[Mlir-commits] [mlir] [mlir][Vectorizer] Added support to Vectorize tensor.unpack (PR #76087)

Balaji V. Iyer. llvmlistbot at llvm.org
Tue Feb 6 15:31:45 PST 2024


================
@@ -419,6 +419,31 @@ module attributes {transform.with_named_sequence} {
 
 // -----
 
+// CHECK-LABEL: func @test_vectorize_unpack
+func.func @test_vectorize_unpack(%0 : tensor<7x1136x16x16xf32>) -> tensor<100x18176xf32>  {
+    // CHECK %[[c0:.*]] = arith.constant 0 : index
+    // CHECK: %[[tr0:.*]] = vector.mask %[[m0:.*]] {{.*}} vector.transfer_read %{{.*}} : tensor<7x1136x16x16xf32>, vector<7x1136x16x16xf32> } : vector<7x1136x16x16xi1> -> vector<7x1136x16x16xf32>
----------------
bviyer wrote:

Yes you are right. When I implemented using `vectorSizes` this fixed itself.

https://github.com/llvm/llvm-project/pull/76087


More information about the Mlir-commits mailing list