[llvm] [RISCV] Sink vp.splat operands of VP intrinsic. (PR #133245)

[via llvm-commits]

================
@@ -2772,6 +2772,40 @@ bool RISCVTTIImpl::canSplatOperand(Instruction *I, int Operand) const {
   }
 }
 
+bool RISCVTTIImpl::tryToSinkVPSplat(VPIntrinsic *VPI,
+                                    SmallVectorImpl<Use *> &Ops) const {
+  Value *EVL = VPI->getVectorLengthParam();
+  if (!EVL)
+    return false;
+
+  for (auto &Op : VPI->operands()) {
+    auto *I = dyn_cast<Instruction>(Op.get());
+    if (!I || I->getParent() == VPI->getParent() ||
----------------
NexMing wrote:

> Why do we have the check for I->getParent() == VPI->getParent()? The non-VP code doesn't seem to need it

Sinking the splat operand within the same basic block has no effect. This optimization typically focuses on cases where the splat operand is not in the same basic block.

https://github.com/llvm/llvm-project/pull/133245