; file firlib\src\fir_sp_asm.s
; brief FIR filtering function in canonical C6600 assembler (single precision IEEE754)
; Advices for assembler procedure development :
; - step 1 : procedure call and return
; - step 2 : check loop exit condition and number of iterations
; - step 3 : check data loading and storage with complex index generation 
; - step 4 : check for dot product implementation
; - step 5 : during test and validation, always check limits values
; author 
; date

	.global fir_sp_asm
	
; C prototype :						match registers :
;
; void fir_sp_asm ( const float * restrict xk, 	-> A4
;		   			const float * restrict a,	-> B4
;		   			float * restrict yk, 		-> A6
;		   			int na,						-> B6
;		   			int nyk);					-> A8
fir_sp_asm:
				;int i, j;
				MV				A8,A0		;i<nyk
				SUB				B6,1,B6
								
fir_sp_asm_l1:	; input array loop			
				;for (i=0; i<nyk; i++) {
				;	tmp = 0;
				ZERO			A5			
				MV				B6,A1		;j<na				
				MV				A4,A19
				MV				B4,B19					
											
fir_sp_asm_l2:	; FIR filter algorithm - dot product	
				;for (j=0; j<na; j++){
				;	tmp += a[j]*xk[i+j];
				LDW				*A19++,A3
				NOP				4				
				LDW				*B19++,B9
				NOP				4				
				MPYSP			A3,B9,A17
				NOP				3
				ADDSP			A17,A5,A5					
				NOP				3
;		[A1] 	SUB				A1,1,A1	
;		[A1] 	B				fir_sp_asm_l2											
		[A1] 	BDEC			fir_sp_asm_l2,A1
				NOP				5				
				; end of FIR filter algorithm - dot product	
				
				STW				A5,*A6++	;yk(i) = tmp;
				ADD				A4,4,A4  	;i+j (float -> 4b)	
;		[A0] 	SUB				A0,1,A0	
;		[A0] 	B				fir_sp_asm_l1														
		[A0]	BDEC			fir_sp_asm_l1,A0
				NOP				5
				; end of input array loop	
						
				B				B3
				NOP				5
				; end of fir_sp_asm procedure