https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127420
--- Comment #12 from H.J. Lu <hjl.tools at gmail dot com> --- (In reply to H.J. Lu from comment #11) > (In reply to Haochen Jiang from comment #10) > > After excluding the table jmps, there are nearly no occurence: > > > > samples hits rate > > jump tables 109,380,239 6,941,545 6.346 % > > everything else 26,946,316 50,767 0.188 % > > all compiler-generated 136,326,555 6,992,312 5.129 % > > > > I suppose we will see no impact on SPEC > > Please compare performance impact using -fno-plt. BTW, even without -fno-plt, 1096 UD2s are generated after sibcall in the cc1plus executable.
