It mainly depends on the compiler and the linker. For exemple most C/C++ compiler will try to inline small functions automatically when compiling with all optimisations. Most of the time this inlining is only possible when both the caller and the target function are compiled from the same source file. However there is a feature of Visual C++ 7 under windows that permits to inline functions from other .c/.cpp source files, and there is another method named Global Optimisation that permits the linker to reorganize object code to optimise register usage and calling conventions accross function calls. I'm not sure gcc provides these features are of now but the effect is often spectacular concerning the resulting speed gain (typically 10 to 15% compared to the same code compiled with VC6 or without these optimisations). Of course sometimes the optimizer guess wrong and it results in strange bugs but it only occured to me once in almost one year of daily use, and i couldn't even prove it was not MY fault :). I giess the simple answer is: yes. Hope it helps, Sebastien ----- Original Message ----- From: "Paul Davis" <paul(a)linuxaudiosystems.com> To: <linux-audio-dev(a)music.columbia.edu> Sent: Sunday, 08 December, 2002 16:48 Subject: [linux-audio-dev] code in different files affects speed?

Not on x86, at least since we have a flat address space. I don't know about other architectures, of course, but I suspect not. The exception is, of course, cache. Closer things will tend to both be in cache, or neither in cache (up to cache-lines). Probably matters more for data than code.