Overwriting vs allocation/deallocation - efficiency

Question

I am writing a C++ application which will require a block of memory (about 1000 bytes) as a temporary buffer for some text processing. The operation may be repeated up to 10,000 times a second.

Could anybody confirm that it would be more expensive to allocate the memory each time I need the buffer (i.e. new with a smart pointer, memory is deallocated when out of scope), than to have a fixed buffer and clear it (write every byte of it with a zero) every time the processing is complete?

It sounds like common sense for C++ but I just cannot find anything on the internet which confirms it.

Is the situation different for computer languages with automatic garbage collection facilities (e.g. Java, .net)?

Answer 1

It's probably more expensive to allocate and free the memory each time you need it, but the bigger question is: does it matter? Write it the easiest way you know how that works correctly and doesn't leak/corrupt memory. Only then if your performance isn't good enough, profile the code to see where you can improve it.