At first sight there is no dependency between lines of code and the performance of that code. Just look at these two statements:
a = b + c;
a
=
b
+
c
;
The first one takes one line, the second one six lines. Nevertheless, they will end up generating exactly the same code.
However, leaving such unusual cases out of consideration, there are some relations between lines of code and the run-time efficiency of that code.
(a) Optimizing a give piece of code ofter leads to more code lines.
One of the optimization techniques is loop unrolling. This often helps to optimize code with an inner loop of a fixed number of iterations. By writing manually down a number of iterations, the loop overhead is reduced and the compiler is given more opportunities to juggle with temporary variables and registers. This clearly leads to more lines of code. Other optimization techniques also have a tendency to lead to more lines of source code.
(b) Refactoring "clumsy" code often leads to better performance and less lines of code.
In all practical cases I have seen, re-implementing a badly written piece of code in a clearer way leads to a reduction of code lines and sometimes better performance. Hence, the rule: "Optimization leads to more code lines", cannot always be applied.
So, there is no general dependency between number of code lines and performance. BUT, it can be said: Doing performance optimization on a well written piece of code often leads to more code lines.