Add tests for read/write operations with permutation mapping

[Copilot]

• Update tests to use 3D tensors with cyclic permutation (d0,d1,d2) -> (d1,d2,d0)
• Use non-self-inverse permutation for more robust testing
• Apply multi-line formatting to dictionary arguments for readability
• Reply to formatting comment

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[ftynse]

Nice! Can you use higher-rank tensors so that the permutation is not equal to its inverse, this makes for a more robust test...

[ftynse]

@copilot Nice! Can you use higher-rank tensors so that the permutation is not equal to its inverse, this makes for a more robust test...

[Copilot]

@copilot Nice! Can you use higher-rank tensors so that the permutation is not equal to its inverse, this makes for a more robust test...

Updated both tests to use 3D tensors with a cyclic permutation (d0, d1, d2) -> (d1, d2, d0). This is a non-self-inverse permutation that requires 3 applications to return to identity, making for a more robust test.

Changes committed in 81a10e1.

[ftynse]

@copilot the code needs a bit of reformatting, please apply the following patch:

diff --git a/lit_tests/kernel/wave/mlir_converter.py b/lit_tests/kernel/wave/mlir_converter.py
index 947d4c9..335b3fc 100644
--- a/lit_tests/kernel/wave/mlir_converter.py
+++ b/lit_tests/kernel/wave/mlir_converter.py
@@ -1111,7 +1111,11 @@ def read_with_mapping_kernel(
     k = tkw.IndexMapping.iterator(2)
     cyclic_mapping = tkw.IndexMapping(
         num_iterators=3,
-        inputs={M: k, N: i, K: j},  # Memory[M,N,K]: permutation maps (d0,d1,d2) -> (d1,d2,d0)
+        inputs={
+            M: k,
+            N: i,
+            K: j,
+        },  # Memory[M,N,K]: permutation maps (d0,d1,d2) -> (d1,d2,d0)
         outputs={N: i, K: j, M: k},  # Register[N,K,M]: N→iter(0), K→iter(1), M→iter(2)
     )
 
@@ -1188,7 +1192,11 @@ def write_with_mapping_kernel(
     cyclic_mapping = tkw.IndexMapping(
         num_iterators=3,
         inputs={N: i, K: j, M: k},  # Register[N,K,M]: N→iter(0), K→iter(1), M→iter(2)
-        outputs={M: k, N: i, K: j},  # Memory[M,N,K]: permutation maps (d0,d1,d2) -> (d1,d2,d0)
+        outputs={
+            M: k,
+            N: i,
+            K: j,
+        },  # Memory[M,N,K]: permutation maps (d0,d1,d2) -> (d1,d2,d0)
     )
 
     # Read from memory (no mapping)

[Copilot]

@copilot the code needs a bit of reformatting, please apply the following patch:

...

Formatting applied as requested in 112f5ed.