[PATCH] D41168: [X86] Lowering X86 avx512 sqrt intrinsics to IR

[Craig Topper via Phabricator via cfe-commits]

craig.topper added inline comments.


================
Comment at: lib/CodeGen/CGBuiltin.cpp:8904
+    unsigned CC = cast<llvm::ConstantInt>(Ops[4])->getZExtValue();
+    // Support only if the rounding mode is 4 (AKA CUR_DIRECTION), otherwise keep the intrinsic.
+    if (CC != 4) {
----------------
80 columns?


================
Comment at: lib/CodeGen/CGBuiltin.cpp:8907
+      return nullptr;
+    }
+    llvm::Value *C0 = llvm::ConstantInt::get(SizeTy, 0);
----------------
Drop the curly braces


================
Comment at: lib/CodeGen/CGBuiltin.cpp:8909
+    llvm::Value *C0 = llvm::ConstantInt::get(SizeTy, 0);
+    Value *A = Builder.CreateExtractElement(Ops[0], C0, "extract");
+    Function *F = CGM.getIntrinsic(Intrinsic::sqrt, A->getType());
----------------
There's a signature for CreateExtractElement that takes a uint64_t for the index. No need to create a ConstantInt explicitly


================
Comment at: lib/CodeGen/CGBuiltin.cpp:8930
+    // Support only if the rounding mode is 4 (AKA CUR_DIRECTION), otherwise keep the intrinsic.
+    if (CC != 4) {
+      return nullptr;
----------------
Drop the curly braces


Repository:
  rC Clang

https://reviews.llvm.org/D41168