https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70048
--- Comment #18 from Jiong Wang <jiwang at gcc dot gnu.org> --- (In reply to Wilco from comment #17) > (In reply to Jiong Wang from comment #16) > I ran this modified patch through a few benchmarks and there are no > regressions. The codesize of SPEC2006 reduces significantly in several cases > (-0.25%), while there are only 2 minor increases of +0.02%, so the patch > reduces instruction counts quite effectively. For the record, I benchmarked the same modified patch on SPEC2000 INT === There are object size improvement (smaller) on near all individual benchmarks, especially, 164.gzip is -1.9% smaller and 256.bzip2 is -0.8% smaller. Only one benchmark 197.parser got very trivial object size regression which is <0.02%. Meanwhile, for runtime perf, 256.bzip2 is +1.1%, no change on others. FP === I only got object size comparision result currently (runtime perf result will be available during weekend). Like INT, there are overall object size reduce, for example, 173.applu is -1.3% smaller, and 177.mesa is -1.8% smaller.
