Why changing the order of struct members can change the size of it?

[babaliaris]

I remember when I was learning assembly as a colleague (I still am by the way), in a programming lab the teachers showed as that if you have a struct in c and then see the size , change the order of the members and see the size again you might notice a different value. They said this was something to do about linear organization of the memory or something and that computers like to organize the memory in even numbers (which are divided by 2) but I can't remember actually why that was.

Can someone explain it to me or give me a link where i can read the reason?
Also is this something that i might be worried about as a programmer?

Thanks!

[2kaud]

What they are referring to is data alignment - which is quite a complicated topic.

Have a look at
https://stackoverflow.com/questions/...casting-pointe
https://stackoverflow.com/questions/...***-determined
https://stackoverflow.com/questions/...er-inheritance

[babaliaris]

What they are referring to is data alignment - which is quite a complicated topic.

Have a look at
https://stackoverflow.com/questions/...casting-pointe
https://stackoverflow.com/questions/...***-determined
https://stackoverflow.com/questions/...er-inheritance

I remembered now. The thing is, as a programmer should I worry about that? Using the sizeof() function I can always get the correct size of a struct or class and also using the macro offsetof(st, m) for structures i can also easily take the offset of each member, so i should not worry about the alignment i guess.

I'm asking these question because lately I'm dealing with low lever programming (Graphics Programming) and I'm playing a lot with raw bytes, so i just want to make sure I understand everything correctly.

[wolle]

I'm asking these question because lately I'm dealing with low lever programming (Graphics Programming) and I'm playing a lot with raw bytes, so i just want to make sure I understand everything correctly.

Well, you need to pack data according to how the Graphics subsystem expects to find it. It means you need to know enough about the programming language and the tools you're using to accomplish the expected data layouts. There is no way around it when you have systems that communicate at a low level by passing addresses to blocks of memory in order to exchange information.

What you need is a good tutorial and lots of examples. I've recently used DirectX 12 to display movable objects in a 3D scene. I could never have done it without the help of the book by Frank D. Luna and even so it was a daunting task.