trsm_kernel_rt.s

Optimized GotoBLAS libraries

	}
	;;
	{ .mib
	(p7) LDFD	f32 = [AOFFSET], 1 * SIZE
	mov	ar.lc = L
	(p6) br.cond.dpnt   .L168
	}
	;;
	.align 8

.L162:
	{ .mmf
	cmp.ne	p4, p5 =  0, L
	(p12) cmp.ne p3, p0 =  0, L
	FMA	f64   = f32, f48, f64	// A1 * B1
	}
	;;
	{ .mmi
	(p3) LDFD	f56 = [BOFFSET], 1 * SIZE
	(p3) LDFD	f40 = [AOFFSET], 1 * SIZE
	nop	__LINE__
	}
	;;
	{ .mmi
	(p4) LDFD	f32 = [AOFFSET],   1 * SIZE
	nop	__LINE__
	adds	L = -1, L
	}
	{ .mfb
	(p4) LDFD	f48 = [BOFFSET],   1 * SIZE
	(p3) FMA	f64   = f40, f56, f64	// A1 * B1
	br.cloop.sptk.few .L162
	}
	;;
	.align 8

.L168:
#if defined(LN) || defined(RT)
#ifdef LN
	adds	r2 = -1, KK
#else
	adds	r2 = -1, KK
#endif
	;;
	shladd	r2 = r2, BASE_SHIFT, r0
	;;
	add	AOFFSET = r2, AORIG
	add	BOFFSET = r2, B
	;;	
#endif

#if defined(LN) || defined(LT)
	{ .mmi
	LDFD	f32 = [BOFFSET]
	LDFD	f33 = [AOFFSET]
#ifdef LN
	adds	C1 = -1 * SIZE, C1
#else
	nop	__LINE__
#endif
	}
	;;
#else
	{ .mmi
	LDFD	f32 = [AOFFSET]
	LDFD	f33 = [BOFFSET]
	nop	__LINE__
	}
	;;
#endif

	{ .mmf
	sub	L = K, KK
#ifdef RT
	shladd	AORIG = K, BASE_SHIFT, AORIG
#else
	nop	__LINE__
#endif
	FSUB	f64  = f32, f64
	}
	;;
#ifdef LT
	adds	KK =  1, KK
#elif defined LN
	adds	KK = -1, KK
#else
	nop	__LINE__
#endif
	;;
#if defined(LT) || defined(RN)
	mov	L = KK
#else
	sub	L = K, KK
#endif
	;;
	FMPY	f64  = f64,  f33
	;;
#if defined(LN) || defined(LT)
	{ .mmf
	STFD	[BOFFSET]  = f64
#ifndef LN
	STFD	[C1 ] = f64, SIZE
#else
	STFD	[C1 ] = f64
#endif
	mov	f64  = f0
	}
	;;
#else
	{ .mmf
	STFD	[AOFFSET]  = f64
	STFD	[C1 ] = f64, SIZE
	mov	f64  = f0
	}
	;;
#endif

#if defined(LT) || defined(RN)
	shladd	AOFFSET = L, BASE_SHIFT, AOFFSET
#else
	nop	__LINE__
#endif
#if defined(LT) || defined(RN)
	shladd	BOFFSET = L, BASE_SHIFT, BOFFSET
#else
	nop	__LINE__
#endif
	;;
	.align 8

.L169:
	{ .mii
#ifdef LN
	shladd	B = K, BASE_SHIFT, B
#elif defined(LT) || defined(RN)
	mov	B =  BOFFSET
#else
	nop	__LINE__
#endif

#ifdef RN
	adds	KK =  1,  KK
#elif defined RT
	adds	KK = -1,  KK
#else
	nop	__LINE__
#endif
	mov	AOFFSET = A
	}
	;;
	.align 16

.L090:
	tbit.z	p6, p0 = N, 1
	(p6)	br.cond.dpnt .L050
	;;

#ifdef RT
       { .mmi
	shladd	r3 = LDC, 1, r0
	nop	__LINE__
	shl	r2 = K, 1 + BASE_SHIFT
	}
	;;
	{ .mmi
	sub	B = B, r2
	sub	C = C, r3
	nop	__LINE__
	}
#endif
	;;
	mov	f64  = f0
	mov	f65  = f0
	mov	f66  = f0
	mov	f67  = f0

	mov	f72  = f0
	mov	f73  = f0
	mov	f74  = f0
	mov	f75  = f0
	;;
	{ .mfi
	shr	I  = M, 3
	} 
	{ .mfi
	mov	C1 = C			// coffset1 = c + 0 * ldc
#ifdef LN
	add	KK = M, OFFSET
#elif defined LT
	mov	KK = OFFSET
#else
	nop	__LINE__
#endif
	}
	;;
	{ .mmf
	cmp.eq	p6, p7 = 0, I
#if defined(LN) || defined(RT)
	mov	AORIG = A
#else
	mov	AOFFSET = A
#endif
	}
	{ .mmf
	add	C2 = LDC, C		// coffset2 = c + 1 * ldc
	}
	;;
	{ .mfi
#ifndef RT
	shladd	C = LDC, 1, C		// coffset += 8 * ldc
#else
	nop	__LINE__
#endif
	mov	f81  = f0
#if defined(LT) || defined(RN)
	mov	L = KK
#else
	sub	L = K, KK
#endif
	}{ .mfb
	(p6)	br.cond.dpnt .L100
	}
	;;
	.align 16

.L092:
	{ .mmi
	cmp.ne	p7, p0 = r0, L
	adds	BOFFSET = 0 * SIZE, B
	shl	r2 = K, 3 + BASE_SHIFT
	}
	{ .mmi
	shladd	r3 = KK, BASE_SHIFT, r0
	nop	__LINE__
	nop	__LINE__
	}
	;;
#if defined(LT) || defined(RN)
	{ .mmi
	(p7) LDFPD	f48, f49 = [BOFFSET], 2 * SIZE
	nop	__LINE__
	nop	__LINE__
	}
	;;
#else
	{ .mfi
	shladd	BOFFSET = r3, 1, B
#ifdef LN
	sub	AORIG = AORIG, r2
#else
	nop	__LINE__
#endif
	}
	;;
	{ .mfi
	(p7) LDFPD	f48, f49 = [BOFFSET], 2 * SIZE
	shladd	AOFFSET = r3, 3, AORIG
	}
	;;
#endif
	(p7) LDFPD	f32, f33 = [AOFFSET], 2 * SIZE
	;;
	{ .mmf
	(p7) LDFPD	f34, f35  = [AOFFSET], 2 * SIZE
	}
	;;
	{ .mmf
	(p7) LDFPD	f36, f37  = [AOFFSET], 2 * SIZE
	}
	{ .mfi
	cmp.eq	p3, p0 = r0, r0
	}
	;;
	{ .mmf
	(p7) LDFPD	f38, f39  = [AOFFSET], 2 * SIZE
	}
	{ .mfi
	adds	PREC = CPREFETCHSIZE * SIZE, C1
	}
	;;
	{ .mmf
	CPREFETCH [PREC], LDC
	}
	{ .mfi
	adds	L =  1, L
	}
	;;
	{ .mmf
	CPREFETCH [PREC]
	}
	{ .mfi
	adds	PREA = (PREFETCHSIZE + 8) * SIZE, AOFFSET
	}
	;;
	{ .mfi
	adds	PREB = (PREFETCHSIZE - 8) * SIZE, BOFFSET
	}
	;;
	{ .mfi
	tbit.z	p12, p0 = L, 0
	}
	{ .mfi
	shr	L = L, 1
	}
	;;
	{ .mfi
	adds	L =  -1, L
	}
	;;
	{ .mfi
	mov	ar.lc = L
	}
	;;
	mov	f68  = f0
	mov	f69  = f0
	mov	f70  = f0
	mov	f71  = f0
	mov	f76  = f0
	mov	f77  = f0
	mov	f78  = f0
	mov	f79  = f0
	;;
	{ .mfb
	cmp.eq  p6, p0 = -1, L
	(p6) br.cond.dpnt   .L098
	}
	;;
	.align 8

.L093:
/*  1 */
	{ .mfi
	lfetch.nt1	[PREA],  16 * SIZE
	FMA	f64   = f32, f48, f64	// A1 * B1
	cmp.ne	p4, p5 =  0, L
	}
	{ .mfi
	nop	__LINE__
	FMA	f72   = f32, f49, f72	// A1 * B2
	(p12) cmp.ne p3, p0 =  0, L
	}
	;;
	{ .mfi
	lfetch.nt1	[PREB],   4 * SIZE
	FMA	f65   = f33, f48, f65	// A2 * B1
	adds	C9  = 4 * SIZE, C1
	}
	{ .mfi
	nop	__LINE__
	FMA	f73   = f33, f49, f73	// A2 * B2
	adds	C10 = 4 * SIZE, C2
	}
	;;
	{ .mfi
	(p3) LDFPD	f56, f57 = [BOFFSET],   2 * SIZE
	FMA	f66   = f34, f48, f66	// A3 * B1
	adds	C11 = 4 * SIZE, C3
	}
	{ .mfi
	nop	__LINE__
	FMA	f74   = f34, f49, f74	// A3 * B2
	adds	C12 = 4 * SIZE, C4
	}
	;;
	{ .mfb
	(p3) LDFPD	f40, f41 = [AOFFSET], 2 * SIZE
	FMA	f67   = f35, f48, f67	// A4 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	FMA	f75   = f35, f49, f75	// A4 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p3) LDFPD	f42, f43 = [AOFFSET], 2 * SIZE
	FMA	f68   = f36, f48, f68	// A5 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	FMA	f76   = f36, f49, f76	// A5 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p3) LDFPD	f44, f45 = [AOFFSET], 2 * SIZE
	FMA	f69   = f37, f48, f69	// A6 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	FMA	f77   = f37, f49, f77	// A6 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p3) LDFPD	f46, f47 = [AOFFSET], 2 * SIZE
	FMA	f70   = f38, f48, f70	// A7 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	FMA	f78   = f38, f49, f78	// A7 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p4) LDFPD	f32, f33 = [AOFFSET],   2 * SIZE
	FMA	f71   = f39, f48, f71	// A8 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	FMA	f79   = f39, f49, f79	// A8 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p4) LDFPD	f48, f49 = [BOFFSET],  2 * SIZE
	(p3) FMA	f64   = f40, f56, f64	// A1 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f72   = f40, f57, f72	// A1 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p4) LDFPD	f34, f35 = [AOFFSET], 2 * SIZE
	(p3) FMA	f65   = f41, f56, f65	// A2 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f73   = f41, f57, f73	// A2 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p4) LDFPD	f36, f37 = [AOFFSET], 2 * SIZE
	(p3) FMA	f66   = f42, f56, f66	// A3 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f74   = f42, f57, f74	// A3 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	(p4) LDFPD	f38, f39 = [AOFFSET], 2 * SIZE
	(p3) FMA	f67   = f43, f56, f67	// A4 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f75   = f43, f57, f75	// A4 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	nop	__LINE__
	(p3) FMA	f68   = f44, f56, f68	// A5 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f76   = f44, f57, f76	// A5 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	nop	__LINE__
	(p3) FMA	f69   = f45, f56, f69	// A6 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f77   = f45, f57, f77	// A6 * B2
	nop	__LINE__
	}
	;;
	{ .mfb
	nop	__LINE__
	(p3) FMA	f70   = f46, f56, f70	// A7 * B1
	nop	__LINE__
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f78   = f46, f57, f78	// A7 * B2
	nop	__LINE__
	}
	;;
	{ .mfi
	nop	__LINE__
	(p3) FMA	f71   = f47, f56, f71	// A8 * B1
	adds	L = -1, L
	}
	{ .mfb
	nop	__LINE__
	(p3) FMA	f79   = f47, f57, f79	// A8 * B2
	br.cloop.sptk.few .L093
	}
	;;
	.align 8

.L098:
#if defined(LN) || defined(RT)
#ifdef LN
	adds	r2 = -8, KK
#else
	adds	r2 = -2, KK
#endif
	;;
	shladd	r2 = r2, BASE_SHIFT, r0
	;;
	shladd	AOFFSET = r2, 3, AORIG
	shladd	BOFFSET = r2, 1, B
	;;	
#endif
 	adds	AOFFSET2 = 4 * SIZE, AOFFSET
 	adds	BOFFSET2 = 4 * SIZE, BOFFSET
	;;

#if defined(LN) || defined(LT)
	LDFPD	f32, f33 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f34, f35 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f36, f37 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f38, f39 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f40, f41 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f42, f43 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f44, f45 = [BOFFSET], 2 * SIZE
	;;
	LDFPD	f46, f47 = [BOFFSET]
	adds	BOFFSET = -14 * SIZE, BOFFSET
	;;
	FSUB	f64  = f32, f64
	FSUB	f72  = f33, f72
	FSUB	f65  = f34, f65
	FSUB	f73  = f35, f73

	FSUB	f66  = f36, f66
	FSUB	f74  = f37, f74
	FSUB	f67  = f38, f67
	FSUB	f75  = f39, f75

	FSUB	f68  = f40, f68
	FSUB	f76  = f41, f76
	FSUB	f69  = f42, f69
	FSUB	f77  = f43, f77

	FSUB	f70  = f44, f70
	FSUB	f78  = f45, f78
	FSUB	f71  = f46, f71
	FSUB	f79  = f47, f79
	;;
#else
	LDFPD	f32, f33 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f34, f35 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f36, f37 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f38, f39 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f40, f41 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f42, f43 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f44, f45 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f46, f47 = [AOFFSET]
	adds	AOFFSET = -14 * SIZE, AOFFSET
	;;
	FSUB	f64  = f32, f64
	FSUB	f65  = f33, f65
	FSUB	f66  = f34, f66
	FSUB	f67  = f35, f67
	FSUB	f68  = f36, f68
	FSUB	f69  = f37, f69
	FSUB	f70  = f38, f70
	FSUB	f71  = f39, f71
	;;
	FSUB	f72  = f40, f72
	FSUB	f73  = f41, f73
	FSUB	f74  = f42, f74
	FSUB	f75  = f43, f75
	FSUB	f76  = f44, f76
	FSUB	f77  = f45, f77
	FSUB	f78  = f46, f78
	FSUB	f79  = f47, f79
	;;
#endif

#ifdef LN
	adds	AOFFSET = 62 * SIZE, AOFFSET
	;;
	LDFPD	f33, f32 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f35, f34 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f37, f36 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f39, f38 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFD	f40 = [AOFFSET], -2 * SIZE
	;;
	LDFPD	f42, f41 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f44, f43 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f46, f45 = [AOFFSET]
	adds	AOFFSET = - 4 * SIZE, AOFFSET
	;;
	LDFPD	f48, f47 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f50, f49 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f52, f51 = [AOFFSET]
	adds	AOFFSET = - 4 * SIZE, AOFFSET
	;;
	LDFD	f53 = [AOFFSET], -2 * SIZE
	;;
	LDFPD	f55, f54 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f57, f56 = [AOFFSET]
	adds	AOFFSET = - 6 * SIZE, AOFFSET
	;;
	LDFPD	f59, f58 = [AOFFSET]
	adds	AOFFSET = - 2 * SIZE, AOFFSET
	;;
	LDFPD	f61, f60 = [AOFFSET]
	adds	AOFFSET = - 6 * SIZE, AOFFSET
	;;
	LDFD	f16 = [AOFFSET], -2 * SIZE
	;;
	LDFPD	f18, f17 = [AOFFSET]
	adds	AOFFSET = - 8 * SIZE, AOFFSET
	;;
	LDFPD	f20, f19 = [AOFFSET]
	adds	AOFFSET = - 8 * SIZE, AOFFSET
	;;
	LDFD	f21 = [AOFFSET]
	;;
	FMPY	f71  = f71,  f32
	FMPY	f79  = f79,  f32
	;;
	FNMA	f70  = f71,  f33, f70
	FNMA	f78  = f79,  f33, f78
	;;
	FNMA	f69  = f71,  f34, f69
	FNMA	f77  = f79,  f34, f77
	;;
	FNMA	f68  = f71,  f35, f68
	FNMA	f76  = f79,  f35, f76
	;;
	FNMA	f67  = f71,  f36, f67
	FNMA	f75  = f79,  f36, f75
	;;
	FNMA	f66  = f71,  f37, f66
	FNMA	f74  = f79,  f37, f74
	;;
	FNMA	f65  = f71,  f38, f65
	FNMA	f73  = f79,  f38, f73
	;;
	FNMA	f64  = f71,  f39, f64
	FNMA	f72  = f79,  f39, f72
	;;
	FMPY	f70  = f70,  f40
	FMPY	f78  = f78,  f40
	;;
	FNMA	f69  = f70,  f41, f69
	FNMA	f77  = f78,  f41, f77
	;;
	FNMA	f68  = f70,  f42, f68
	FNMA	f76  = f78,  f42, f76
	;;
	FNMA	f67  = f70,  f43, f67
	FNMA	f75  = f78,  f43, f75
	;;
	FNMA	f66  = f70,  f44, f66
	FNMA	f74  = f78,  f44, f74
	;;
	FNMA	f65  = f70,  f45, f65
	FNMA	f73  = f78,  f45, f73
	;;
	FNMA	f64  = f70,  f46, f64
	FNMA	f72  = f78,  f46, f72
	;;
	FMPY	f69  = f69,  f47
	FMPY	f77  = f77,  f47
	;;
	FNMA	f68  = f69,  f48, f68
	FNMA	f76  = f77,  f48, f76
	;;
	FNMA	f67  = f69,  f49, f67
	FNMA	f75  = f77,  f49, f75
	;;
	FNMA	f66  = f69,  f50, f66
	FNMA	f74  = f77,  f50, f74
	;;
	FNMA	f65  = f69,  f51, f65
	FNMA	f73  = f77,  f51, f73
	;;
	FNMA	f64  = f69,  f52, f64
	FNMA	f72  = f77,  f52, f72
	;;
	FMPY	f68  = f68,  f53
	FMPY	f76  = f76,  f53
	;;
	FNMA	f67  = f68,  f54, f67
	FNMA	f75  = f76,  f54, f75
	;;
	FNMA	f66  = f68,  f55, f66
	FNMA	f74  = f76,  f55, f74
	;;
	FNMA	f65  = f68,  f56, f65
	FNMA	f73  = f76,  f56, f73
	;;
	FNMA	f64  = f68,  f57, f64
	FNMA	f72  = f76,  f57, f72
	;;
	FMPY	f67  = f67,  f58
	FMPY	f75  = f75,  f58
	;;
	FNMA	f66  = f67,  f59, f66
	FNMA	f74  = f75,  f59, f74
	;;
	FNMA	f65  = f67,  f60, f65
	FNMA	f73  = f75,  f60, f73
	;;
	FNMA	f64  = f67,  f61, f64
	FNMA	f72  = f75,  f61, f72
	;;
	FMPY	f66  = f66,  f16
	FMPY	f74  = f74,  f16
	;;
	FNMA	f65  = f66,  f17, f65
	FNMA	f73  = f74,  f17, f73
	;;
	FNMA	f64  = f66,  f18, f64
	FNMA	f72  = f74,  f18, f72
	;;
	FMPY	f65  = f65,  f19
	FMPY	f73  = f73,  f19
	;;
	FNMA	f64  = f65,  f20, f64
	FNMA	f72  = f73,  f20, f72
	;;
	FMPY	f64  = f64,  f21
	FMPY	f72  = f72,  f21
	;;

	adds	BOFFSET  =  8 * SIZE, BOFFSET
	adds	BOFFSET2 =  8 * SIZE, BOFFSET2
	;;
	STFD	[BOFFSET]  = f68, SIZE
	STFD	[BOFFSET2] = f70, SIZE
	;;
	STFD	[BOFFSET]  = f76, SIZE
	STFD	[BOFFSET2] = f78, SIZE
	;;
	STFD	[BOFFSET]  = f69, SIZE
	STFD	[BOFFSET2] = f71, SIZE
	;;
	STFD	[BOFFSET]  = f77, - 11 * SIZE
	STFD	[BOFFSET2] = f79, - 11 * SIZE
	;;
	STFD	[BOFFSET]  = f64, SIZE
	STFD	[BOFFSET2] = f66, SIZE
	;;
	STFD	[BOFFSET]  = f72, SIZE
	STFD	[BOFFSET2] = f74, SIZE
	;;
	STFD	[BOFFSET]  = f65, SIZE
	STFD	[BOFFSET2] = f67, SIZE
	;;
	STFD	[BOFFSET]  = f73, - 3 * SIZE
	STFD	[BOFFSET2] = f75, - 3 * SIZE
	;;
	adds	C1 = -8 * SIZE, C1
	adds	C2 = -8 * SIZE, C2
	;;
#endif

#ifdef LT
	LDFPD	f32, f33 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f34, f35 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f36, f37 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f38, f39 = [AOFFSET]
	adds	AOFFSET = 3 * SIZE, AOFFSET
	;;
	LDFD	f40 = [AOFFSET], 1 * SIZE
	;;
	LDFPD	f41, f42 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f43, f44 = [AOFFSET], 2 * SIZE
	;;
	LDFPD	f45, f46 = [AOFFSET]
	adds	AOFFSET = 4 * SIZE, AOFFSET
	;;
	LDFPD	f47, f48 = [AOFFSET], 2 * SIZE
	;;