Simple asm delay ?

[ultiblade]

Google came up with an interesting piece of code.
I did some tests and it seems to do the trick.

If called using "@ wait 0" it delays 1uS,
if called using "@ wait 1" it delays about 2,5uS,
if called using "@ wait 2" it delays about 3uS.

asm
shortwait macro del  ; used by main wait macro
    if (del>3) && (del<8) 
        call del7+(7-(del)) 
    endif 
    if del <4 
        if del>=2 
            goto $+1
        endif
        if ((del) & 1) == 1
            nop
        endif
    endif
    endm
endasm

asm
wait macro del 
;    if (del)<0 error 'negative delay' 
;    endif 
    if (del)>=8 
        movlw ((del)-8) /4 
        call delay
        shortwait ((del-8) % 4) 
    else
        shortwait (del)
     endif
    endm

delay  
    addlw 0xff 
    skpnc 
    goto delay
    return
 
del7 goto del5      ; 7 cycle delay
del6 goto del4      ; 6 cycle delay
del5 nop            ; 5 cycle delay
del4 return         ; 4 cycle delay
ENDasm

So, the delay is not the problem anymore.
However, when called from picbasic like :

led1 var word
led2 var word

led1 = 2
led2 = 2

start:

GPIO = %01
@ wait _led1
GPIO = %10
@ wait _led2

goto start

The delay is about 33uS instead of 2. I understand that the delay would be longer, but 30uS extra ??

Can someone, please, point me in the right direction ?