On Mon, 12 Apr 2010 13:56:23 -0400, Jérôme M. Berger <jeber...@free.fr>
wrote:
Here's a fast 100% precise solution:
==============================8<------------------------------
uint32_t maxOr (uint32_t minA, uint32_t minB,
uint32_t maxA, uint32_t maxB)
{
assert (minA <= maxA);
assert (minB <= maxB);
if (maxA == 0) return maxB;
if (maxB == 0) return maxA;
uint32_t a = maxA ^ minA;
if (a != 0) {
a = ((1 << (highbit (a)+1)) - 1) & maxA & maxB;
if (a != 0)
a = (1 << highbit (a)) - 1;
}
uint32_t b = maxB ^ minB;
if (b != 0) {
b = ((1 << (highbit (b)+1)) - 1) & maxA & maxB;
if (b != 0)
b = (1 << highbit (b)) - 1;
}
return maxA|maxB|a|b;
}
------------------------------>8==============================
Fails for test case:
minA = 4, maxA = 4, minB = 4, maxB = 6 (outputs 7, accurate result is 6).
I'm not quite sure what you are trying to do, but I think you are on the
right path to getting a faster solution.
-Steve