Removing _fltused from import table
Author
Hello,
My compiler (Intel C++ compiler 9.1, MSVC 6.0/2003/2005 compatible) generates object files which imports symbol `_fltused` when floating point constants/functions are used in source code. Is there any way to disable this using compiler parameters? I don't need any functionality from run-time library and therefore I don't need any stuff related to it in my object files.
Thank you.
Why is it a bad thing?
Also, if you use floating point at all in your program, there is a very good chance you do need support from the runtime because many normal floating point operations (such as float-to-int conversion) are much more complicated than simply using store-float-as-int assembly instructions due to the requirements of the C++ language specification. In MSVC 2002, at least, that operation is actually performed by a very long function that is part of the runtime library.
Also, if you use floating point at all in your program, there is a very good chance you do need support from the runtime because many normal floating point operations (such as float-to-int conversion) are much more complicated than simply using store-float-as-int assembly instructions due to the requirements of the C++ language specification. In MSVC 2002, at least, that operation is actually performed by a very long function that is part of the runtime library.
Author
Quote:Original post by Extrarius
Why is it a bad thing?
Also, if you use floating point at all in your program, there is a very good chance you do need support from the runtime because many normal floating point operations (such as float-to-int conversion) are much more complicated than simply using store-float-as-int assembly instructions due to the requirements of the C++ language specification. In MSVC 2002, at least, that operation is actually performed by a very long function that is part of the runtime library.
I have just tried to convert floating point number to integer and linker didn't show any errors about unresolved externals. Code was not removed by optimizer - I can see it in actual assembly output (fistp FPU instruction is used) so I'm sure I can live without run-time libraries.
I think you are talking about emulating floating point functionality (long functions to convert float to integer, etc.).
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
