When profiling my application I see references to the function __intel_fast_memcpy. Is this function users accessible? Does it provide superior performance to just calling the compiler memcpy() function?
This function is not accessable from user. It's provided by the Intel compiler for better performance. When the compiler sees "memcpy()", it translates to this function.
So you should be able to use nm or dumpbin to verify that memcpy() has in fact been replaced by the intel optimized function.It is intended to recognize cases which benefit from code for specific Intel CPUs. Needless to say, you might be interested in profiling to compare performance with alternate libraries on your application.