Sean McBride wrote: The technique apple suggests, which I have done successfully, is to limit your floating point work to a shared library. Then this lib can be built with appropriate project settings. I forget the name, but one of Apple's sample shows this. Its some coreaudio sample driver I think. Originally, it was recommended to save cpu cycles, I think (because the fpu emulation is a little expensive, especially if used to crunch numbers). Now, the reasoning could be extended to "because the compiler's buggy". However, if I have float calculations scattered across my code, it really makes no difference if I call a library to do the calculations. For now, I like the convenience better than the speed gain of having all my float code concentrated in one place. Plus, I'm hoping the soft-fpu will get fixed eventually. So, I still need to know how to build a .kext with working float calculations. Is it possible without the library? Heiko _______________________________________________ darwin-kernel mailing list | darwin-kernel@lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.