.section	__TEXT,__text,regular,pure_instructions
	.build_version macos, 11, 0	sdk_version 11, 3
	.globl	__Z13CalcCorrCoef_PfS_PKfS1_mf  ; -- Begin function _Z13CalcCorrCoef_PfS_PKfS1_mf
	.p2align	2
__Z13CalcCorrCoef_PfS_PKfS1_mf:         ; @_Z13CalcCorrCoef_PfS_PKfS1_mf
	.cfi_startproc
; %bb.0:
	; InlineAsm Start




	cbz	x4, LInvalidArg	; if n == 0 goto END 
	tst	x2, #0xf	; if (x2 != 15)  
	b.ne	LInvalidArg	;   goto END         
	tst	x3, #0xf	; if (x3 != 15)  
	b.ne	LInvalidArg	;   goto END         
	mov	x5, x4	; save n to x5       

	eor.16b	v16, v16, v16	; sum_x = 0    
	eor.16b	v17, v17, v17	; sum_y = 0    
	eor.16b	v18, v18, v18	; sum_xx = 0   
	eor.16b	v19, v19, v19	; sum_yy = 0   
	eor.16b	v20, v20, v20	; sum_xy = 0   

	cmp	x4, #16	; if n<=16              ; =16
	b.lo	LSkipLoop1	;   goto SkipLoop1   

LLoop1:
	ld1.4s	{ v0, v1, v2, v3 }, [x2], #64	; load x[0:16] 
	ld1.4s	{ v4, v5, v6, v7 }, [x3], #64	; load y[0:16] 

	fadd.4s	v16, v16, v0	; sum_x: v16 += VregX  
	fadd.4s	v17, v17, v4	; sum_y: v17 += VregX  
	fmla.4s	v18, v0, v0	; sum_xx: v18 += VregX^2 
	fmla.4s	v19, v4, v4	; sum_yy: v19 += VregY^2 
	fmla.4s	v20, v0, v4	; sum_xy: v20 += VregX * VregY 

	fadd.4s	v16, v16, v1	; sum_x: v16 += VregX  
	fadd.4s	v17, v17, v5	; sum_y: v17 += VregX  
	fmla.4s	v18, v1, v1	; sum_xx: v18 += VregX^2 
	fmla.4s	v19, v5, v5	; sum_yy: v19 += VregY^2 
	fmla.4s	v20, v1, v5	; sum_xy: v20 += VregX * VregY 

	fadd.4s	v16, v16, v2	; sum_x: v16 += VregX  
	fadd.4s	v17, v17, v6	; sum_y: v17 += VregX  
	fmla.4s	v18, v2, v2	; sum_xx: v18 += VregX^2 
	fmla.4s	v19, v6, v6	; sum_yy: v19 += VregY^2 
	fmla.4s	v20, v2, v6	; sum_xy: v20 += VregX * VregY 

	fadd.4s	v16, v16, v3	; sum_x: v16 += VregX  
	fadd.4s	v17, v17, v7	; sum_y: v17 += VregX  
	fmla.4s	v18, v3, v3	; sum_xx: v18 += VregX^2 
	fmla.4s	v19, v7, v7	; sum_yy: v19 += VregY^2 
	fmla.4s	v20, v3, v7	; sum_xy: v20 += VregX * VregY 

	sub	x4, x4, #16	; n -= 16       ; =16
	cmp	x4, #16	; if x4 >= 16           ; =16
	b.hs	LLoop1	;   goto Loop     

LSkipLoop1:
	faddp.4s	v16, v16, v16	; lane0=lane0+lane1,lane1=lane2+lane3 
	faddp.4s	v16, v16, v16	; s16 = lane0+lane1      
	faddp.4s	v17, v17, v17	; lane0=lane0+lane1,lane1=lane2+lane3 
	faddp.4s	v17, v17, v17	; s17 = lane0+lane1      
	faddp.4s	v18, v18, v18	; lane0=lane0+lane1,lane1=lane2+lane3 
	faddp.4s	v18, v18, v18	; s18 = lane0+lane1      
	faddp.4s	v19, v19, v19	; lane0=lane0+lane1,lane1=lane2+lane3 
	faddp.4s	v19, v19, v19	; s19 = lane0+lane1      
	faddp.4s	v20, v20, v20	; lane0=lane0+lane1,lane1=lane2+lane3 
	faddp.4s	v20, v20, v20	; s20 = lane0+lane1      
	cbz	x4, LSkipLoop2	; if x4==0 goto SkipLoop2      

LLoop2:
	ldr	s1, [x2], #4	; s1 = x; x+=4     
	ldr	s2, [x3], #4	; s2 = y; y+=4     
	fadd	s16, s16, s1	; s16 += s1        
	fadd	s17, s17, s2	; s17 += s2        
	fmla.4s	v18, v1, v1[0]	; v18 += s1 * s1   
	fmla.4s	v19, v2, v2[0]	; f19 += s2 * s2   
	fmla.4s	v20, v1, v2[0]	; f20 += s1 * s2   
	subs	x4, x4, #1	; if (--n !=0)      ; =1
	b.ne	LLoop2	;   goto Loop2     

LSkipLoop2:
	stp	s16, s17, [x1], #8	; [x1]=s16,s17; x1+=8 
	stp	s18, s19, [x1], #8	; [x1]=s18,s19; x1+=8 
	str	s20, [x1]	; [x1]=s20; x1+=8  

	; rho numerator                              
	scvtf	s21, x5	; s21 = n          
	fmul	s1, s21, s20	; s1 = n * sum_xy  
	fmls.4s	v1, v16, v17[0]	; s1 -= sum_x * sum_y 

	; rho denominator                            
	fmul	s2, s21, s18	; s2 = n * sum_xx  
	fmsub	s2, s16, s16, s2	; s2 = s2 - sum_x * sum_x 
	fsqrt	s2, s2	; s2 = sqrt(s2)    

	fmul	s3, s21, s19	; s2 = n * sum_yy 
	fmsub	s3, s17, s17, s3	; s2 = s3 - sum_y * sum_y 
	fsqrt	s3, s3	; s3 = sqrt(s3)     

	fmul	s4, s2, s3	; s4 = s2 * s3      
	fcmp	s4, s0	; if rho_den < epsilon 
	b.lo	LBadRhoDen	;   goto BadRhoDen  

	fdiv	s5, s1, s4	; s5 = rho          
	str	s5, [x0]	; [x0] = s5         
	mov	w8, #1	; return code: success 
	b	LReturn

LBadRhoDen:
	eor.16b	v5, v5, v5	; rho = 0   
	str	s5, [x0]	; [x0] = rho        

LInvalidArg:
	mov	w8, #0	; return code: fail  
LReturn:

	; InlineAsm End
	and	w0, w8, #0x1
	ret
	.cfi_endproc
                                        ; -- End function
.subsections_via_symbols