C++ compile time (again)

[Matt Austern]

On Monday, June 16, 2003, at 09:40  PM, Mike Stump wrote:

> On Friday, June 13, 2003, at 08:31 AM, Steven Bosscher wrote:
>>> Why is it that everyone is always looking at C code for compile time?
>>>
>> If C speeds up, it usually helps _all_ languages, so speedin gup C is 
>> a good thing :-)
>
> The problems encountered in C++ are fairly unique to C++.  Speeding up 
> C isn't going to help much.

That's true of C++ front end problems, anyway.

As I've commented before, there's a bit of a disconnect here: when some 
people say "speed up" they usually think they're talking about front 
end speedups, when other people say "speed up" they usually think 
they're talking about back end/optimizer speedups.  Both groups tend to 
think their focus is so obvious that it's not worth mentioning.

I'm usually in the former group.  My experience is that at least at 
-O0, and sometimes even at higher optimization levels, time spent in 
the back end and assembler is completely negligible for C++.  For now, 
at least my opinion is that the only places worth looking at for -O0 
C++ speedups are the preprocessor and the C++ front end.

(And even if my opinion is wrong, at least I've clarified what I'm 
talking about!)

			--Matt