I would put very short update, regarding my PWM trial

Corresponding ASM code is as follows. Indeed I greatly refer the C code at avrfreak and tried to convert to assembler by myself (so such a redundant code, ;-)

.def temp1 = R16
.def temp2 = R17
.def count_L = R18
.def count_H = R19
.def final_L = R20
.def final_H = R21
.def temp3 = R22
.def temp4 = R23
.def temp5 = R24
.def temp6 = R25
.equ topnum = 1000
.equ count_start = 0xff
.equ count_end = 0x01
.equ subnum = 0x01
; Replace with your application code
setup:
	ldi temp1, (1<<PB0)+(1<<PB1);
	out DDRB, temp1
	ldi temp1, (1<<COM0A1)+(1<<COM0B1)+(1<<WGM01);
	out TCCR0A, temp1
	ldi temp1, low(topnum);1111101000
	out ICR0L, temp1
	ldi temp1, byte2(topnum)
	out ICR0H, temp1
	ldi temp1,(1<<CS01)+(1<<WGM03)+(1<<WGM02)
	out TCCR0B, temp1

start:
    ;inc r16
	ldi count_L, 0
	ldi count_H, 0
	ldi final_L, low(topnum)
	ldi final_H, byte2(topnum)
	ldi temp1, 0x01
	ldi temp2, 0x00
loop1:
	add count_L, temp1
	adc	count_H, temp2
	out OCR0AL, count_L
    out OCR0AH, count_H

	mov temp3, count_L
	mov temp4, count_H
	ldi temp5, low(topnum)
	ldi temp6, byte2(topnum)
	sub temp5,temp3
	sbc temp6,temp4
	out OCR0BL,	temp5
	out OCR0BH, temp6
	rcall delay
	cp final_H, count_H
	cpc final_L, count_L
	brne loop1
    rjmp start


delay:	; subroutine DELAY
	ldi	temp3, count_end
	ldi	temp4, count_start
	ldi	temp5, subnum
counter:
	sub	temp4, temp5
	cpse	temp4, temp3
	rjmp counter	
	ret