[Mlir-commits] [mlir] [mlir][vector] Add foldInsertUseChain folder function to insert op (PR #147045)

[Andrzej Warzyński]

================
@@ -3446,3 +3446,68 @@ func.func @fold_insert_constant_indices(%arg : vector<4x1xi32>) -> vector<4x1xi3
   %res = vector.insert %1, %arg[%0, %0] : i32 into vector<4x1xi32>
   return %res : vector<4x1xi32>
 }
+
+// -----
+
+// CHECK-LABEL: @fold_insert_use_chain_static_pos(
+//  CHECK-SAME:   %[[ARG:.*]]: vector<4xf32>,
+//  CHECK-SAME:   %[[VAL:.*]]: f32) -> vector<4xf32> {
+//       CHECK:   %[[RES:.*]] = vector.insert %[[VAL]], %[[ARG]] [0] : f32 into vector<4xf32>
+//       CHECK:    return %[[RES]] : vector<4xf32>
+func.func @fold_insert_use_chain_static_pos(%v : vector<4xf32>, %value : f32) -> vector<4xf32> {
+  %v_0 = vector.insert %value, %v[0] : f32 into vector<4xf32>
+  %v_1 = vector.insert %value, %v_0[0] : f32 into vector<4xf32>
+  %v_2 = vector.insert %value, %v_1[0] : f32 into vector<4xf32>
+  return %v_2 : vector<4xf32>  
+}
+
+// -----
+
+// CHECK-LABEL: @fold_insert_use_chain_dynamic_pos(
+//  CHECK-SAME:   %[[ARG:.*]]: vector<4x4xf32>,
+//  CHECK-SAME:   %[[VAL:.*]]: f32,
+//  CHECK-SAME:   %[[POS:.*]]: index) -> vector<4x4xf32> {
+//       CHECK:   %[[RES:.*]] = vector.insert %[[VAL]], %[[ARG]] {{\[}}%[[POS]], 0] : f32 into vector<4x4xf32>
+//       CHECK:   return %[[RES]] : vector<4x4xf32>
+func.func @fold_insert_use_chain_dynamic_pos(%arg : vector<4x4xf32>, %value : f32, %pos: index) -> vector<4x4xf32> {
+  %v_0 = vector.insert %value, %arg[%pos, 0] : f32 into vector<4x4xf32>
+  %v_1 = vector.insert %value, %v_0[%pos, 0] : f32 into vector<4x4xf32>
+  %v_2 = vector.insert %value, %v_1[%pos, 0] : f32 into vector<4x4xf32>
+  return %v_2 : vector<4x4xf32>  
+}
----------------
banach-space wrote:

Compared to `@fold_insert_use_chain_static_pos`, this is:
*  adding _dynamic positions_, **and** 
* switches from 1D to _2D vectors_. 

Could you split this into two functions and test one change at a time? Also, is there any point checking static positions at all?

https://github.com/llvm/llvm-project/pull/147045