[Mlir-commits] [mlir] [mlir][sparse] support tensor.pad on CSR tensors (PR #90687)

[Aart Bik]

================
@@ -30,16 +30,8 @@
 // Do the same run, but now with direct IR generation and VLA vectorization.
 // RUN: %if mlir_arm_sve_tests %{ %{compile_sve} | %{run_sve} | FileCheck %s %}
 
-#CCCC = #sparse_tensor.encoding<{
-  map = (d0, d1, d2, d3) -> (d0 : compressed, d1 : compressed, d2 : compressed, d3 : compressed)
-}>
-
-#CDCD = #sparse_tensor.encoding<{
-  map = (d0, d1, d2, d3) -> (d0 : compressed, d1 : dense, d2 : compressed, d3 : dense)
-}>
-
-#DCCD = #sparse_tensor.encoding<{
-  map = (d0, d1, d2, d3) -> (d0 : dense, d1 : compressed, d2 : compressed, d3 : dense)
+#CDCC_NHWC = #sparse_tensor.encoding<{
----------------
aartbik wrote:

can we also add a CHECK test

you can make very simple structural tests (not the full thing), but just to make sure the essentials are tehre?

https://github.com/llvm/llvm-project/pull/90687