[Bug tree-optimization/93056] Poor codegen for heapsort in stephanov_vector benchmark

[amonakov at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93056

Alexander Monakov <amonakov at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |amonakov at gcc dot gnu.org

--- Comment #1 from Alexander Monakov <amonakov at gcc dot gnu.org> ---
The benchmark sorts a 2000-entry random array, so GCC's version runs with high
branch misprediction rate. Clang's version is if-converted, it issues one extra
load compared to gcc.

PRE makes it very difficult to if-convert this on RTL, with -fno-tree-pre we
even get nicer code but still not if-converted, so slower than Clang.