[PATCH] D99750: [LV, VP] RFC: VP intrinsics support for the Loop Vectorizer (Proof-of-Concept)

[Alexey Bataev via Phabricator via llvm-commits]

ABataev added inline comments.


================
Comment at: llvm/include/llvm/IR/Intrinsics.td:1951
+  DefaultAttrsIntrinsic<[llvm_i32_ty],
+            [llvm_anyint_ty, llvm_i64_ty, llvm_i64_ty],
+            [IntrNoMem, IntrNoSync, IntrWillReturn]>;
----------------
craig.topper wrote:
> Any reason for the arguments to be i64 instead of i32?
> 
> Are they always constants?
1. Yes, let's make them i32, forgot about 32 bit targets.
2. Currently, yes.


================
Comment at: llvm/lib/Transforms/Vectorize/VPlanRecipes.cpp:1389
+    Constant *WidthArg = State.Builder.getInt64(*State.ElementWidth);
+    Constant *RWFactorArg = State.Builder.getInt64(*State.RegWidthFactor);
+
----------------
craig.topper wrote:
> What is RegWidthFactor supposed to represent?
> 
> Can we make this the VF instead?
Generally speaking, LMUL. We can pass VF here, I assume, instead, it should be more portable.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D99750/new/

https://reviews.llvm.org/D99750