https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66573
--- Comment #1 from Jason McG <jmcguiness at liquidcapital dot com> --- Note that clang++ for all tested versions (3.0, 3.2, 3.3, 3.4.1, 3.5, 3.5.1, 3.6 (rc2), 3.7 (experimental)) produce the same assembler output for -O1, -O2 & -O3: foo(bool): # @foo(bool) testb %dil, %dil je .LBB0_2 jmp bar1() # TAILCALL .LBB0_2: jmp bar2() # TAILCALL With bar1() being the statically-predicted branch, as per icc and Intel's own documentation (as referenced in the link in the initial comment).