It's been said before, but it merits repeating.
Developers never know where the performance bottleneck is until they run performance tests.
The same is true for comparing technique A to technique B. Unless there is a dramatic difference then you just have to test it. It might be obvious if you have an O(n) vs O(n^x) scenario, but since the LINQ stuff is mostly compiler witchcraft, it merits a profiling.
Besides, unless your project is in production and you have profiled the code and found that that loop is slowing down your execution, leave it as whichever is your preference for readability and maintenance. Premature optimization is the devil.
