max and min of 2's complement

pgbpsu · 2009-04-16 01:57

I'm trying to grab the max and min values (in SPIN) from an array of 24-bit 2's complement data. Any clever suggestions? I'm afraid I need to convert all the data into signed ints, then scan them. Is there a better way, a way to determine this without converting them first?

Thanks,
Peter

Mike Green · 2009-04-16 02:11

I assume you have 24 bit 2's complement values stored in an array of 32 bit longs. I suggest you convert them in the process of scanning for the maximum and minimum like this:

CON tableSize = 32
VAR long theMax, theMin, table[noparse][[/noparse]tableSize]
PRI findMinMax | i, temp
   theMin := POSX   ' Anything is lower than this
   theMax := NEGX   ' Anything is higher than this
   repeat i from 0 to tableSize-1   ' Go through the table
      temp := (table[noparse][[/noparse] i ] << 8) ~> 8   ' Convert to signed long
      theMin <#= temp   ' Find the minimum value in the table
      theMax #>= temp   ' Find the maximum value in the table

Phil Pilgrim (PhiPi) · 2009-04-16 02:18

You don't need to do complete conversions during your scan. Just shift them all left by 8 bits to compare; then, when you're done, arithmetically shift the min and max right by eight bits to get your result. Spin makes this kind of testing easy with its limit minimum and limit maximum operators:

x := candidate << 8
minval <#= x
maxval #>= x

-Phil

pgbpsu · 2009-04-16 13:19

Mike and Phil-

Thank you both for the prompt and useful suggestions. I should have know shifting was required because my sign bit isn't in bit 31. I think either one of these solutions will work quite well.

Thanks,
Peter

max and min of 2's complement

Comments