Vincent Lefevre wrote:
No, in my tests, I was using the cos() function from glibc, which is *not* implemented by the fcos instruction of the processor, but by a software algorithm from IBM.
okay for me. The question of Christoph was if his problem was because of a bug in gcc or somewhere in the software path. But taking his angle and calling fcos from an assembly program (to get rid of other sources of problems) gives different results on AMD or intel 64b CPUs. So my answer was: don't look for a software problem. AMD and Intel CPUs don't round the same for your angle. Now, which is right and which is wrong... So we had the same binary and different results (last bit). But maybe I misunderstand your point, so let me go back to work. :-)
