在win32，win64下，float运算结果不同是什么原因

[秋风之刀]

有个c++程序分别编译为32bit和64bit，分别在win7-32bit和win7-64bit上运行，该程序中存在大量的float运算，运算结果中，有一部分结果会发生差异，这是什么原因造成的？

[paschen]

float在32位和64位下不都是4字节 你先确定你问题是由于float导致的？ 另外浮点数可能有精度丢失问题，但如果代码一样，32位与64位也应该是一样的

[赵4老师]

Collapse AllExpand All Code: All Code: Multiple Code: Visual Basic Code: C# Code: Visual C++ Code: J# Code: JScript Visual Basic C# Visual C++ J# JScript Run-Time Library Reference _set_controlfp See Also Send Feedback Sets the floating-point control word. void __cdecl _set_controlfp( unsigned int newControl, unsigned int mask ); Parameters newControl New control-word bit values. mask Mask for new control-word bits to set. Return Value None. Remarks The _set_controlfp is similar to _control87, but it only sets the floating-point control word to newControl. The bits in the values indicate the floating-point control state. The floating-point control state allows the program to change the precision, rounding, and infinity modes in the floating-point math package. You can also mask or unmask floating-point exceptions using _set_controlfp. For more information, see _control87, _controlfp, __control87_2. This function is deprecated when compiling with /clr (Common Language Runtime Compilation) or /clr:pure because the common language runtime only supports the default floating-point precision. Requirements Routine Required header Compatibility _set_controlfp <float.h> x86 processor only For more compatibility information, see Compatibility in the Introduction. See Also Concepts Floating-Point Support _clear87, _clearfp _status87, _statusfp, _statusfp2 Send feedback on this topic to Microsoft.

[fefe82]

浮点运算是有误差的。 分别按照 32bit 和 64bit 编译的时候，有可能生成不同的指令，计算结果也可能有细微的差别。浮点运算很多的时候，这些误差有可能积累起来，导致运行结果不同。

[u010231493]

64位也有很多种， l64、f64 32位也同样。 所以跨平台编程不会使用int long float这种模糊的类型。。size32 size 16等这种才会被使用