|
From: Julian S. <js...@ac...> - 2007-02-09 17:07:52
|
> > Specialise VG_(ssort) for 4-word elements. This removes about 80% of > > all calls to VG_(memcpy). Thanks to cachegrind for showing somebody > > was calling VG_(memcpy) a huge number of times, and to callgrind for > > finding out who :-) > > Cool. > Just curious: about which code / VG tool are you talking about? VG_(ssort) (a generic sort routine) is used mostly by the debuginfo readers, to sort the line-number / symbol-table arrays by address, and so it the debuginfo reader which mostly is marginally faster. (both the line number info and symbol table info is contained in 4-word records and that is why I did the specialisation for 4-word units.) > Is the runtime improvement similar impressive? Unfortunately no. J |