I have been trying to use the gcc extension __int128, but although I am able
to declare a variable, when I test it by assigning a value, gcc warns me:
"constant value is too large for its type".
I am using gcc 4.4.4 on 64 bit Slackware 13.1 with a core 2 duo processor.
This is the code Im using to test it:
#include <iostream>
int main ()
{
__int128_t wibble;
wibble=170141183460469231731687303715884105727;
std::cout << "size of wibble: " << sizeof(wibble) << std::endl;
return 0;
}
Unless Im mistaken, the maximum value that a signed int128 should be able to
cope with is 170141183460469231731687303715884105727. However I tried
knocking off a few digits to see what happened and the compiler is still
complaining. The output of the above program displays a sizeof result of 16
for "wibble" indicating that it is seeing __int128_t as a 16 byte (128 bit)
integer. So I guess the problem is with the assignment, perhaps something
Ive missed?
I put this on the gcc list because although it is probably a c++ problem,
the only resources I can find relevant to this are related to external math
libraries.
Has anyone else come accross this problem / can see what I have done wrong
and if so, please can you advise.
Regards
TS
