padding tradeoff

Posted on 2013-06-02
Last Modified: 2013-06-04
Compilers do padding of struture members to improve the performance with the expense of memory.  (its a well-known trade-off)


typedef struct {
      uint32_t a;   //4 byte
      uint64_t b;   // 8 byte

On a 32 bit machine, sizeof(TEST) will be 12.
On a 64-bit machine, sizeof(TEST) will be 16. (because of the extra padding added for after member a)

1) What is the default behavior of compiler in gcc? Will it add paading by default or will it complie it a packed structure without padding? Is there a #define to contol this behavior?

2) Lets say if it is adding padding, does it "zeroed" those extra padded bytes? How would the runtime know how mich data to be read for member "a" and what is the starting address of member "b" as there is 4 bytes padded after "a"
Question by:perlperl
  • 2
LVL 86

Accepted Solution

jkr earned 500 total points
Comment Utility
The default behaviour is to add packing - if you don't want that, you can turn that off or fine-tune the behaviour by using '-fpack-struct[=n]' (see also the docs at

And to address the 2nd part of your question: The runtime does not care if *your* structs are padded otr not, since it is solely your code (compiled by gcc/g++) that accesses it, and every code that is supposed to deal with these has be properly instrumentated by gcc/g++ during the compile phase.

Author Comment

Comment Utility
i am little confused.
so basically if we dont specify any option to gcc during compile time, it will do "PADDING" for struct members for performance optimization. Correct?
LVL 86

Expert Comment

Comment Utility
Yes, that's right. IMO turning off padding nowadays only makes sense on embedded systemms with extremely low amounts of memory, and you'll hardly encounter these "in the wild" any more.
LVL 32

Expert Comment

Comment Utility
To reduce padding in large structures, cluster the largest data types first (e.g., double), followed by a cluster of the next largest data type (e.g., float or long), and so on down to char. One area where you may want to add a lot of padding is when you are working in a multithreading program and have locking and shared data variables near each other. You may want to add, say, char padding[128], between the locking variable (e.g., semaphore or mutex) and the shared data variable so that the synchronization variables are in a different cache line than the shared variable. Then, when one thread modifies the shared variable, the other threads do not have to invalidate their cache lines for the synchronization variable.

Featured Post

Enabling OSINT in Activity Based Intelligence

Activity based intelligence (ABI) requires access to all available sources of data. Recorded Future allows analysts to observe structured data on the open, deep, and dark web.

Join & Write a Comment

Have you thought about creating an iPhone application (app), but didn't even know where to get started? Here's how: ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Important pre-programming comments: I’ve never tri…
Container Orchestration platforms empower organizations to scale their apps at an exceptional rate. This is the reason numerous innovation-driven companies are moving apps to an appropriated datacenter wide platform that empowers them to scale at a …
The viewer will learn how to pass data into a function in C++. This is one step further in using functions. Instead of only printing text onto the console, the function will be able to perform calculations with argumentents given by the user.
The viewer will learn how to user default arguments when defining functions. This method of defining functions will be contrasted with the non-default-argument of defining functions.

772 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

10 Experts available now in Live!

Get 1:1 Help Now