[llvm] [RISCV][TTI] Model the cost of insert/extractelt when the vector split into multiple register group and idx exceed single group. (PR #118401)

[Philip Reames via llvm-commits]

================
@@ -926,10 +926,10 @@ define void @extractelement_int_lmul(i32 %x) {
 ; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i32_15 = extractelement <32 x i32> undef, i32 15
 ; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i32_31 = extractelement <32 x i32> undef, i32 31
 ; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i32_63 = extractelement <64 x i32> undef, i32 63
-; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v128i8 = extractelement <128 x i8> undef, i32 %x
-; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v256i8 = extractelement <256 x i8> undef, i32 %x
-; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v32i32 = extractelement <32 x i32> undef, i32 %x
-; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v64i32 = extractelement <64 x i32> undef, i32 %x
+; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v128i8 = extractelement <128 x i8> undef, i32 %x
+; RV64ZVE64X-NEXT:  Cost Model: Found an estimated cost of 129 for instruction: %v256i8 = extractelement <256 x i8> undef, i32 %x
----------------
preames wrote:

This cost seems way too big.

256 x i8 on V is 16 registers worth of data.  Storing that should be ~16 in cost.  A single scalar load and addressing is ~2-3.  So, I'd expect something in the order of ~20, not ~129 here.  

https://github.com/llvm/llvm-project/pull/118401