[PATCH] D39292: [X86][AVX512] Adding new patterns for extract_subvector of vXi1

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: lib/Target/X86/X86InstrAVX512.td:2944
+
+foreach Index = 0-1 in{
+def : Pat<(v1i1 (extract_subvector (v2i1 VK2:$src), (iPTR Index))),
----------------
Why can't we use an SDNodeXForm to rewrite the immediate instead of generating so many patterns? See getInsertVINSERTImmediate in X86ISelDAGToDAG.cpp.


================
Comment at: test/CodeGen/X86/avx512-extract-subvector-load-store.ll:2
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=skx -O2 | FileCheck %s --check-prefix=AVX512
+
----------------
I believe we're trying to avoid using -mcpu in tests. Please use -mattr


https://reviews.llvm.org/D39292