[Mlir-commits] [mlir] [mlir][vector] add tensor.concat, bitcast, expand_shape, collapse_shape vectorization support (PR #97297)

[Oleksandr Alex Zinenko]

================
@@ -1055,3 +1055,195 @@ func.func @test_vectorize_unpack_no_vector_sizes_permute(%source: tensor<4x7x4xf
     transform.yield
   } 
  }
+
+  // -----
+
+ // CHECK-LABEL: func @test_vectorize_collapseshape
+func.func @test_vectorize_collapseshape(%source: tensor<8x8x32x16xf32>, %dest: tensor<64x512xf32>) -> tensor<64x512xf32> {
+    // CHECK: %[[CST:.*]] = arith.constant 0.000000e+00 : f32
+    // CHECK: %[[C0:.*]]= arith.constant 0 : index
+    // CHECK: %[[C8:.*]] = arith.constant 8 : index
+    // CHECK: %[[C80:.*]] = arith.constant 8 : index
+    // CHECK: %[[C32:.*]] = arith.constant 32 : index
+    // CHECK: %[[C16:.*]] = arith.constant 16 : index
+    // CHECK: %[[MSK0:.*]] = vector.create_mask %[[C8]], %[[C80]], %[[C32]], %[[C16]] : vector<8x8x32x32xi1>
+    // CHECK: %[[READ0:.*]] = vector.mask %[[MSK0]] {{.*}} : vector<8x8x32x32xi1> -> vector<8x8x32x32xf32>
----------------
ftynse wrote:

This doesn't check that we actually read inside the mask. Same below.

https://github.com/llvm/llvm-project/pull/97297