Tim Armstrong has posted comments on this change. ( http://gerrit.cloudera.org:8080/8802 )
Change subject: IMPALA-6291: disable AVX512 codegen in LLVM ...................................................................... Patch Set 2: (4 comments) http://gerrit.cloudera.org:8080/#/c/8802/2/be/src/codegen/llvm-codegen.cc File be/src/codegen/llvm-codegen.cc: http://gerrit.cloudera.org:8080/#/c/8802/2/be/src/codegen/llvm-codegen.cc@110 PS2, Line 110: DEFINE_string(llvm_cpu_attr_whitelist, "sse2,cx16,prfchw,bmi2,fsgsbase,popcnt,aes,smap," > And as the recommended LLVM version increases I think we should be doing that manually when we are confident that we'll be routinely testing on systems with those features. It'll be a bit annoying to have to do this but it avoids surprises. http://gerrit.cloudera.org:8080/#/c/8802/2/be/src/codegen/llvm-codegen.cc@111 PS2, Line 111: "mmx,rdseed,hle,xsave,invpcid,avx,rtm,fma,bmi,rdrnd,sse4.1,sse4.2,avx2,sse,lzcnt," I sorted these lists to make them more readable. http://gerrit.cloudera.org:8080/#/c/8802/2/be/src/codegen/llvm-codegen.cc@1663 PS2, Line 1663: result.emplace_back(attr); > nit: I think this is the same as push_back, due to reference collapsing. Do Switched to push_back() everywhere since I thing it's the same -- all the args are either const string& or string&&. The move() doesn't make sense for the const string& argument here. http://gerrit.cloudera.org:8080/#/c/8802/2/be/src/codegen/llvm-codegen.cc@1666 PS2, Line 1666: string attr_name = attr.substr(1); > nit: const string Done -- To view, visit http://gerrit.cloudera.org:8080/8802 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Ic7c3ee3e370bafc50d855113485b7e6925f7bf6a Gerrit-Change-Number: 8802 Gerrit-PatchSet: 2 Gerrit-Owner: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Impala Public Jenkins Gerrit-Reviewer: Jim Apple <jbapple-imp...@apache.org> Gerrit-Reviewer: Michael Ho <k...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Comment-Date: Fri, 08 Dec 2017 19:37:14 +0000 Gerrit-HasComments: Yes
