Bitshift on 64-bit integers.

Originally Posted by quzah

Code:

uint64_t shift(uint64_t to_shift, int pos) {
    if( pos > 32 )
    {
        to_shift <<= 32;
        pos -= 32;
    }
    return to_shift << pos;
}

Something like that perhaps?

Quzah.

My fix was to switch from lcc-32 to gcc. It now recognizes shifts using variables denoting the amount to shift, shrinking my code by alot.

Originally Posted by maththeorylvr

but uint_64 doesnt shift correctly when I try to use a variable to represent the number of bits to shift

I built the following with lcc-win32 and it generated the expected results. It shifted by an amount contained in a variable, and the amount could be greater than 32.

Code:

#include <stdio.h>
#include <stdint.h>
#include <inttypes.h>

uint64_t shift(uint64_t value, int bits)
{
   return value << bits;
}

int main( void )
{
   uint64_t value = ~0ULL;
   int bits;
   for ( bits = 0; bits < 64; ++bits )
   {
      uint64_t result = shift(value, bits);
      printf("shift(%016"PRIx64", %2d) = %016"PRIx64"\n", value, bits, result);
   }
   return 0;
}

I'm only mentioning this because...

Originally Posted by maththeorylvr

My fix was to switch from lcc-32 to gcc. It now recognizes shifts using variables denoting the amount to shift

...I get nervous about blaming a compiler for my woes; more often there is something I missed in the original code. FWIW.