Re: Any more graceful way to saturate adds?

Jim Leonard wrote:
> I am working in 16-bit 808x assembler (yes, on an old 808x CPU) and
> adding two 8-bit numbers to each other and saturating the result.
> Here is the best I could come up with:
> 
> ; AL and DL have unsigned 8-bit values to add
> ; DL is actually extended into DX (ie. 00xxh)
>    cbw            ; Make AL a signed word
>    add   dx,ax    ; Do word sized add, into DX
>    jns   @[EMAIL PROTECTED]
   ; jump if result not signed/negative
>    xor   dx,dx    ; Underflowed; make 0
> @[EMAIL PROTECTED]
>    or   dh,dh     ; Did it overflow?
>    jz   @[EMAIL PROTECTED]
      ; if not, we're ok
>    mov   dx,00FFh ; Overflowed; force our saturation value (FFh)
> @[EMAIL PROTECTED]
> 
> This is part of an ADPCM inner loop.
> 
> Is this about as fast as I can expect?  Or am I missing an obvious
> technique?  (I am not an 808x assembler expert)

What's wrong with the obvious approach?

  add dl,al	; Carry set if it overflows!
  sbb al,al	; AL = 0xff if overflow
  or dl,al	; Turns any overflow into 0xff

OK?

Terje
-- 
- <Terje.Mathisen@[EMAIL PROTECTED]
>
"almost all programming can be viewed as an exercise in caching"