memory boundry alignment with stuctures

This is a Linux gcc question.
When defining a structure, do I have to be
concerned with how the variables aligned
themselves in memory? For example, if the,
if all the data types within the structure added
up to 32 bytes, would that be more efficient
then a structure of 33 bytes?
This used to be an issue ( the 16 byte memory
boundary used by Intel x86 as segments ) way
back in the dark ages, but since this is still the
age of Intel x86 processors, does memory
alignment in structures matter or no?

Yes, memory alignment in C structures still matter. There are various (non-portable) ways to ensure the alignment of the variables in memory will be aligned from one machine to the next. Some methods would be using #pragma or taking the union.

> do I have to be concerned with how the variables aligned themselves in memory?
No, that's the job of the compiler.

Because the compiler is free to add padding bytes between members of a structure, the size of a struct from one machine to another can be different.

If you write/read your structures to/from files in binary mode, alignement does matter. But then endianness matters as well.
Kurt