Optimization Tips - Mo’ Hustle Mo’ Problems - CppCon 2014

• Smarter C/C++ inlining with __attribute__((flatten))

// GCC
#define ALWAYS_INLINE inline __attribute__((__always_inline__))
#define NEVER_INLINE         __attribute__((__noinline__))
//#define NEVER_INLINE

Options That Control Optimization

• An interesting bechmark: always_inline performance. - SO

• GCC 4.6.2 inlining behavior

• -foptimize-sibling-calls - Optimize sibling and tail recursive calls.
• -finline-small-functions -
• -findirect-inlining
• -finline-functions - Consider all functions for inlining, even if they are not declared inline.
• -fearly-inlining - Inline functions marked by always_inline and functions whose body seems smaller than the function call overhead
• -finline-limit=n - This flag allows coarse control of this limit. n is the size of functions that can be inlined in number of pseudo instructions.
• -fwhole-program - Assume that the current compilation unit represents the whole program being compiled.

Common Function Attributes (GCC)

• __cold__ - The cold attribute on functions is used to inform the compiler that the function is unlikely to be executed.
• __hot__ - The hot attribute on a function is used to inform the compiler that the function is a hot spot of the compiled program.
• __leaf__ - Calls to external functions with this attribute must return to the current compilation unit only by return or by exception handling. For example, the sin function is a leaf function, but qsort is not.
• __pure__ - Calls to functions that have no observable effects on the state of the program other than to return a value may lend themselves to optimizations such as common subexpression elimination.