Working 2D tiling

unit_tests/recipies/Test_MatMulTiling.cpp

@@ -22,6 +22,7 @@
 #include "aidge/operator/Producer.hpp"
 #include "aidge/graph/OpArgs.hpp"
 #include "aidge/scheduler/SequentialScheduler.hpp"
+#include "aidge/graph/Matching.hpp"
 #include "aidge/utils/TensorUtils.hpp"
 
 using namespace Aidge;
@@ -79,10 +80,22 @@ TEST_CASE("[MatMulTiling]") {
     // Tiling
     fmt::println("Tiling");
     matMulTiling(matmul1, {16, 16});
+    removeIdentity(g1);
 
     g1->setBackend("cpu");
     g1->save("MatMulSplitting_graph_split");
 
+    auto gm = SinglePassGraphMatching(g1);
+    gm.addNodeLambda("16x16", [](const NodePtr& node) {
+        const auto op =
+            std::static_pointer_cast<OperatorTensor>(node->getOperator());
+        const auto dims = op->getOutput(0)->dims();
+        return (dims.end()[-2] == 16 && dims.end()[-1] == 16);
+    });
+
+    const auto results = gm.match("MatMul[16x16]");
+    REQUIRE(results.size() == 25);
+
     // Check result
     fmt::println("Schedule forward tiled graph");
     s1 = SequentialScheduler(g1);