We maintain a list of issues in critical path:
functionality:
#I3ONI7:[mplfeC] 前端关于支持neon instrinisics编译调研和相关问题跟踪
#I3TJV5:dpdk中的int128类型使用及inline asemmbly使用
optimization:
Register allocation issues especially the localra related ones are top 1 during our CPU2017 performance diagnose. Will update details soon.
2021/9/22
#I4BL1L:restrict pointer effect may improve 557.xz
2021/8/16
#I45PLQ:spec-500-502-ref性能问题整理&&Huawei后续优化计划
Summary of SPEC2017 performance diagnose and future plan
#I43ZJ5:【spec性能分析】store merging
#I43ZO4:【spec性能分析】特别场景下div/mul指令优化
#I3TXT7:寄存器压力太大时,lra spill过多,导致性能不好 (closed)
2021/06/03 update
#I3U16H:【spec性能分析】iv冗余的swtx
2021/06/04 update
#I3UAYZ:【RA LOOP】大函数里,循环内一些实际执行频率高的变量,没有被分配到寄存器。 (major)
2021/06/23 update
#I3XAX3:RA里的caller save 操作插入最小化问题。 (major)
2021/06/24 update
#I3SPQA:【spec性能分析】ivopt优化(major)
I have added support for inline assembly to clang2mpl in the branch, "inline-asm". It cannot be merged into master until the backend supports the new AsmNode, since some of the ctorture tests use inline assembly and will fail with this change. Let me know if have any feedback.
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。
登录 后才可以发表评论