On Fri, 2013-08-30 15:36:03 +0200, Jakub Jelinek <ja...@redhat.com> wrote: > On Fri, Aug 30, 2013 at 03:26:59PM +0200, Paolo Carlini wrote: > > On 08/30/2013 02:05 PM, Tim Shen wrote: > > >+ const _TraitsT& _M_traits; > > >+ _FlagT _M_flags; > > >+ bool _M_is_non_matching; > > >+ std::set<_CharT> _M_char_set; > > >+ std::set<pair<_StringT, _StringT>> _M_range_set; > > >+ _CharClassT _M_class_set; > > another, very general comment: now that nothing is decided ABI-wise > > for these features, let's pay attention to the layouts, let's make > > sure that the data members are ordered in the best way to minimize > > the size. For example, when I see a bool sandwiched between big > > objects something seems at least weird... Let's make measurements > > and optimize for 64-bit but let's double check on 32-bit too. > > You could use pahole utility from dwarves for this, though not sure how far > it is with support for DWARF4 and extensions recent gcc emit, so perhaps you > need -gdwarf-3 -gstrict-dwarf or similar.
But please keep in mind that size is not everything. Cachelines count,
too! If you save four or eight bytes off of a stackframe (or struct or
whatever), but sacrifice a cacheline or two to access badly layouted
variables in eg. a long-running loop, you even lost performance.
MfG, JBG
--
Jan-Benedict Glaw jbg...@lug-owl.de +49-172-7608481
Signature of: GDB has a 'break' feature; why doesn't it have 'fix' too?
the second :
signature.asc
Description: Digital signature
