.text
	.allow_optimizations
	.file	"3"
	.section	.text.__runCodelet_AddLayerNormMoments,"ax",@progbits
	.globl	__runCodelet_AddLayerNormMoments # -- Begin function __runCodelet_AddLayerNormMoments
	.p2align	2
	.type	__runCodelet_AddLayerNormMoments,@function
__runCodelet_AddLayerNormMoments:       # @__runCodelet_AddLayerNormMoments
	.worker
.Lfunc_begin0:
# %bb.0:                                # %entry
	setzi $m6, 16
	cmpult $m6, $m6, __worker_stack_max_growth_plus_one
	brz $m6, _stackoverflow
	add $m11, $m12, -16
	get $m0, 1
	ld32 $m3, $m13, $m15, 4
	and $m1, $m0, 7
	ld32 $m0, $m13, $m15, 3
	ld32 $m2, $m13, $m15, 2
	brz $m3, .LBB0_1
# %bb.11:                               # %if.end.i
	ld64 $ap0, $m0, $m15, 0
	{
		ld64 $ap1, $m0, $m15, 1
		f32v2add $ap0, $ap0, $apzero
	}
	{
		ld64 $ap1, $m0, $m15, 2
		f32v2add $ap0, $ap0, $ap1
	}
	{
		ld64 $ap1, $m0, $m15, 3
		f32v2add $ap0, $ap0, $ap1
	}
	{
		ld64 $ap1, $m0, $m15, 4
		f32v2add $ap0, $ap0, $ap1
	}
	{
		ld64 $ap1, $m0, $m15, 5
		f32v2add $ap0, $ap0, $ap1
	}
	{
		cmpeq $m3, $m3, 1
		f32v2add $ap0, $ap0, $ap1
	}
	{
		st32 $m2, $m11, $m15, 1
		f32add $a0, $a0, $a1
	}
	ld32 $a1, $m11, $m15, 1
	f32fromui32 $a1, $a1
	{
		brz $m3, .LBB0_18
		f32div $a0, $a0, $a1
	}
# %bb.12:                               # %if.then15.i
	cmpult $m5, $m2, 96
	ld32 $m4, $m13, $m15, 6
	shl $m3, $m1, 3
	brnz $m5, .LBB0_15
# %bb.13:                               # %if.then15.i
	or $m5, $m2, $m4
	and $m5, $m5, 7
	brnz $m5, .LBB0_15
# %bb.14:                               # %if.then.i.i
	mov	$m5, $m15
	{
		shr $m2, $m2, 2
		f32sub $a7, $azero, $a0
	}
	sub $m2, $m2, $m1
	add $m1, $m11, 8
	add $m2, $m2, 5
	setzi $m6, 43691
	sort4x16lo $m7, $m2, $m15
	shr $m8, $m2, 16
	add $m2, $m4, $m3
	mul $m4, $m7, $m6
	mul $m6, $m8, $m6
	shr $m4, $m4, 16
	add $m4, $m6, $m4
	setzi $m6, 43690
	mul $m7, $m7, $m6
	mul $m6, $m8, $m6
	sort4x16lo $m8, $m4, $m15
	shr $m4, $m4, 16
	add $m7, $m7, $m8
	add $m4, $m6, $m4
	shr $m6, $m7, 16
	add $m4, $m4, $m6
	shr $m4, $m4, 2
	#APP
	setzi $a0, 8
	put	1, $a0
	mov	$a6, $a7
	{
		rpt $m4, 5
		fnop
	}

	{
		ld64step $ap2, $m15, $m2+=, 6
		fnop
	}

	{
		nop
		f16v2tof32 $ap0, $a4
	}

	{
		nop
		f16v2tof32 $ap1, $a5
	}

	{
		nop
		f32v2add $ap0, $a6:B, $ap0
	}

	{
		nop
		f32v2add $ap1, $a6:B, $ap1
	}

	{
		nop
		f32v4sqacc $aq0
	}

	f32v2gina $ap0, $apzero, 0
	f32v2gina $ap1, $apzero, 0
	f32v2add $ap0, $ap0, $ap1
	st64 $ap0, $m1, $m15, 0

	#NO_APP
	ld64 $ap3, $m11, $m15, 1
	st64 $ap3, $m3, $m0, 6
	exitz $m15
.LBB0_1:                                # %if.then.i
	{
		add $m3, $m2, 7
		zero	$ap0
	}
	shr $m3, $m3, 3
	cmpult $m6, $m1, $m3
	ld32 $m4, $m13, $m15, 0
	st64 $apzero, $m11, $m15, 1
	ld32 $m5, $m13, $m15, 5
	ld32 $m3, $m13, $m15, 6
	brz $m6, .LBB0_10
# %bb.2:                                # %if.end.i.i
	shr $m6, $m2, 3
	cmpult $m7, $m1, $m6
	brz $m7, .LBB0_3
# %bb.4:                                # %cond.true.i.i
	sub $m6, $m6, $m1
	add $m6, $m6, 5
	setzi $m7, 43691
	sort4x16lo $m8, $m6, $m15
	shr $m6, $m6, 16
	mul $m9, $m8, $m7
	mul $m7, $m6, $m7
	shr $m9, $m9, 16
	add $m7, $m7, $m9
	setzi $m9, 43690
	sort4x16lo $m10, $m7, $m15
	mul $m8, $m8, $m9
	shr $m7, $m7, 16
	mul $m6, $m6, $m9
	add $m8, $m8, $m10
	add $m6, $m6, $m7
	shr $m7, $m8, 16
	add $m6, $m6, $m7
	shr $m6, $m6, 2
	bri .LBB0_5
.LBB0_18:                               # %if.else.i
	brz $m1, .LBB0_19
# %bb.20:                               # %_ZN19AddLayerNormMoments7computeEj.exit
	exitz $m15
.LBB0_15:                               # %if.end.i60.i
	{
		shr $m5, $m2, 1
		zero	$ap3
	}
	cmpult $m2, $m1, $m5
	brz $m2, .LBB0_17
# %bb.16:                               # %if.then10.i66.i
	mov	$m2, $m15
	{
		sub $m5, $m5, $m1
		f32sub $a3, $azero, $a0
	}
	add $m5, $m5, 5
	setzi $m6, 43691
	sort4x16lo $m7, $m5, $m15
	shr $m5, $m5, 16
	mul $m8, $m7, $m6
	mul $m6, $m5, $m6
	shr $m8, $m8, 16
	add $m6, $m6, $m8
	setzi $m8, 43690
	mul $m7, $m7, $m8
	mul $m5, $m5, $m8
	sort4x16lo $m8, $m6, $m15
	shr $m6, $m6, 16
	add $m7, $m7, $m8
	add $m5, $m5, $m6
	shr $m6, $m7, 16
	add $m5, $m5, $m6
	shl $m1, $m1, 2
	shr $m5, $m5, 2
	add $m1, $m4, $m1
	#APP
	mov	$a4, $a3
	{
		rpt $m5, 4
		fnop
	}

	{
		ld32step $a0, $m15, $m1+=, 6
		fnop
	}

	{
		nop
		f16v2tof32 $ap0, $a0
	}

	{
		nop
		f32v2add $ap0, $a4:B, $ap0
	}

	{
		nop
		f32v2mul $ap0, $ap0, $ap0
	}

	{
		nop
		f32v2add $ap3, $ap3, $ap0
	}


	#NO_APP
.LBB0_17:                               # %_ZL7normSumPKDhS0_jjfb.exit.i
	st64 $ap3, $m3, $m0, 6
	exitz $m15
.LBB0_19:                               # %for.body25.preheader.i
	ld64 $ap1, $m0, $m15, 6
	{
		ld64 $ap2, $m0, $m15, 7
		f32v2add $ap1, $ap1, $apzero
	}
	{
		ld64 $ap2, $m0, $m15, 8
		f32v2add $ap1, $ap1, $ap2
	}
	{
		ld64 $ap2, $m0, $m15, 9
		f32v2add $ap1, $ap1, $ap2
	}
	{
		ld64 $ap2, $m0, $m15, 10
		f32v2add $ap1, $ap1, $ap2
	}
	{
		ld64 $ap2, $m0, $m15, 11
		f32v2add $ap1, $ap1, $ap2
	}
	{
		ld32 $m0, $m13, $m15, 1
		f32v2add $ap1, $ap1, $ap2
	}
	f32add $a1, $a2, $a3
	st64 $ap0, $m0, $m15, 0
	exitz $m15
.LBB0_3:
	mov	$m6, $m15
.LBB0_5:                                # %cond.end.i.i
	shl $m8, $m1, 4
	#APP
	setzi $a0, 8
	put	1, $a0
	#NO_APP
	shl $m7, $m1, 3
	add $m4, $m4, $m8
	add $m5, $m5, $m8
	add $m3, $m3, $m8
	brz $m6, .LBB0_7
# %bb.6:                                # %if.then10.i.i
	#APP
	{
		rpt $m6, 6
		fnop
	}

	{
		ld64step $ap0, $m15, $m4+=, 1
		fnop
	}

	{
		ld64step $ap1, $m15, $m4+=, 11
		fnop
	}

	{
		ld64step $ap2, $m15, $m5+=, 1
		fnop
	}

	{
		ld64step $ap3, $m15, $m5+=, 11
		f16v4add $ap0, $ap0, $ap2
	}

	{
		nop
		f16v4add $ap1, $ap1, $ap3
	}

	{
		st64step $ap0, $m15, $m3+=, 1
		fnop
	}

	{
		st64step $ap1, $m15, $m3+=, 11
		f16v8acc $aq0
	}


	#NO_APP
.LBB0_7:                                # %if.end13.i.i
	mul $m6, $m6, 48
	add $m6, $m6, $m7
	cmpult $m2, $m6, $m2
	brz $m2, .LBB0_9
# %bb.8:                                # %if.then18.i.i
	#APP
	ld64 $ap0, $m4, $m15, 0
	ld64 $ap2, $m5, $m15, 0
	f16v4add $ap0, $ap0, $ap2
	zero	$ap1
	{
		st64 $ap0, $m3, $m15, 0
		f16v8acc $aq0
	}


	#NO_APP
.LBB0_9:                                # %if.end19.i.i
	add $m2, $m11, 8
	#APP
	f32v2gina $ap0, $apzero, 0
	f32v2gina $ap1, $apzero, 0
	f32v2add $ap0, $ap0, $ap1
	f32v2gina $ap1, $apzero, 0
	f32v2add $ap0, $ap0, $ap1
	f32v2gina $ap1, $apzero, 0
	f32v2add $ap0, $ap0, $ap1
	st64 $ap0, $m2, $m15, 0

	#NO_APP
	ld64 $ap0, $m11, $m15, 1
.LBB0_10:                               # %_ZL15normAddAndStorePKDhS0_PDhjj.exit.i
	st64 $ap0, $m0, $m15, $m1
	exitz $m15
.Lfunc_end0:
	.size	__runCodelet_AddLayerNormMoments, .Lfunc_end0-__runCodelet_AddLayerNormMoments
	.section	.stack_sizes,"o",@progbits,.text.__runCodelet_AddLayerNormMoments
	.long	.Lfunc_begin0
	.byte	16
	.section	.text.__runCodelet_AddLayerNormMoments,"ax",@progbits
                                        # -- End function
	.ident	"clang version 16.0.0 (git@github.com:Software-GCAI/llvm-project.git 6343a843cd5af31429c62a83b6ba249a74d36969)"
	.section	".note.GNU-stack","",@progbits
	.addrsig