On 08/27/2018 09:49 AM, Ningsheng Jian wrote: > Yes, some benchmark data is unstable. We did analyze the arm64 and > aarch64 codegen difference with microbenchmarks. AArch64 backend > generates better code on most cases, and I believe that we have > addressed those minor worse cases we found. E.g. JDK-8183533, > JDK-8185786, JDK-8187601.
Mmm, and it's not really possible to show for sure that changes generalize well. I'm still not convinced by tweaks to unrolling, but they work spectacularly well in some small benchmarks. -- Andrew Haley Java Platform Lead Engineer Red Hat UK Ltd. <https://www.redhat.com> EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671
