Jtdcftul

I am wondering will compiler make padding result different between 32bit and 64bit system. So I write the code below in a simple VS2019 C++ console project today.

struct Z

{

    char s;

    __int64 i;

};



int main()

{

    std::cout << sizeof(Z) <<"n"; 

}

What I expected on each "Platform" setting

x86: 12

X64: 16

Actual result

x86: 16

X64: 16

Since the memory word size on x86 is 4 byte. This means it has to store the 16 bytes 'i' in two different words. So I thought the compiler will do padding in this way:

struct Z

{

    char s;

    char _pad[3];

    __int64 i;

};

So may I know what's the reason behind this?

1. For forward-compatibility with the 64bit system?


2. Due to the limitation of supporting 64bit number on the 32bit processor?

The padding is not determined by the word size, but by the alignment of each data type.

In most cases, the alignment requirement is equal to the type's size. So for a 64 bit type like int64 you will get an 8 byte (64 bit) alignment. Padding needs to be inserted into the struct to make sure that the storage for the type ends up at an address that is properly aligned.

You may see a difference in padding between 32 bit and 64 bit when using built-in datatypes that have different sizes on both architectures, for instance pointer types (int*).

"32-bit" and "64-bit" refer to the size/alignment of pointers. If you use void* instead of __int64, you will see your expected result.

Any x86-based system could store its __int64 field starting from an address not divisible by 8, but that would be inefficient - for example, this could split it between two cache lines. So all "basic" data types require their alignment to be a multiple of their size.

This is a matter of alignment requirement of the data type as specified in
Padding and Alignment of Structure Members

Every data object has an alignment-requirement. The alignment-requirement for all data except structures, unions, and arrays is either the size of the object or the current packing size (specified with either /Zp or the pack pragma, whichever is less).

And the default value for structure member alignment is specified in /Zp (Struct Member Alignment)

The available packing values are described in the following table:

/Zp argument Effect

1 Packs structures on 1-byte boundaries. Same as /Zp.

2 Packs structures on 2-byte boundaries.

4 Packs structures on 4-byte boundaries.
8 Packs structures on 8-byte boundaries (default for x86, ARM, and ARM64).

16 Packs structures on 16-byte boundaries (default for x64).

Since the default for x86 is /Zp8 which is 8 bytes, the output is 16.

However, you can specify a different packing size with /Zp option.

Here is a Live Demo with /Zp4 which gives the output as 12 instead of 16.