diff --git a/tests/datasets/LargeBatchMatMulDataset.h b/tests/datasets/LargeMatMulDataset.h
similarity index 82%
rename from tests/datasets/LargeBatchMatMulDataset.h
rename to tests/datasets/LargeMatMulDataset.h
index 0d8ff91..cbc97d5 100644
--- a/tests/datasets/LargeBatchMatMulDataset.h
+++ b/tests/datasets/LargeMatMulDataset.h
@@ -21,12 +21,12 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ACL_TESTS_DATASETS_LARGEBATCHMATMULDATASET
-#define ACL_TESTS_DATASETS_LARGEBATCHMATMULDATASET
+#ifndef ACL_TESTS_DATASETS_LARGEMATMULDATASET
+#define ACL_TESTS_DATASETS_LARGEMATMULDATASET
 
 #include "arm_compute/core/TensorShape.h"
 #include "arm_compute/core/Types.h"
-#include "tests/datasets/BatchMatMulDataset.h"
+#include "tests/datasets/MatMulDataset.h"
 
 namespace arm_compute
 {
@@ -34,10 +34,10 @@
 {
 namespace datasets
 {
-class LargeBatchMatMulDataset final : public BatchMatMulDataset
+class LargeMatMulDataset final : public MatMulDataset
 {
 public:
-    LargeBatchMatMulDataset()
+    LargeMatMulDataset()
     {
         add_config(TensorShape(21U, 13U, 3U, 2U), TensorShape(33U, 21U, 3U, 2U), TensorShape(33U, 13U, 3U, 2U));
         add_config(TensorShape(38U, 12U, 1U, 5U), TensorShape(21U, 38U, 1U, 5U), TensorShape(21U, 12U, 1U, 5U));
@@ -45,10 +45,10 @@
     }
 };
 
-class HighDimensionalBatchMatMulDataset final : public BatchMatMulDataset
+class HighDimensionalMatMulDataset final : public MatMulDataset
 {
 public:
-    HighDimensionalBatchMatMulDataset()
+    HighDimensionalMatMulDataset()
     {
         add_config(TensorShape(5U, 5U, 2U, 2U, 2U, 2U), TensorShape(5U, 5U, 2U, 2U, 2U, 2U), TensorShape(5U, 5U, 2U, 2U, 2U, 2U)); // 6D tensor
     }
@@ -57,4 +57,4 @@
 } // namespace datasets
 } // namespace test
 } // namespace arm_compute
-#endif /* ACL_TESTS_DATASETS_LARGEBATCHMATMULDATASET */
+#endif /* ACL_TESTS_DATASETS_LARGEMATMULDATASET */
diff --git a/tests/datasets/BatchMatMulDataset.h b/tests/datasets/MatMulDataset.h
similarity index 90%
rename from tests/datasets/BatchMatMulDataset.h
rename to tests/datasets/MatMulDataset.h
index dad7cc0..9c1c5fb 100644
--- a/tests/datasets/BatchMatMulDataset.h
+++ b/tests/datasets/MatMulDataset.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef TESTS_DATASETS_BATCHMATMULDATASET
-#define TESTS_DATASETS_BATCHMATMULDATASET
+#ifndef ACL_TESTS_DATASETS_MATMULDATASET
+#define ACL_TESTS_DATASETS_MATMULDATASET
 
 #include "arm_compute/core/TensorShape.h"
 #include "utils/TypePrinter.h"
@@ -33,7 +33,7 @@
 {
 namespace datasets
 {
-class BatchMatMulDataset
+class MatMulDataset
 {
 public:
     using type = std::tuple<TensorShape, TensorShape, TensorShape>;
@@ -58,7 +58,7 @@
             return description.str();
         }
 
-        BatchMatMulDataset::type operator*() const
+        MatMulDataset::type operator*() const
         {
             return std::make_tuple(*_a_it, *_b_it, *_dst_it);
         }
@@ -96,8 +96,8 @@
     }
 
 protected:
-    BatchMatMulDataset()                      = default;
-    BatchMatMulDataset(BatchMatMulDataset &&) = default;
+    MatMulDataset()                 = default;
+    MatMulDataset(MatMulDataset &&) = default;
 
 private:
     std::vector<TensorShape> _a_shapes{};
@@ -107,4 +107,4 @@
 } // namespace datasets
 } // namespace test
 } // namespace arm_compute
-#endif /* TESTS_DATASETS_BATCHMATMULDATASET */
+#endif /* ACL_TESTS_DATASETS_MATMULDATASET */
diff --git a/tests/datasets/SmallBatchMatMulDataset.h b/tests/datasets/SmallMatMulDataset.h
similarity index 78%
rename from tests/datasets/SmallBatchMatMulDataset.h
rename to tests/datasets/SmallMatMulDataset.h
index cfe76be..ae92b9a 100644
--- a/tests/datasets/SmallBatchMatMulDataset.h
+++ b/tests/datasets/SmallMatMulDataset.h
@@ -21,12 +21,12 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ACL_TESTS_DATASETS_SMALLBATCHMATMULDATASET
-#define ACL_TESTS_DATASETS_SMALLBATCHMATMULDATASET
+#ifndef ACL_TESTS_DATASETS_SMALLMATMULDATASET
+#define ACL_TESTS_DATASETS_SMALLMATMULDATASET
 
 #include "arm_compute/core/TensorShape.h"
 #include "arm_compute/core/Types.h"
-#include "tests/datasets/BatchMatMulDataset.h"
+#include "tests/datasets/MatMulDataset.h"
 
 namespace arm_compute
 {
@@ -34,10 +34,10 @@
 {
 namespace datasets
 {
-class SmallBatchMatMulDataset final : public BatchMatMulDataset
+class SmallMatMulDataset final : public MatMulDataset
 {
 public:
-    SmallBatchMatMulDataset()
+    SmallMatMulDataset()
     {
         add_config(TensorShape(3U, 4U, 2U, 2U), TensorShape(2U, 3U, 2U, 2U), TensorShape(2U, 4U, 2U, 2U));
         add_config(TensorShape(9U, 6U), TensorShape(5U, 9U), TensorShape(5U, 6U));
@@ -46,7 +46,18 @@
         add_config(TensorShape(32U, 2U), TensorShape(17U, 32U), TensorShape(17U, 2U));
     }
 };
+
+class TinyMatMulDataset final : public MatMulDataset
+{
+public:
+    TinyMatMulDataset()
+    {
+        add_config(TensorShape(1U), TensorShape(1U), TensorShape(1U));
+        add_config(TensorShape(2U, 2U), TensorShape(2U, 2U), TensorShape(2U, 2U));
+    }
+};
+
 } // namespace datasets
 } // namespace test
 } // namespace arm_compute
-#endif /* ACL_TESTS_DATASETS_SMALLBATCHMATMULDATASET */
+#endif /* ACL_TESTS_DATASETS_SMALLMATMULDATASET */
diff --git a/tests/validation/CL/BatchMatMul.cpp b/tests/validation/CL/BatchMatMul.cpp
deleted file mode 100644
index fd84526..0000000
--- a/tests/validation/CL/BatchMatMul.cpp
+++ /dev/null
@@ -1,239 +0,0 @@
-/*
- * Copyright (c) 2023 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-
-#include "arm_compute/runtime/CL/CLTensor.h"
-#include "src/gpu/cl/kernels/ClNativeMatMulKernel.h"
-#include "tests/datasets/LargeBatchMatMulDataset.h"
-#include "tests/datasets/SmallBatchMatMulDataset.h"
-#include "tests/framework/Macros.h"
-#include "tests/framework/datasets/Datasets.h"
-#include "tests/validation/Validation.h"
-#include "tests/validation/fixtures/BatchMatMulFixture.h"
-
-namespace arm_compute
-{
-namespace test
-{
-namespace validation
-{
-namespace
-{
-RelativeTolerance<float> tolerance_f32(0.001f); /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
-constexpr float          abs_tolerance_f32(
-    0.0001f); /**< Absolute tolerance value for comparing reference's output against implementation's output for floating point data types in case using relative tolerance fails because of small values */
-constexpr float abs_tolerance_f16(
-    0.001f);                                                   /**< Absolute tolerance value for comparing reference's output against implementation's output for fp16  data types in case using relative tolerance fails because of small values */
-RelativeTolerance<half_float::half> tolerance_f16(half(0.01)); /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
-} // namespace
-
-/** M0 values to test --precommit*/
-const auto m0_values_precommit = framework::dataset::make("M0", { 1, 3 });
-
-/** N0 values to test --precommit*/
-const auto n0_values_precommit = framework::dataset::make("N0", { 2, 4 });
-
-/** K0 values to test --precommit*/
-const auto k0_values_precommit = framework::dataset::make("K0", { 2, 3 });
-
-/** M0 values to test --nightly*/
-const auto m0_values_nightly_lhs_nt = framework::dataset::make("M0", { 1, 2, 3, 4, 5, 6, 7, 8 });
-// const auto m0_values_nightly_lhs_t = framework::dataset::make("M0", { 1, 2, 3, 4, 8 }); // To be enabled
-
-/** N0 values to test --nightly*/
-const auto n0_values_nightly_rhs_nt = framework::dataset::make("N0", { 1, 2, 3, 4, 8, 16 });
-const auto n0_values_nightly_rhs_t  = framework::dataset::make("N0", { 1, 2, 3, 4, 8 });
-
-/** K0 values to test --nightly*/
-const auto k0_values_nightly_lhs_nt_rhs_nt = framework::dataset::make("K0", { 1, 2, 3, 4, 8, 16 });
-const auto k0_values_nightly_lhs_nt_rhs_t  = framework::dataset::make("K0", { 1, 2, 3, 4, 8 });
-// const auto k0_values_nightly_lhs_t_rhs_nt = framework::dataset::make("K0", { 1, 2, 3, 4, 5, 6, 7, 8 }); // To be enabled
-
-template <typename T>
-using CLBatchMatMulFixture = BatchMatMulValidationFixture<T>;
-
-TEST_SUITE(CL)
-TEST_SUITE(BatchMatMul)
-DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(zip(
-                                                                      framework::dataset::make("LhsInfo",
-{
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::S32), // Unsupported data type
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-    TensorInfo(TensorShape(27U, 13U), 1, DataType::F32),
-}),
-framework::dataset::make("RhsInfo",
-{
-    TensorInfo(TensorShape(8U, 27U), 1, DataType::S32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32), TensorInfo(TensorShape(8U, 27U), 1, DataType::F32),
-})),
-framework::dataset::make("OutputInfo",
-{
-    TensorInfo(TensorShape(8U, 13U), 1, DataType::S32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32), TensorInfo(TensorShape(8U, 13U), 1, DataType::F32),
-})),
-framework::dataset::make("MatMulInfo",
-{
-    MatMulKernelInfo(false, false, 2, 2, 2, false), MatMulKernelInfo(false, false, 2, 2, 2, false), MatMulKernelInfo(false, false, 9, 2, 2, false), MatMulKernelInfo(false, false, 0, 2, 2, false), // M0 cannot be < 1
-    MatMulKernelInfo(false, true, 4, 5, 2, false),                                                                                                                                                  // For LHS NT RHS NT: N0 cannot be 5
-    MatMulKernelInfo(false, true, 4, 6, 2, false),                                                                                                                                                  // For LHS NT RHS NT: N0 cannot be 6
-    MatMulKernelInfo(false, true, 4, 9, 2, false),                                                                                                                                                  // For LHS NT RHS NT: N0 cannot be 9
-    MatMulKernelInfo(false, true, 4, 10, 2, false),                                                                                                                                                 // For LHS NT RHS NT: N0 cannot be 10
-    MatMulKernelInfo(false, true, 4, 11, 2, false),                                                                                                                                                 // For LHS NT RHS NT: N0 cannot be 11
-    MatMulKernelInfo(false, true, 4, 17, 2, false),                                                                                                                                                 // For LHS NT RHS NT: N0 cannot be 17
-})),
-framework::dataset::make("Expected", { false, true, true, false, false, false, false, false, false, false })),
-lhs_info, rhs_info, output_info, matmul_info, expected)
-{
-    bool is_valid = bool(ClNativeMatMulKernel::validate(&lhs_info, &rhs_info, &output_info, matmul_info));
-    ARM_COMPUTE_EXPECT(is_valid == expected, framework::LogLevel::ERRORS);
-}
-TEST_SUITE(Float)
-TEST_SUITE(FP32)
-FIXTURE_DATA_TEST_CASE(RunSmallNoTranspose, CLBatchMatMulFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallBatchMatMulDataset(),
-                                                                                                                      framework::dataset::make("pretransose_A", { false })),
-                                                                                                                      framework::dataset::make("pretransose_B", { false })),
-                                                                                                                      m0_values_precommit),
-                                                                                                                      n0_values_precommit),
-                                                                                                                      k0_values_precommit),
-                                                                                                              framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-FIXTURE_DATA_TEST_CASE(RunSmallRhsTransposed, CLBatchMatMulFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallBatchMatMulDataset(),
-                                                                                                                        framework::dataset::make("pretransose_A", { false })),
-                                                                                                                        framework::dataset::make("pretransose_B", { true })),
-                                                                                                                        m0_values_precommit),
-                                                                                                                        n0_values_precommit),
-                                                                                                                        k0_values_precommit),
-                                                                                                                framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-FIXTURE_DATA_TEST_CASE(RunLargeNoTranspose, CLBatchMatMulFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeBatchMatMulDataset(),
-                                                                                                                  framework::dataset::make("pretransose_A", { false })),
-                                                                                                                  framework::dataset::make("pretransose_B", { false })),
-                                                                                                                  m0_values_nightly_lhs_nt),
-                                                                                                                  n0_values_nightly_rhs_nt),
-                                                                                                                  k0_values_nightly_lhs_nt_rhs_nt),
-                                                                                                                  framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-// Running High Dimensional test is enough for FP32, because we're stressing the number of dimensions, not data type or M0/N0/K0
-// It's a good idea to test for each Lhs/Rhs T/NT combinations because they're different CL kernels
-FIXTURE_DATA_TEST_CASE(RunHighDimNoTranspose, CLBatchMatMulFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::HighDimensionalBatchMatMulDataset(),
-                                                                                                                        framework::dataset::make("pretransose_A", { false })),
-                                                                                                                        framework::dataset::make("pretransose_B", { false })),
-                                                                                                                        framework::dataset::make("M0", { 2 })),
-                                                                                                                        framework::dataset::make("N0", { 2 })),
-                                                                                                                        framework::dataset::make("K0", { 2 })),
-                                                                                                                framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-FIXTURE_DATA_TEST_CASE(RunLargeRhsTransposed, CLBatchMatMulFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeBatchMatMulDataset(),
-                                                                                                                    framework::dataset::make("pretransose_A", { false })),
-                                                                                                                    framework::dataset::make("pretransose_B", { true })),
-                                                                                                                    m0_values_nightly_lhs_nt),
-                                                                                                                    n0_values_nightly_rhs_t),
-                                                                                                                    k0_values_nightly_lhs_nt_rhs_t),
-                                                                                                                    framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-FIXTURE_DATA_TEST_CASE(RunHighDimRhsTransposed, CLBatchMatMulFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::HighDimensionalBatchMatMulDataset(),
-                                                                                                                  framework::dataset::make("pretransose_A", { false })),
-                                                                                                                  framework::dataset::make("pretransose_B", { true })),
-                                                                                                                  framework::dataset::make("M0", { 2 })),
-                                                                                                                  framework::dataset::make("N0", { 2 })),
-                                                                                                                  framework::dataset::make("K0", { 2 })),
-                                                                                                                  framework::dataset::make("DataType", DataType::F32)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
-}
-TEST_SUITE_END() // FP32
-
-TEST_SUITE(FP16)
-FIXTURE_DATA_TEST_CASE(RunSmallNoTranspose, CLBatchMatMulFixture<half>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallBatchMatMulDataset(),
-                                                                                                                     framework::dataset::make("pretransose_A", { false })),
-                                                                                                                     framework::dataset::make("pretransose_B", { false })),
-                                                                                                                     m0_values_precommit),
-                                                                                                                     n0_values_precommit),
-                                                                                                                     k0_values_precommit),
-                                                                                                             framework::dataset::make("DataType", DataType::F16)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
-}
-FIXTURE_DATA_TEST_CASE(RunSmallRhsTransposed, CLBatchMatMulFixture<half>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallBatchMatMulDataset(),
-                                                                                                                       framework::dataset::make("pretransose_A", { false })),
-                                                                                                                       framework::dataset::make("pretransose_B", { true })),
-                                                                                                                       m0_values_precommit),
-                                                                                                                       n0_values_precommit),
-                                                                                                                       k0_values_precommit),
-                                                                                                               framework::dataset::make("DataType", DataType::F16)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
-}
-FIXTURE_DATA_TEST_CASE(RunLargeNoTranspose, CLBatchMatMulFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeBatchMatMulDataset(),
-                                                                                                                 framework::dataset::make("pretransose_A", { false })),
-                                                                                                                 framework::dataset::make("pretransose_B", { false })),
-                                                                                                                 m0_values_nightly_lhs_nt),
-                                                                                                                 n0_values_nightly_rhs_nt),
-                                                                                                                 k0_values_nightly_lhs_nt_rhs_nt),
-                                                                                                                 framework::dataset::make("DataType", DataType::F16)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
-}
-FIXTURE_DATA_TEST_CASE(RunLargeRhsTransposed, CLBatchMatMulFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeBatchMatMulDataset(),
-                                                                                                                   framework::dataset::make("pretransose_A", { false })),
-                                                                                                                   framework::dataset::make("pretransose_B", { true })),
-                                                                                                                   m0_values_nightly_lhs_nt),
-                                                                                                                   n0_values_nightly_rhs_t),
-                                                                                                                   k0_values_nightly_lhs_nt_rhs_t),
-                                                                                                                   framework::dataset::make("DataType", DataType::F16)))
-{
-    // Validate output
-    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
-}
-TEST_SUITE_END() // FP16
-
-TEST_SUITE_END() // Float
-TEST_SUITE_END() // BatchMatMul
-TEST_SUITE_END() // CL
-} // namespace validation
-} // namespace test
-} // namespace arm_compute
diff --git a/tests/validation/CL/MatMulKernel.cpp b/tests/validation/CL/MatMulKernel.cpp
new file mode 100644
index 0000000..5d2e59a
--- /dev/null
+++ b/tests/validation/CL/MatMulKernel.cpp
@@ -0,0 +1,391 @@
+/*
+ * Copyright (c) 2023 Arm Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "src/gpu/cl/kernels/ClNativeMatMulKernel.h"
+#include "tests/datasets/LargeMatMulDataset.h"
+#include "tests/datasets/SmallMatMulDataset.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/MatMulKernelFixture.h"
+#include "tests/validation/reference/Permute.h"
+
+#include <tuple>
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+RelativeTolerance<float> tolerance_f32(0.001f); /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
+constexpr float          abs_tolerance_f32(
+    0.0001f); /**< Absolute tolerance value for comparing reference's output against implementation's output for floating point data types in case using relative tolerance fails because of small values */
+constexpr float abs_tolerance_f16(
+    0.001f);                                                   /**< Absolute tolerance value for comparing reference's output against implementation's output for fp16  data types in case using relative tolerance fails because of small values */
+RelativeTolerance<half_float::half> tolerance_f16(half(0.01)); /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
+} // namespace
+
+/** M0 values to test --precommit*/
+const auto m0_values_precommit = framework::dataset::make("M0", { 1, 3 });
+
+/** N0 values to test --precommit*/
+const auto n0_values_precommit = framework::dataset::make("N0", { 2, 4 });
+
+/** K0 values to test --precommit*/
+const auto k0_values_precommit = framework::dataset::make("K0", { 2, 3 });
+
+/** M0 values to test --nightly*/
+const auto m0_values_nightly_lhs_nt = framework::dataset::make("M0", { 1, 2, 3, 4, 5, 6, 7, 8 });
+const auto m0_values_nightly_lhs_t  = framework::dataset::make("M0", { 1, 2, 3, 4, 8 });
+
+/** N0 values to test --nightly*/
+const auto n0_values_nightly_rhs_nt = framework::dataset::make("N0", { 1, 2, 3, 4, 8, 16 });
+const auto n0_values_nightly_rhs_t  = framework::dataset::make("N0", { 1, 2, 3, 4, 8 });
+
+/** K0 values to test --nightly*/
+const auto k0_values_nightly_lhs_nt_rhs_nt = framework::dataset::make("K0", { 1, 2, 3, 4, 8, 16 });
+const auto k0_values_nightly_rhs_t         = framework::dataset::make("K0", { 1, 2, 3, 4, 8 });
+const auto k0_values_nightly_lhs_t_rhs_nt  = framework::dataset::make("K0", { 1, 2, 3, 4, 5, 6, 7, 8 });
+
+template <typename T>
+using CLMatMulKernelFixture = MatMulKernelValidationFixture<T>;
+
+TEST_SUITE(CL)
+TEST_SUITE(MatMulKernel)
+TEST_SUITE(Validate)
+
+TEST_CASE(SupportedBlockSizes, framework::DatasetMode::ALL)
+{
+    using MatMulConfigurationPair = std::pair<MatMulKernelInfo, bool>;
+
+    const std::vector<MatMulConfigurationPair> supported_block_sizes =
+    {
+        // MatMulKernelInfo(adj_lhs, adj_rhs, M0, N0, K0, export_rhs_to_cl_image = false)
+        // Lhs not-transposed, Rhs-not-transposed
+        { MatMulKernelInfo(false, false, 0, 1, 1), false },  // M0 should be > 0
+        { MatMulKernelInfo(false, false, 3, 5, 1), false },  // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, false, 3, 6, 1), false },  // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, false, 3, 3, 17), false }, // K0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, false, 3, 3, 7), false },  // K0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, false, 9, 1, 2), true },
+        { MatMulKernelInfo(false, false, 3, 16, 3), true },
+        { MatMulKernelInfo(false, false, 7, 3, 4), true },
+
+        // Lhs not-transposed, Rhs transposed
+        { MatMulKernelInfo(false, true, 0, 1, 1), false },  // M0 should be > 0
+        { MatMulKernelInfo(false, true, 3, 11, 1), false }, // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, true, 3, 7, 1), false },  // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, true, 3, 3, 12), false }, // K0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, true, 3, 3, 6), false },  // K0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(false, true, 5, 1, 2), true },
+        { MatMulKernelInfo(false, true, 3, 3, 3), true },
+        { MatMulKernelInfo(false, true, 2, 4, 8), true },
+
+        // // Lhs transposed, Rhs-not-transposed
+        { MatMulKernelInfo(true, false, 1, 1, 0), false },  // K0 should be > 0
+        { MatMulKernelInfo(true, false, 3, 11, 1), false }, // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, false, 3, 7, 1), false },  // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, false, 6, 3, 12), false }, // M0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, false, 5, 3, 6), false },  // M0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, false, 4, 1, 22), true },
+        { MatMulKernelInfo(true, false, 3, 3, 3), true },
+        { MatMulKernelInfo(true, false, 2, 4, 8), true },
+
+        // // Lhs transposed, Rhs-transposed
+        { MatMulKernelInfo(true, true, 2, 1, 5), false },  // K0 should in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 1, 8, 7), false },  // K0 should in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 3, 11, 1), false }, // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 3, 7, 1), false },  // N0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 6, 3, 12), false }, // M0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 5, 3, 6), false },  // M0 not in {1, 2, 3, 4, 8, 16}
+        { MatMulKernelInfo(true, true, 4, 8, 16), true },
+        { MatMulKernelInfo(true, true, 3, 3, 4), true },
+        { MatMulKernelInfo(true, true, 16, 4, 8), true },
+    };
+
+    // Set big enough shapes so that block sizes are not truncated. Also, set all dimensions equal
+    // so that it doesn't fail for different NT/T configurations. We aim to test the block sizes here,
+    // not the shapes themselves.
+    const TensorInfo lhs_info = TensorInfo(TensorShape(100U, 100U), 1, DataType::F32);
+    const TensorInfo rhs_info = TensorInfo(TensorShape(100U, 100U), 1, DataType::F32);
+
+    for(auto &pair : supported_block_sizes)
+    {
+        TensorInfo output_info;
+        Status     status = ClNativeMatMulKernel::validate(&lhs_info, &rhs_info, &output_info, pair.first);
+
+        ARM_COMPUTE_EXPECT(bool(status) == pair.second, framework::LogLevel::ERRORS);
+    }
+}
+
+TEST_CASE(ValidateInputShapes, framework::DatasetMode::ALL)
+{
+    // Configurations are assumed to be Nt/Nt, but will be transposed inside the test to test other configurations
+    using ShapeConfigurationTuple = std::tuple<TensorShape, TensorShape, bool>;
+    const std::vector<ShapeConfigurationTuple> shape_configurations =
+    {
+        { TensorShape(5U, 1U), TensorShape(3U, 5U), true },
+        { TensorShape(10U, 12U), TensorShape(3U, 10U), true },
+        { TensorShape(8U, 4U), TensorShape(2U, 8U), true },
+        { TensorShape(8U, 4U), TensorShape(2U, 5U), false }, // Mismatch in the K dimension
+        { TensorShape(5U, 0U), TensorShape(2U, 5U), false }, // Invalid dimension
+        { TensorShape(5U, 4U, 3U, 4U, 5U, 6U), TensorShape(2U, 5U, 3U, 4U, 5U, 6U), true },
+        { TensorShape(5U, 4U, 3U, 4U, 5U, 1U), TensorShape(2U, 5U, 3U, 4U, 5U, 6U), false }, // no batch broadcasting
+        { TensorShape(5U, 4U, 3U, 4U, 9U, 6U), TensorShape(2U, 5U, 3U, 4U, 5U, 6U), false }, // mismatch in batch dimension
+    };
+
+    for(auto &tuple : shape_configurations)
+    {
+        const bool expected = std::get<2>(tuple);
+
+        for(bool adj_lhs :
+            {
+                false, true
+            })
+        {
+            for(bool adj_rhs :
+                {
+                    false, true
+                })
+            {
+                TensorShape lhs_shape = std::get<0>(tuple);
+                TensorShape rhs_shape = std::get<1>(tuple);
+
+                if(adj_lhs)
+                {
+                    permute(lhs_shape, PermutationVector(1U, 0U));
+                }
+
+                if(adj_rhs)
+                {
+                    permute(rhs_shape, PermutationVector(1U, 0U));
+                }
+
+                const TensorInfo lhs_info = TensorInfo(lhs_shape, 1, DataType::F32);
+                const TensorInfo rhs_info = TensorInfo(rhs_shape, 1, DataType::F32);
+                TensorInfo       output_info;
+
+                MatMulKernelInfo matmul_kernel_info{ adj_lhs, adj_rhs, 1, 1, 1, false /* export_rhs_to_cl_image */ };
+
+                Status status = ClNativeMatMulKernel::validate(&lhs_info, &rhs_info, &output_info, matmul_kernel_info);
+                ARM_COMPUTE_EXPECT(bool(status) == expected, framework::LogLevel::ERRORS);
+            }
+        }
+    }
+}
+
+TEST_CASE(ValidateDataTypes, framework::DatasetMode::ALL)
+{
+    // Configurations are assumed to be Nt/Nt, but will be transposed inside the test to test other configurations
+    using DataTypeConfigurationTuple = std::tuple<DataType, DataType, DataType, bool>;
+    const std::vector<DataTypeConfigurationTuple> data_type_configurations =
+    {
+        { DataType::F32, DataType::F32, DataType::F32, true },
+        { DataType::F16, DataType::F16, DataType::F16, true },
+        { DataType::F16, DataType::F32, DataType::F32, false },                                              // no mixed precision
+        { DataType::F64, DataType::F64, DataType::F64, false },                                              // no double precision
+        { DataType::QASYMM8, DataType::QASYMM8, DataType::QASYMM8, false },                                  // no quantized types
+        { DataType::QASYMM8_SIGNED, DataType::QASYMM8_SIGNED, DataType::QASYMM8_SIGNED, false },             // no quantized types
+        { DataType::QSYMM8_PER_CHANNEL, DataType::QSYMM8_PER_CHANNEL, DataType::QSYMM8_PER_CHANNEL, false }, // no quantized types
+        { DataType::QASYMM16, DataType::QASYMM16, DataType::QASYMM16, false },                               // no quantized types
+        { DataType::QSYMM16, DataType::QSYMM16, DataType::QSYMM16, false },                                  // no quantized types
+        { DataType::QSYMM8, DataType::QSYMM8, DataType::QSYMM8, false },                                     // no quantized types
+        { DataType::S64, DataType::S64, DataType::S64, false },                                              // no integral types
+        { DataType::S32, DataType::S32, DataType::S32, false },                                              // no integral types
+        { DataType::S16, DataType::S16, DataType::S16, false },                                              // no integral types
+        { DataType::S8, DataType::S8, DataType::S8, false },                                                 // no integral types
+        { DataType::U64, DataType::U64, DataType::U64, false },                                              // no integral types
+        { DataType::U32, DataType::U32, DataType::U32, false },                                              // no integral types
+        { DataType::U16, DataType::U16, DataType::U16, false },                                              // no integral types
+        { DataType::U8, DataType::U8, DataType::U8, false },                                                 // no integral types
+    };
+
+    const TensorShape      shape = TensorShape(10U, 10U);
+    const MatMulKernelInfo matmul_kernel_info{ false, false, 1, 1, 1, false };
+    for(auto &tuple : data_type_configurations)
+    {
+        const bool expected = std::get<3>(tuple);
+
+        const TensorInfo lhs_info(shape, 1, std::get<0>(tuple));
+        const TensorInfo rhs_info(shape, 1, std::get<1>(tuple));
+        TensorInfo       output_info(shape, 1, std::get<2>(tuple));
+
+        Status status = ClNativeMatMulKernel::validate(&lhs_info, &rhs_info, &output_info, matmul_kernel_info);
+        ARM_COMPUTE_EXPECT(bool(status) == expected, framework::LogLevel::ERRORS);
+    }
+}
+
+TEST_SUITE_END() // Validate
+
+TEST_SUITE(Float)
+TEST_SUITE(FP32)
+FIXTURE_DATA_TEST_CASE(RunTiny, CLMatMulKernelFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::TinyMatMulDataset(),
+                                                                                                                   framework::dataset::make("pretransose_A", { false, true })),
+                                                                                                                   framework::dataset::make("pretransose_B", { false, true })),
+                                                                                                                   m0_values_precommit),
+                                                                                                                   n0_values_precommit),
+                                                                                                           k0_values_precommit),
+                                                                                                   framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+FIXTURE_DATA_TEST_CASE(RunSmall, CLMatMulKernelFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallMatMulDataset(),
+                                                                                                                    framework::dataset::make("pretransose_A", { false, true })),
+                                                                                                                    framework::dataset::make("pretransose_B", { false, true })),
+                                                                                                                    m0_values_precommit),
+                                                                                                                    n0_values_precommit),
+                                                                                                            k0_values_precommit),
+                                                                                                    framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeNoTranspose, CLMatMulKernelFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                   framework::dataset::make("pretransose_A", { false })),
+                                                                                                                   framework::dataset::make("pretransose_B", { false })),
+                                                                                                                   m0_values_nightly_lhs_nt),
+                                                                                                                   n0_values_nightly_rhs_nt),
+                                                                                                                   k0_values_nightly_lhs_nt_rhs_nt),
+                                                                                                                   framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeRhsTransposed, CLMatMulKernelFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                     framework::dataset::make("pretransose_A", { false })),
+                                                                                                                     framework::dataset::make("pretransose_B", { true })),
+                                                                                                                     m0_values_nightly_lhs_nt),
+                                                                                                                     n0_values_nightly_rhs_t),
+                                                                                                                     k0_values_nightly_rhs_t),
+                                                                                                                     framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeLhsTransposed, CLMatMulKernelFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                     framework::dataset::make("pretransose_A", { true })),
+                                                                                                                     framework::dataset::make("pretransose_B", { false })),
+                                                                                                                     m0_values_nightly_lhs_t),
+                                                                                                                     n0_values_nightly_rhs_nt),
+                                                                                                                     k0_values_nightly_lhs_t_rhs_nt),
+                                                                                                                     framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeLhsTransposedRhsTransposed, CLMatMulKernelFixture<float>, framework::DatasetMode::NIGHTLY,
+                       combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                       framework::dataset::make("pretransose_A", { true })),
+                                                               framework::dataset::make("pretransose_B", { true })),
+                                                       m0_values_nightly_lhs_t),
+                                               n0_values_nightly_rhs_t),
+                                       k0_values_nightly_rhs_t),
+                               framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+// Running High Dimensional test is enough for FP32, because we're stressing the number of dimensions, not data type or M0/N0/K0
+// It's a good idea to test for each Lhs/Rhs T/NT combinations because they're different CL kernels
+FIXTURE_DATA_TEST_CASE(RunHighDimensional, CLMatMulKernelFixture<float>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::HighDimensionalMatMulDataset(),
+                                                                                                                      framework::dataset::make("pretransose_A", { false, true })),
+                                                                                                                      framework::dataset::make("pretransose_B", { false, true })),
+                                                                                                                      framework::dataset::make("M0", { 2 })),
+                                                                                                                      framework::dataset::make("N0", { 2 })),
+                                                                                                                      framework::dataset::make("K0", { 2 })),
+                                                                                                              framework::dataset::make("DataType", DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32, 0.f, abs_tolerance_f32);
+}
+TEST_SUITE_END() // FP32
+
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLMatMulKernelFixture<half>, framework::DatasetMode::ALL, combine(combine(combine(combine(combine(combine(datasets::SmallMatMulDataset(),
+                                                                                                                   framework::dataset::make("pretransose_A", { false, true })),
+                                                                                                                   framework::dataset::make("pretransose_B", { false, true })),
+                                                                                                                   m0_values_precommit),
+                                                                                                                   n0_values_precommit),
+                                                                                                           k0_values_precommit),
+                                                                                                   framework::dataset::make("DataType", DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeNoTranspose, CLMatMulKernelFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                  framework::dataset::make("pretransose_A", { false })),
+                                                                                                                  framework::dataset::make("pretransose_B", { false })),
+                                                                                                                  m0_values_nightly_lhs_nt),
+                                                                                                                  n0_values_nightly_rhs_nt),
+                                                                                                                  k0_values_nightly_lhs_nt_rhs_nt),
+                                                                                                                  framework::dataset::make("DataType", DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeRhsTransposed, CLMatMulKernelFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                    framework::dataset::make("pretransose_A", { false })),
+                                                                                                                    framework::dataset::make("pretransose_B", { true })),
+                                                                                                                    m0_values_nightly_lhs_nt),
+                                                                                                                    n0_values_nightly_rhs_t),
+                                                                                                                    k0_values_nightly_rhs_t),
+                                                                                                                    framework::dataset::make("DataType", DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeLhsTransposed, CLMatMulKernelFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                                                                                                                    framework::dataset::make("pretransose_A", { true })),
+                                                                                                                    framework::dataset::make("pretransose_B", { false })),
+                                                                                                                    m0_values_nightly_lhs_t),
+                                                                                                                    n0_values_nightly_rhs_nt),
+                                                                                                                    k0_values_nightly_lhs_t_rhs_nt),
+                                                                                                                    framework::dataset::make("DataType", DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeLhsTransposedRhsTransposed, CLMatMulKernelFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(combine(combine(combine(combine(datasets::LargeMatMulDataset(),
+                       framework::dataset::make("pretransose_A", { true })),
+                       framework::dataset::make("pretransose_B", { true })),
+                       m0_values_nightly_lhs_t),
+                       n0_values_nightly_rhs_t),
+                       k0_values_nightly_rhs_t),
+                       framework::dataset::make("DataType", DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f16, 0.f, abs_tolerance_f16);
+}
+TEST_SUITE_END() // FP16
+TEST_SUITE_END() // Float
+TEST_SUITE_END() // MatMulKernel
+TEST_SUITE_END() // CL
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/fixtures/BatchMatMulFixture.h b/tests/validation/fixtures/MatMulKernelFixture.h
similarity index 94%
rename from tests/validation/fixtures/BatchMatMulFixture.h
rename to tests/validation/fixtures/MatMulKernelFixture.h
index 9fb2dcc..4595646 100644
--- a/tests/validation/fixtures/BatchMatMulFixture.h
+++ b/tests/validation/fixtures/MatMulKernelFixture.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ACL_TESTS_VALIDATION_FIXTURES_BATCHMATMULFIXTURE
-#define ACL_TESTS_VALIDATION_FIXTURES_BATCHMATMULFIXTURE
+#ifndef ACL_TESTS_VALIDATION_FIXTURES_MATMULKERNELFIXTURE
+#define ACL_TESTS_VALIDATION_FIXTURES_MATMULKERNELFIXTURE
 
 #include "arm_compute/core/KernelDescriptors.h"
 #include "src/gpu/cl/kernels/ClNativeMatMulKernel.h"
@@ -44,7 +44,7 @@
 using namespace arm_compute::opencl::kernels;
 
 template <typename T>
-class BatchMatMulValidationFixture : public framework::Fixture
+class MatMulKernelValidationFixture : public framework::Fixture
 {
 public:
     template <typename...>
@@ -96,7 +96,7 @@
         CLTensor b   = create_tensor<CLTensor>(shape_b, data_type, 1);
         CLTensor dst = create_tensor<CLTensor>(output_shape, data_type, 1);
 
-        CLSynthetizeOperator<ClNativeMatMulKernel> batchMatMul{};
+        CLSynthetizeOperator<ClNativeMatMulKernel> matMul{};
         MatMulKernelInfo                           matmul_info;
         matmul_info.adj_lhs = pretranspose_a;
         matmul_info.adj_rhs = pretranspose_b;
@@ -104,7 +104,7 @@
         matmul_info.n0      = N0;
         matmul_info.k0      = K0;
 
-        batchMatMul.configure(a.info(), b.info(), dst.info(), matmul_info);
+        matMul.configure(a.info(), b.info(), dst.info(), matmul_info);
         ARM_COMPUTE_ASSERT(a.info()->is_resizable());
         ARM_COMPUTE_ASSERT(b.info()->is_resizable());
         ARM_COMPUTE_ASSERT(dst.info()->is_resizable());
@@ -122,12 +122,12 @@
         fill(CLAccessor(a), 0);
         fill(CLAccessor(b), 1);
 
-        // Compute batchMatMul kernel
+        // Compute matMul kernel
         ITensorPack tensors_pack({ { ACL_SRC_0, &a },
             { ACL_SRC_1, &b },
             { ACL_DST, &dst }
         });
-        batchMatMul.run(tensors_pack);
+        matMul.run(tensors_pack);
 
         return dst;
     }
@@ -200,4 +200,4 @@
 } // namespace validation
 } // namespace test
 } // namespace arm_compute
-#endif /* ACL_TESTS_VALIDATION_FIXTURES_BATCHMATMULFIXTURE */
+#endif /* ACL_TESTS_VALIDATION_FIXTURES_MATMULKERNELFIXTURE */