What is the Real Performance Impact of Virtual Functions in C ?

Virtual Functions and Performance in C : A Closer Examination

Your concern regarding the performance impact of virtual functions is a valid one. To address it, let's explore the implications of virtual functions on program efficiency.

Firstly, it's important to clarify that virtual functions are dynamically bound, meaning that their implementation is determined at runtime based on the actual object type. This adds some overhead compared to regular function calls, which are statically bound and resolved at compile time.

To quantify this overhead, let's examine some empirical data from experiments conducted on a 3GHz PowerPC CPU. In these tests, a simple 4D vector class was created with inline, virtual, and regular function definitions. By repeatedly performing operations on arrays of these vectors, the following results were obtained:

• Inline: 0.65ns per call
• Direct: 5.53ns per call
• Virtual: 13ns per call

The results indicate that virtual function calls were indeed about 20 times slower than inline calls in this particular scenario. However, it's crucial to consider the significance of this difference.

The experiments involved a massive number of function calls (12,288,000 in total), and the overhead per call was a mere 7 nanoseconds. Therefore, unless you anticipate calling virtual functions at extremely high frequencies (over ten million calls per second), the performance impact is likely to be negligible.

While virtual functions do introduce some performance overhead, it's typically insignificant in most practical scenarios. Thus, premature optimization by avoiding virtual functions solely based on performance concerns is generally not advisable.

The above is the detailed content of What is the Real Performance Impact of Virtual Functions in C ?. For more information, please follow other related articles on the PHP Chinese website!