COMPMID-592: Added CL GEMMInterleave4x4 tests.

Change-Id: I4df011b451df56990e86edcd248b7d53f57b2ba7
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/112545
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Jenkins <bsgcomp@arm.com>

2 files changed, 109 insertions, 1 deletions

diff --git a/tests/CL/Helper.h b/tests/CL/Helper.h
new file mode 100644
index 0000000000..78b6d0734d
--- /dev/null
+++ b/tests/CL/Helper.h
@@ -0,0 +1,65 @@
+/*
+ * Copyright (c) 2017 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.
+ */
+#ifndef __ARM_COMPUTE_TEST_CL_HELPER_H__
+#define __ARM_COMPUTE_TEST_CL_HELPER_H__
+
+#include "arm_compute/runtime/CL/ICLSimpleFunction.h"
+#include "support/ToolchainSupport.h"
+
+namespace arm_compute
+{
+namespace test
+{
+// This template synthetizes an ICLSimpleFunction which runs the given kernel K
+template <typename K>
+class CLSynthetizeFunction : public ICLSimpleFunction
+{
+public:
+    template <typename... Args>
+    void configure(Args &&... args)
+    {
+        auto k = arm_compute::support::cpp14::make_unique<K>();
+        k->configure(std::forward<Args>(args)...);
+        _kernel = std::move(k);
+    }
+};
+
+// As above but this also setups a Zero border on the input tensor of the specified bordersize
+template <typename K, int bordersize>
+class CLSynthetizeFunctionWithZeroConstantBorder : public ICLSimpleFunction
+{
+public:
+    template <typename T, typename... Args>
+    void configure(T first, Args &&... args)
+    {
+        auto k = arm_compute::support::cpp14::make_unique<K>();
+        k->configure(first, std::forward<Args>(args)...);
+        _kernel = std::move(k);
+        _border_handler.configure(first, BorderSize(bordersize), BorderMode::CONSTANT, PixelValue(0));
+    }
+};
+
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_CL_HELPER_H__ */
diff --git a/tests/validation/CL/GEMM.cpp b/tests/validation/CL/GEMM.cpp
index 62671e34d7..255f5d934a 100644
--- a/tests/validation/CL/GEMM.cpp
+++ b/tests/validation/CL/GEMM.cpp
@@ -21,11 +21,13 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
+#include "arm_compute/core/CL/kernels/CLGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/Types.h"
 #include "arm_compute/runtime/CL/CLTensor.h"
 #include "arm_compute/runtime/CL/CLTensorAllocator.h"
 #include "arm_compute/runtime/CL/functions/CLGEMM.h"
 #include "tests/CL/CLAccessor.h"
+#include "tests/CL/Helper.h"
 #include "tests/PaddingCalculator.h"
 #include "tests/datasets/LargeGEMMDataset.h"
 #include "tests/datasets/SmallGEMMDataset.h"
@@ -34,6 +36,7 @@
 #include "tests/framework/datasets/Datasets.h"
 #include "tests/validation/Validation.h"
 #include "tests/validation/fixtures/GEMMFixture.h"
+#include "tests/validation/fixtures/GEMMInterleave4x4Fixture.h"
 
 namespace arm_compute
 {
@@ -46,7 +49,8 @@ namespace
 RelativeTolerance<float>            tolerance_f32(0.001f);    /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
 RelativeTolerance<half_float::half> tolerance_f16(half(0.2)); /**< Tolerance value for comparing reference's output against implementation's output for floating point data types */
 constexpr AbsoluteTolerance<float>  tolerance_q(1.0f);        /**< Tolerance value for comparing reference's output against implementation's output for fixed point data types */
-constexpr float                     tolerance_num = 0.02f;    /**< Tolerance number */
+constexpr float                     tolerance_num   = 0.02f;  /**< Tolerance number */
+const auto                          data_interleave = framework::dataset::make("M", 8, 14) * framework::dataset::make("N", 7, 14);
 
 /** CNN data types */
 const auto CNNDataTypes = framework::dataset::make("DataType",
@@ -61,6 +65,45 @@ const auto CNNDataTypes = framework::dataset::make("DataType",
 TEST_SUITE(CL)
 TEST_SUITE(GEMM)
 
+TEST_SUITE(INTERLEAVE_4X4)
+using CLGEMMInterleave4x4 = CLSynthetizeFunctionWithZeroConstantBorder<CLGEMMInterleave4x4Kernel, 4>;
+
+TEST_SUITE(FP32)
+using CLGEMMInterleave4x4Fixture = GEMMInterleave4x4ValidationFixture<CLTensor, CLAccessor, CLGEMMInterleave4x4, float>;
+FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMInterleave4x4Fixture, framework::DatasetMode::PRECOMMIT, data_interleave * framework::dataset::make("DataType", DataType::F32))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END() // FP32
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QS8)
+using CLGEMMInterleave4x4Fixture = GEMMInterleave4x4ValidationFixedPointFixture<CLTensor, CLAccessor, CLGEMMInterleave4x4, int8_t>;
+FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMInterleave4x4Fixture, framework::DatasetMode::PRECOMMIT, data_interleave *
+                       framework::dataset::make("DataType", DataType::QS8)
+                       * framework::dataset::make("FractionalBits", 1, 7))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(QS16)
+using CLGEMMInterleave4x4Fixture = GEMMInterleave4x4ValidationFixedPointFixture<CLTensor, CLAccessor, CLGEMMInterleave4x4, int16_t>;
+FIXTURE_DATA_TEST_CASE(RunSmall, CLGEMMInterleave4x4Fixture, framework::DatasetMode::PRECOMMIT, data_interleave *
+                       framework::dataset::make("DataType", DataType::QS16)
+                       * framework::dataset::make("FractionalBits", 1, 14))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE_END()
+
+TEST_SUITE_END() // INTERLEAVE_4X4
+
 DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallGEMMDataset(), datasets::LargeGEMMDataset()), CNNDataTypes),
                shape_a, shape_b, shape_c, output_shape, alpha, beta, data_type)
 {