[Mlir-commits] [mlir] [mlir][x86vector] AVX10 I8 Dot Op (PR #178807)

[Adam Siemieniuk]

================
@@ -614,40 +683,6 @@ module attributes {transform.with_named_sequence} {
 
 // -----
 
-!vecA = vector<1x1x1x4xi8>
-!vecB = vector<1x1x16x4xi8>
-!vecC = vector<1x1x16xi32>
-#map = affine_map<(d0, d4, d1, d2, d3) -> (d0, d1, d3, d4)>
-#map1 = affine_map<(d0, d4, d1, d2, d3) -> (d0, d3, d2, d4)>
-#map2 = affine_map<(d0, d4, d1, d2, d3) -> (d0, d1, d2)>
-func.func @negative_wrong_vector_shape_int8(
----------------
adam-smnk wrote:

Let's keep the negative test. Just please update the shape to an invalid one.

https://github.com/llvm/llvm-project/pull/178807