[Bug middle-end/122913] [16 Regression?] pr12392.cpp -Ofast compile time went up 66% when switching to -std=c++20

[rguenth at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122913

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2026-02-06
     Ever confirmed|0                           |1

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.

But:

> ~/install/gcc-16/usr/local/bin/g++ -S t.C -std=c++20 -Ofast -fno-checking -E 
> -dD | grep _GLIBCXX_EXTERN_TEMPLATE
#define _GLIBCXX_EXTERN_TEMPLATE 1
> ~/install/gcc-16/usr/local/bin/g++ -S t.C -std=c++17 -Ofast -fno-checking -E 
> -dD | grep _GLIBCXX_EXTERN_TEMPLATE
#define _GLIBCXX_EXTERN_TEMPLATE 1

so not sure what comment#2 wants to tell us?

C++20:

   text    data     bss     dec     hex filename
 796421       0       0  796421   c2705 t.o

vs. C++17:

   text    data     bss     dec     hex filename
 571556       0       0  571556   8b8a4 t.o

so there's also 72% more code generated.

-fopt-info differs a lot, the only thing I can say that C++20 optimizes a lot
less, 16894 "optimized" vs. 21277 "optimized" using C++17.  I see substantial
less inlining happening.

> nm t.o.C++20 | wc -l
267
> nm t.o.C++17 | wc -l
264

Are there possibly new "patterns" with C++20 we've not yet seen and thus
fail to optimize?