCH + 8) * SIZE, AO2
	mov	ar.ec= 2
	}
	{ .mmi
	cmp.eq	p16, p0 = r0, r0
	add	I = I, I
	adds	AO21 = 7 * SIZE, AO2
	}
	;;
	{ .mmf
	adds	WPRE =  4 * SIZE, CO
	adds	PREB  = RPREFETCH * SIZE, BO
	(p8) FMPY	f8  = f40, f32
	}
	{ .mmf
	adds	RPRE3  = RPREFETCH * SIZE, AO3
	adds	I = -1, I
	(p8) FMPY	f10 = f40, f33
	}
	;;
	{ .mfi
	adds	AO41 = 7 * SIZE, AO4
	(p8) FMPY	f12 = f40, f34
	mov	ar.lc = I
	}
	{ .mfb
	adds	RPRE4  = (RPREFETCH + 8) * SIZE, AO4
	(p8) FMPY	f14 = f40, f35
	(p6) br.cond.dpnt .L125
	}
	;;
	.align 16

.L122:
	{ .mmf
	(p17) LDFPD	f72, f87 = [AO4]
	(p17) LDFPD	f110, f111 = [BO], 2 * SIZE
	(p17) FMA	f8  = f104, f33, f8
	}
	{ .mfi
	(p17) adds	AO4 = 3 * SIZE, AO4
	(p17) FMA	f9  = f105, f34, f9
	(p16) tbit.nz.unc p14, p15 = I, 0
	}
	;;
	{ .mmf
	(p14) PREFETCH [RPRE1], 16 * SIZE
	(p16) LDFPD	f32, f33 = [AO1], 2 * SIZE
	(p17) FMA	f10 = f104, f35, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f105, f36, f11
	}
	;;
	{ .mmf
	(p15) PREFETCH [RPRE2], 16 * SIZE
	(p16) LDFD	f34      = [AO2], 1 * SIZE
	(p17) FMA	f12 = f104, f37, f12
	}
	{ .mmf
	(p17) LDFD	f84      = [AO21], 8 * SIZE
	nop	__LINE__
	(p17) FMA	f13 = f105, f38, f13
	}
	;;
	{ .mmf
	(p14) PREFETCH [RPRE3], 16 * SIZE
	(p16) LDFPD	f36, f37 = [AO3], 2 * SIZE
	(p17) FMA	f14 = f104, f39, f14
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f15 = f105, f40, f15
	}
	;;
	{ .mmf
	(p15) PREFETCH [RPRE4], 16 * SIZE
	(p16) LDFD	f38      = [AO4], 1 * SIZE
	(p17) FMA	f8  = f106, f49, f8
	}
	{ .mmf
	(p17) LDFD	f88      = [AO41], 8 * SIZE
	nop	__LINE__
	(p17) FMA	f9  = f107, f50, f9
	}
	;;
	{ .mmf
	(p14) PREFETCH [PREB], 16 * SIZE
	(p16) LDFPD	f48, f49 = [AO1], 2 * SIZE
	(p17) FMA	f10 = f106, f51, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f107, f52, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f35, f50 = [AO2], 2 * SIZE
	(p16) LDFPD	f103, f104 = [BO], 2 * SIZE
	(p17) FMA	f12 = f106, f53, f12
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f13 = f107, f54, f13
	}
	;;
	{ .mmf
	(p16) LDFPD	f52, f53 = [AO3], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f14 = f106, f55, f14
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f15 = f107, f56, f15
	}
	;;
	{ .mmf
	(p16) LDFPD	f39, f54 = [AO4], 2 * SIZE
	(p16) LDFPD	f105, f106 = [BO], 2 * SIZE
	(p17) FMA	f8  = f108, f65, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f109, f66, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f64, f65 = [AO1], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f10 = f108, f67, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f109, f68, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f51, f66 = [AO2], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f12 = f108, f69, f12
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f13 = f109, f70, f13
	}
	;;
	{ .mmf
	(p16) LDFPD	f68, f69 = [AO3], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f14 = f108, f71, f14
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f15 = f109, f72, f15
	}
	;;
	{ .mmf
	(p16) LDFPD	f55, f70 = [AO4], 2 * SIZE
	(p16) LDFPD	f107, f108 = [BO], 2 * SIZE
	(p17) FMA	f8  = f110, f81, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f111, f82, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f80, f81 = [AO1], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f10 = f110, f83, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f111, f84, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f67, f82 = [AO2]
	nop	__LINE__
	(p17) FMA	f12 = f110, f85, f12
	}
	{ .mmf
	nop	__LINE__
	(p16) adds	AO2 = 3 * SIZE, AO2
	(p17) FMA	f13 = f111, f86, f13
	}
	;;
	{ .mmf
	(p16) LDFPD	f84, f85 = [AO3], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f14 = f110, f87, f14
	}
	{ .mfb
	adds	I = -1, I
	(p17) FMA	f15 = f111, f88, f15
	br.ctop.sptk.few .L122
	}
	;;
	.align 16

.L125:
	and	I = 15, MIN_M
	mov	pr.rot= 0
	;;
	cmp.eq	p6,  p0 = 0, I
	cmp.eq	p16, p15 = r0, r0
	;;
	adds	I = 1, I
	adds	AO21 = 1 * SIZE, AO2
	adds	AO41 = 1 * SIZE, AO4
	;;
	shr	I = I, 1
	;;
	adds	I = -1, I
	;;
	mov	ar.lc = I
	mov	ar.ec= 3
	and	I = 15, MIN_M
	(p6) br.cond.dpnt .L128
	;;
	.align 16

.L126:
	{ .mmf
	(p16) LDFPD	f104, f107 = [BO], 2 * SIZE
	(p16) LDFPD	f32,  f35  = [AO1], 2 * SIZE
	(p18) FMA	f8  = f106, f34, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p15) FMA	f9  = f109, f37, f9
	}
	;;
	{ .mmf
	(p17) LDFD	f42        = [AO21], 2 * SIZE
	(p16) LDFD	f38        = [AO2], 2 * SIZE
	(p18) FMA	f10 = f106, f40, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p15) FMA	f11 = f109, f43, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f44,  f47  = [AO3], 2 * SIZE
	nop	__LINE__
	(p18) FMA	f12 = f106, f46, f12
	}
	{ .mmf
	nop	__LINE__
	(p17) adds	I = -2, I
	(p15) FMA	f13 = f109, f49, f13
	}
	;;
	{ .mmf
	(p17) LDFD	f54        = [AO41], 2 * SIZE
	(p16) LDFD	f50        = [AO4], 2 * SIZE
	(p15) FMA	f15 = f109, f55, f15
	}
	{ .mfb
	(p17) cmp.ne.unc p15, p0 = -1, I
	(p18) FMA	f14 = f106, f52, f14
	br.ctop.sptk.few .L126
	}
	;;

.L128:
	{ .mmf
	mov	AO1 = CO
	LDFD	f32 = [CO], INCY
	FADD	f8  = f8,  f9
	}
	;;
	{ .mmf
	LDFD	f33 = [CO], INCY
	nop	__LINE__
	FADD	f10 = f10, f11
	}
	;;
	{ .mmf
	LDFD	f34 = [CO], INCY
	nop	__LINE__
	FADD	f12 = f12, f13
	}
	;;
	{ .mmf
	LDFD	f35 = [CO], INCY
	nop	__LINE__
	FADD	f14 = f14, f15
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f32 = ALPHA, f8,  f32
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f33 = ALPHA, f10, f33
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f34 = ALPHA, f12, f34
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f35 = ALPHA, f14, f35
	}
	;;
	{ .mmf
	STFD [AO1] = f32
	add	AO1 = AO1, INCY
	}
	;;
	{ .mmf
	STFD [AO1] = f33
	add	AO1 = AO1, INCY
	}
	;;
	{ .mmf
	STFD [AO1] = f34
	add	AO1 = AO1, INCY
	}
	;;
	{ .mmf
	STFD [AO1] = f35
	add	AO1 = AO1, INCY
	}
	;;
	.align 16

.L130:
	{ .mfi
	mov	AO1 = A
	mov	f8  = f0
	mov	pr.rot= 0
	}
	{ .mfi
	add	AO2 = LDA, A
	mov	f10 = f0
	tbit.z	p6, p0  = N, 1
	}
	;;
	{ .mfi
	mov	BO  = BUFFER
	mov	f12 = f0
	shr	I = MIN_M, 4
	}
	{ .mfb
	adds	WPRE =  4 * SIZE, CO
	mov	f14 = f0
	(p6) br.cond.dpnt .L140
	}
	;;
	{ .mmf
	(p8) LDFD f32 = [AO1], SIZE
	(p8) LDFD f33 = [AO2], SIZE
	mov	f9  = f0
	}
	{ .mfi
	shladd	A   = LDA, 1, A
	mov	f11 = f0
	mov	ar.ec= 2
	}
	;;
	{ .mmf
	(p8) LDFD f40 = [BO], 2 * SIZE
	cmp.eq	p6, p0 = 0, I
	mov	f13 = f0
	}
	{ .mmf
	adds	RPRE1  = RPREFETCH * SIZE, AO1
	add	I = I, I
	mov	f15 = f0
	}
	;;
	{ .mmi
	cmp.eq	p16, p0 = r0, r0
	adds	RPRE2  = (RPREFETCH + 8) * SIZE, AO2
	adds	I = -1, I
	}
	;;
	{ .mfi
	adds	AO21 = 7 * SIZE, AO2
	(p8) FMPY	f8  = f40, f32
	mov	ar.lc = I
	}
	{ .mfb
	adds	PREB  = RPREFETCH * SIZE, BO
	(p8) FMPY	f10 = f40, f33
	(p6) br.cond.dpnt .L135
	}
	;;
	.align 16

.L132:
	{ .mmf
	(p17) LDFPD	f68, f83 = [AO2]
	(p17) LDFPD	f110, f111 = [BO], 2 * SIZE
	(p17) FMA	f8  = f104, f33, f8
	}
	{ .mfi
	(p17) adds	AO2 = 3 * SIZE, AO2
	(p17) FMA	f9  = f105, f34, f9
	(p16) tbit.nz.unc p14, p15 = I, 0
	}
	;;
	{ .mmf
	(p14) PREFETCH [RPRE1], 16 * SIZE
	(p16) LDFPD	f32, f33 = [AO1], 2 * SIZE
	(p17) FMA	f10 = f104, f35, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f105, f36, f11
	}
	;;
	{ .mmf
	(p15) PREFETCH [RPRE2], 16 * SIZE
	(p16) LDFD	f34      = [AO2], 1 * SIZE
	(p17) FMA	f8  = f106, f49, f8
	}
	{ .mmf
	(p17) LDFD	f84      = [AO21], 8 * SIZE
	nop	__LINE__
	(p17) FMA	f9  = f107, f50, f9
	}
	;;
	{ .mmf
	(p14) PREFETCH [PREB], 16 * SIZE
	(p16) LDFPD	f48, f49 = [AO1], 2 * SIZE
	(p17) FMA	f10 = f106, f51, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f107, f52, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f35, f50 = [AO2], 2 * SIZE
	(p16) LDFPD	f103, f104 = [BO], 2 * SIZE
	(p17) FMA	f8  = f108, f65, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f109, f66, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f105, f106 = [BO], 2 * SIZE
	(p16) LDFPD	f64, f65 = [AO1], 2 * SIZE
	(p17) FMA	f10 = f108, f67, f10
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f11 = f109, f68, f11
	}
	;;
	{ .mmf
	(p16) LDFPD	f51, f66 = [AO2], 2 * SIZE
	(p16) LDFPD	f107, f108 = [BO], 2 * SIZE
	(p17) FMA	f8  = f110, f81, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f111, f82, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f80, f81 = [AO1], 2 * SIZE
	nop	__LINE__
	(p17) FMA	f10 = f110, f83, f10
	}
	{ .mfb
	adds	I = -1, I
	(p17) FMA	f11 = f111, f84, f11
	br.ctop.sptk.few .L132
	}
	;;
	.align 16

.L135:
	and	I = 15, MIN_M
	;;
	cmp.eq	p6,  p0 = 0, I
	(p6) br.cond.dpnt .L138
	;;
	tbit.nz	p12, p0 = MIN_M, 3
	tbit.nz	p13, p0 = MIN_M, 2
	tbit.nz	p14, p0 = MIN_M, 1
	tbit.nz	p15, p0 = MIN_M, 0
	;;
	(p12) LDFPD	f100, f101 = [BO], 2 * SIZE
	(p12) LDFPD	f32,  f33  = [AO1], 2 * SIZE
	(p12) LDFD	f34        = [AO2], 1 * SIZE
	;;
	(p12) LDFPD	f36,  f37  = [AO1], 2 * SIZE
	(p12) LDFPD	f35,  f38  = [AO2], 2 * SIZE
	;;
	(p12) LDFPD	f102, f103 = [BO],  2 * SIZE
	(p12) LDFPD	f39,  f42  = [AO2], 2 * SIZE
	;;
	(p12) LDFPD	f40,  f41  = [AO1], 2 * SIZE
	(p12) LDFPD	f43,  f46  = [AO2], 2 * SIZE
	;;
	(p12) LDFPD	f104, f105 = [BO], 2 * SIZE
	(p12) LDFPD	f44,  f45  = [AO1], 2 * SIZE
	(p12) LDFD	f47        = [AO2], 1 * SIZE
	;;
	(p12) LDFPD	f106, f107 = [BO], 2 * SIZE
	(p13) LDFD	f50        = [AO2], 1 * SIZE
	(p13) LDFPD	f48,  f49  = [AO1], 2 * SIZE
	;;
	(p13) LDFPD	f108, f109 = [BO], 2 * SIZE
	(p13) LDFPD	f51,  f54  = [AO2], 2 * SIZE
	;;
	(p13) LDFPD	f110, f111 = [BO], 2 * SIZE
	(p13) LDFPD	f52,  f53  = [AO1], 2 * SIZE
	(p13) LDFD	f55        = [AO2], 1 * SIZE
	;;
	(p14) LDFPD	f56,  f57  = [AO1], 2 * SIZE
	(p14) LDFD	f58        = [AO2], 1 * SIZE
	;;
	(p14) LDFPD	f112, f113 = [BO], 2 * SIZE
	(p15) LDFD	f60        = [AO1]
	(p14) LDFD	f59        = [AO2], 1 * SIZE
	;;
	(p15) LDFD	f61        = [AO2]
	(p15) LDFD	f114       = [BO]
	;;
	(p12) FMA	f8  = f100, f32, f8
	(p12) FMA	f9  = f101, f33, f9
	(p12) FMA	f10 = f100, f34, f10
	(p12) FMA	f11 = f101, f35, f11
	;;
	(p12) FMA	f12 = f102, f36, f12
	(p12) FMA	f13 = f103, f37, f13
	(p12) FMA	f14 = f102, f38, f14
	(p12) FMA	f15 = f103, f39, f15
	;;
	(p12) FMA	f8  = f104, f40, f8
	(p12) FMA	f9  = f105, f41, f9
	(p12) FMA	f10 = f104, f42, f10
	(p12) FMA	f11 = f105, f43, f11
	;;
	(p12) FMA	f12 = f106, f44, f12
	(p12) FMA	f13 = f107, f45, f13
	(p12) FMA	f14 = f106, f46, f14
	(p12) FMA	f15 = f107, f47, f15
	;;
	(p13) FMA	f8  = f108, f48, f8
	(p13) FMA	f9  = f109, f49, f9
	(p13) FMA	f10 = f108, f50, f10
	(p13) FMA	f11 = f109, f51, f11
	;;
	(p13) FMA	f12 = f110, f52, f12
	(p13) FMA	f13 = f111, f53, f13
	(p13) FMA	f14 = f110, f54, f14
	(p13) FMA	f15 = f111, f55, f15
	;;
	(p14) FMA	f8  = f112, f56, f8
	(p14) FMA	f9  = f113, f57, f9
	(p14) FMA	f10 = f112, f58, f10
	(p14) FMA	f11 = f113, f59, f11
	;;
	(p15) FMA	f12 = f114, f60, f12
	(p15) FMA	f14 = f114, f61, f14
	;;
.L138:
	FADD	f8  = f8,  f9
	FADD	f10 = f10, f11
	FADD	f12 = f12, f13
	FADD	f14 = f14, f15
	;;
	FADD	f8  = f8,  f12
	FADD	f10 = f10, f14
	;;
	{ .mmf
	mov	AO1 = CO
	LDFD	f32 = [CO], INCY
	}
	;;
	{ .mmf
	LDFD	f33 = [CO], INCY
	nop	__LINE__
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f32 = ALPHA, f8,  f32
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f33 = ALPHA, f10, f33
	}
	;;
	{ .mmf
	STFD [AO1] = f32
	add	AO1 = AO1, INCY
	}
	;;
	{ .mmf
	STFD [AO1] = f33
	}
	;;
	.align 16

.L140:
	{ .mfi
	mov	AO1 = A
	mov	f8  = f0
	shr	I = MIN_M, 4
	}
	{ .mfi
	mov	BO  = BUFFER
	mov	f10 = f0
	tbit.z	p7, p0  = N, 0
	}
	;;
	{ .mfi
	cmp.eq	p6, p0 = 0, I
	mov	f12 = f0
	mov	pr.rot= 0
	}
	{ .mfb
	add	I = I, I
	mov	f14 = f0
	(p7) br.cond.dpnt .L199
	}
	;;
	{ .mfi
	(p8) LDFD f32 = [AO1], SIZE
	mov	f9  = f0
	mov	ar.ec= 2
	}
	{ .mmf
	(p8) LDFD f40 = [BO], 2 * SIZE
	add	A   = A, LDA
	mov	f11 = f0
	}
	;;
	{ .mmf
	adds	WPRE =  1 * SIZE, CO
	adds	PREB  = RPREFETCH * SIZE, BO
	mov	f13 = f0
	}
	{ .mmf
	cmp.eq	p16, p0 = r0, r0
	adds	I = -1, I
	mov	f15 = f0
	}
	;;
	{ .mfi
	lfetch.excl.nt1	[WPRE]
	(p8) FMPY	f8  = f40, f32
	mov	ar.lc = I
	}
	{ .mmb
	nop	__LINE__
	nop	__LINE__
	(p6) br.cond.dpnt .L145
	}
	;;
	.align 16

.L142:
	{ .mmf
	(p17) LDFPD	f81, f82   = [AO1], 2 * SIZE
	(p17) LDFPD	f110, f111 = [BO], 2 * SIZE
	(p17) FMA	f8  = f104, f33, f8
	}
	{ .mfi
	nop	__LINE__
	(p17) FMA	f9  = f105, f34, f9
	(p16) tbit.nz.unc p14, p15 = I, 0
	}
	;;
	{ .mmf
	(p16) LDFPD	f32, f33   = [AO1], 2 * SIZE
	(p16) LDFPD	f103, f104 = [BO], 2 * SIZE
	(p17) FMA	f8  = f106, f49, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f107, f50, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f105, f106 = [BO], 2 * SIZE
	(p16) LDFPD	f48, f49 = [AO1], 2 * SIZE
	(p17) FMA	f8  = f108, f65, f8
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	(p17) FMA	f9  = f109, f66, f9
	}
	;;
	{ .mmf
	(p16) LDFPD	f64, f65  = [AO1], 2 * SIZE
	(p16) LDFPD	f107, f108 = [BO], 2 * SIZE
	(p17) FMA	f8  = f110, f81, f8
	}
	{ .mfb
	adds	I = -1, I
	(p17) FMA	f9  = f111, f82, f9
	br.ctop.sptk.few .L142
	}
	;;
	.align 16

.L145:
	and	I = 15, MIN_M
	;;
	cmp.eq	p6,  p0 = 0, I
	(p6) br.cond.dpnt .L148
	;;
	tbit.nz	p12, p0 = MIN_M, 3
	tbit.nz	p13, p0 = MIN_M, 2
	tbit.nz	p14, p0 = MIN_M, 1
	tbit.nz	p15, p0 = MIN_M, 0
	;;
	(p12) LDFPD	f32,  f33  = [AO1], 2 * SIZE
	(p12) LDFPD	f100, f101 = [BO], 2 * SIZE
	;;
	(p12) LDFPD	f36,  f37  = [AO1], 2 * SIZE
	(p12) LDFPD	f102, f103 = [BO], 2 * SIZE
	;;
	(p12) LDFPD	f40,  f41  = [AO1], 2 * SIZE
	(p12) LDFPD	f104, f105 = [BO], 2 * SIZE
	;;
	(p12) LDFPD	f44,  f45  = [AO1], 2 * SIZE
	(p12) LDFPD	f106, f107 = [BO], 2 * SIZE
	;;
	(p13) LDFPD	f48,  f49  = [AO1], 2 * SIZE
	(p13) LDFPD	f108, f109 = [BO], 2 * SIZE
	;;
	(p13) LDFPD	f52,  f53  = [AO1], 2 * SIZE
	(p13) LDFPD	f110, f111 = [BO], 2 * SIZE
	;;
	(p14) LDFPD	f56,  f57  = [AO1], 2 * SIZE
	(p14) LDFPD	f112, f113 = [BO], 2 * SIZE
	;;
	(p15) LDFD	f60        = [AO1]
	(p15) LDFD	f114       = [BO]
	;;
	(p12) FMA	f8  = f100, f32, f8
	(p12) FMA	f9  = f101, f33, f9
	(p12) FMA	f10 = f102, f36, f10
	(p12) FMA	f11 = f103, f37, f11
	(p12) FMA	f12 = f104, f40, f12
	(p12) FMA	f13 = f105, f41, f13
	(p12) FMA	f14 = f106, f44, f14
	(p12) FMA	f15 = f107, f45, f15
	;;
	(p13) FMA	f8  = f108, f48, f8
	(p13) FMA	f9  = f109, f49, f9
	(p13) FMA	f10 = f110, f52, f10
	(p13) FMA	f11 = f111, f53, f11
	(p14) FMA	f12 = f112, f56, f12
	(p14) FMA	f13 = f113, f57, f13
	(p15) FMA	f14 = f114, f60, f14
	;;
.L148:
	{ .mmf
	LDFD	f32 = [CO]
	nop	__LINE__
	FADD	f8  = f8,  f9
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f10 = f10, f11
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f12 = f12, f13
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f14 = f14, f15
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f8  = f8,  f12
	}
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f10 = f10, f14
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FADD	f8  = f8,  f10
	}
	;;
	{ .mmf
	nop	__LINE__
	nop	__LINE__
	FMA	f32 = ALPHA, f8,  f32
	}
	;;
	{ .mmf
	STFD [CO] = f32
	nop	__LINE__
	nop	__LINE__
	}
	;;
	.align 16

.L199:
	adds	IS = P, IS
	shladd	A  = LDAP, BASE_SHIFT, A
	;;
	cmp.gt	p6, p0 = M, IS
	(p6) br.cond.dptk .LIs_loop
	.align 4
	;;

.L999:
	mov	r8 = r0
	adds	r9 = 1 * 16, SP
	;;
	ldf.fill  f16 = [SP], 32
	ldf.fill  f17 = [r9], 32
	mov	 ar.lc = ARLC
	;;	
	ldf.fill  f18 = [SP], 32
	ldf.fill  f19 = [r9], 32
	mov pr    = PR, -1
	;;	
	ldf.fill  f20 = [SP], 32
	ldf.fill  f21 = [r9], 32
	mov	ar.pfs = ARPFS
	;;	
	ldf.fill  f22 = [SP], 32
	ldf.fill  f23 = [r9]
	br.ret.sptk.many b0
	;;
	EPILOGUE