AVR Arithmatic

Hi,

the code in your attachment sqrt32_Fb.txt looks different compared to the one at https://www.mikrocontroller.net/articles/AVR_Arithmetik#Wurzel

1

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

2

;  Fast and short 32 bits AVR sqrt routine, avr-gcc ABI compliant

3

;  R25:R24 = SQRT (R25:R24:R23:R22)

4

;  rounded down to integer

5

;     Destroys R26,R27,R22,R23,R18,R19

6

;  Cycles incl call & ret = 260-300

7

;  Stack incl call = 2-3

8

.text

9

.global sqrt32_floor

10

.type sqrt32_floor, @function

11

12

sqrt32_floor:

13

    ldi   R19, 0xc0

14

    clr   R18               ; rotation mask in R19:R18

15

    ldi   R27, 0x40

16

    sub   R26, R26          ; developing sqrt in R27:R26, C=0

17

1:  brcs  2f                ; C --> Bit is always 1

18

    cp    R24, R26

19

    cpc   R25, R27          ; Does test value fit?

20

    brcs  3f                ; C --> nope, bit is 0

21

2:  sub   R24, R26

22

    sbc   R25, R27          ; Adjust argument for next bit

23

    or    R26, R18

24

    or    R27, R19          ; Set bit to 1

25

3:  lsr   R19

26

    ror   R18               ; Shift right mask, C --> end loop

27

    eor   R27, R19

28

    eor   R26, R18          ; Shift right only test bit in result

29

    rol   R22               ; Bit 0 only set if end of loop

30

    rol   R23

31

    rol   R24

32

    rol   R25               ; Shift left remaining argument (C used at 1:)

33

    sbrs  R22, 0            ; Skip if 15 bits developed

34

    rjmp  1b                ; Develop 15 bits of the sqrt

35

36

    brcs  4f                ; C--> Last bits always 1

37

    lsl   R23               ; Need bit 7 in C for cpc

38

    cpc   R26, R24

39

    cpc   R27, R25          ; After this C is last bit

40

41

4:  adc   R26, R19          ; Round up if C (R19=0)

42

    mov   R25, R27          ; return in R25:R24 as for avr-gcc ABI

43

    mov   R24, R26

44

    ret

45

46

.size sqrt32_floor, .-sqrt32_floor

Ok, thanks

but where is jump locations ?

see line 16, brcs 2f see line 20, brcs 3f

assembler shows an error on this instructions

Kishor S. wrote:

Ok, thanks

but where is jump locations ?

see line 16, brcs 2f see line 20, brcs 3f

assembler shows an error on this instructions

These notations are from C inline assembly language, which is very different from an directly usable assembly language.

The understanding of this crude notation is simple: the number describes the relative branch distance (in instruction words), the trailing character denotes, if it is a forward branch or a backward branch. Of course the 'f' stands for 'forward'.

So your task is: count instruction words up from the branch instruction and insert a label at found point in source code. Finally replace the branch target parameter by the name of this label.

C-hater wrote:

The understanding of this crude notation is simple: the number describes the relative branch distance (in instruction words), the trailing character denotes, if it is a forward branch or a backward branch. Of course the 'f' stands for 'forward'.

So your task is: count instruction words up from the branch instruction and insert a label at found point in source code. Finally replace the branch target parameter by the name of this label.

That's so wrong, even the opposite wouldn't be any better!

1f or 1b (or any other number) simply points to a numbered label in forward or backward direction.

Thanks a lot,

I learn this new thing, after 25 yrs. of coding practice..........

after so many efforts, I can't be able to assemble this code in assembly language.

I cant interpret this keyword "brcs 2f" "brcs 2b" and sbrs R22,0 (at line 33)

can anyone explain ?