[llvm] [LV][EVL] Introduce MergeUntilPivot VPInstruction to enable out-loop reduction in EVL vectorization. (PR #101641)

[Elvis Wang via llvm-commits]

================
@@ -1125,6 +1125,8 @@ RISCVTTIImpl::getIntrinsicInstrCost(const IntrinsicCostAttributes &ICA,
         *FOp, ICA.getArgTypes()[0], UI->getPointerAlignment(),
         UI->getOperand(1)->getType()->getPointerAddressSpace(), CostKind);
   }
+  case Intrinsic::vp_merge:
+    return 1;
----------------
ElvisWang123 wrote:

PR #112327 landed. We can get the instruction cost for `vp_merge` from `getintrinsicInstrCost()` in RISCVTTI.

https://github.com/llvm/llvm-project/pull/101641