This is a familiar pattern in any language, not just scala. If you doubt me just check out the source code for memcpy.c. Witness the loop unrolling, the byte fiddling the register level manipulations and the several hundreds, if not thousands, of lines of code required to implement the function. Wouldn't a simple for loop be enough?

High performance code anywhere will force you to drop down to the lowest level possible and work from there. The rationale of functional programming is that you can work largely at a high level and drop down to low level techniques in the 1-10% of the code that really matter for performance. One measure of the success of Scala is that it is being benchmarked against Java unlike Python/Ruby/Groovy etc. It is a reasonably high level language, yet the benchmark comparisons are always against Java. If necessary the performance required can be had by writing your while loops to avoid the object allocation required to build a closure. I don't see this as a disadvantage at all, as I can use the high level techniques in the remaining 90% of the code.