Merge branch 'dev' into 'main'

version 0.2.0

See merge request !15

src/utils/CudaUtils.cpp

@@ -40,7 +40,7 @@ void Aidge::Cuda::setMultiDevicePeerAccess(unsigned int size, unsigned int* devi
                     CHECK_CUDA_STATUS(cudaSetDevice(devices[j]));
                     const cudaError_t status = cudaDeviceEnablePeerAccess(devices[i], 0);
                     if (status == cudaErrorPeerAccessAlreadyEnabled) {
-                        printf("Peer access already enabled between device %d and device %d\n", devices[j], devices[i]);
+                        fmt::print("Peer access already enabled between device {} and device {}\n", devices[j], devices[i]);
                     } else {
                         CHECK_CUDA_STATUS(status);
                     }

unit_tests/Test_AvgPoolingImpl.cpp

+/********************************************************************************
+ * Copyright (c) 2024 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <array>
+
+#include <catch2/catch_test_macros.hpp>
+#include <cuda_fp16.h>
+#include <numeric>   // std::accumulate
+#include <random>    // std::random_device, std::mt19937, std::uniform_real_distribution
+
+#include "Test_cuda.hpp"
+
+#include "aidge/data/half.hpp"
+#include "aidge/data/Tensor.hpp"
+
+#include "aidge/backend/cpu.hpp"
+#include "aidge/backend/cuda.hpp"
+
+using namespace Aidge;
+
+TEST_CASE("[gpu/operator] AvgPooling(forward)", "[AvgPooling][GPU]") {
+    std::shared_ptr<Tensor> myInput = std::make_shared<Tensor>(Array4D<float,2,2,5,5> { //NCHW
+        {
+            {
+                {{  0,   1,   2,   3,   4},
+                 {  5,   6,   7,   8,   9},
+                 { 10,  11,  12,  13,  14},
+                 { 15,  16,  17,  18,  19},
+                 { 20,  21,  22,  23,  24}},
+
+                {{ 25,  26,  27,  28,  29},
+                 { 30,  31,  32,  33,  34},
+                 { 35,  36,  37,  38,  39},
+                 { 40,  41,  42,  43,  44},
+                 { 45,  46,  47,  48,  49}}
+            },
+            {
+                {{100, 101, 102, 103, 104},
+                 {105, 106, 107, 108, 109},
+                 {110, 111, 112, 113, 114},
+                 {115, 116, 117, 118, 119},
+                 {120, 121, 122, 123, 124}},
+
+                {{125, 126, 127, 128, 129},
+                 {130, 131, 132, 133, 134},
+                 {135, 136, 137, 138, 139},
+                 {140, 141, 142, 143, 144},
+                 {145, 146, 147, 148, 149}}
+            }
+        }
+    });
+    SECTION("Stride") {
+        std::shared_ptr<Node> myAvgPool = AvgPooling({2,2}, "myAvgPool", {2,2});
+        auto op = std::static_pointer_cast<OperatorTensor>(myAvgPool -> getOperator());
+
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array4D<float,2,2,2,2> {
+            {
+                {
+                    {{  3,   5},
+                     { 13,  15}},
+                    {{ 28,  30},
+                     { 38,  40}}
+                },
+                {
+                    {{103, 105},
+                     {113, 115}},
+                    {{128, 130},
+                     {138, 140}}
+                }
+            }
+        });
+        op->associateInput(0,myInput);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myAvgPool->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+
+    SECTION("Stride >= feature dim") {
+        std::shared_ptr<Tensor> myInput2 = std::make_shared<Tensor>(Array4D<float,1,1,3,3> { //NCHW
+        {
+            {
+                {{0.3745, 0.9507, 0.7320},
+                 {0.5987, 0.1560, 0.1560},
+                 {0.0581, 0.8662, 0.6011}}
+            }
+        }
+        });
+        std::shared_ptr<Node> myAvgPool = AvgPooling({3,3}, "myAvgPool", {3,3});
+        auto op = std::static_pointer_cast<OperatorTensor>(myAvgPool -> getOperator());
+
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array4D<float,1,1,1,1> {
+            {{{{(0.3745 + 0.9507 + 0.7320 + 0.5987 + 0.1560 + 0.1560 + 0.0581 + 0.8662 + 0.6011)/9.0}}}}
+        });
+        op->associateInput(0,myInput2);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myAvgPool->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+
+    SECTION("half") {
+        std::shared_ptr<Tensor> myInput2 = std::make_shared<Tensor>(Array4D<half_float::half,1,1,3,3> { //NCHW
+        {
+            {
+                {{half_float::half(0.3745), half_float::half(0.9507), half_float::half(0.7320)},
+                 {half_float::half(0.5987), half_float::half(0.1560), half_float::half(0.1560)},
+                 {half_float::half(0.0581), half_float::half(0.8662), half_float::half(0.6011)}}
+            }
+        }
+        });
+        myInput2->setBackend("cuda");
+
+        std::shared_ptr<Node> myAvgPool = AvgPooling({3,3}, "mymyAvgPoolcdw", {3,3});
+        auto op = std::static_pointer_cast<OperatorTensor>(myAvgPool -> getOperator());
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array4D<half_float::half,1,1,1,1> {
+            {{{{(half_float::half(0.3745) + half_float::half(0.9507) + half_float::half(0.7320) + half_float::half(0.5987) + half_float::half(0.1560) + half_float::half(0.1560) + half_float::half(0.0581) + half_float::half(0.8662) + half_float::half(0.6011))/half_float::half(9.0)}}}}
+        });
+        op->associateInput(0,myInput2);
+        op->setDataType(DataType::Float16);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myAvgPool->forward();
+
+        half_float::half* computedOutput   = new half_float::half[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(half_float::half) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const half_float::half targetOutput = *(static_cast<half_float::half*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+}
\ No newline at end of file

unit_tests/Test_CastMove.cpp

@@ -18,8 +18,8 @@
 #include "aidge/graph/Node.hpp"
 #include "aidge/graph/GraphView.hpp"
 #include "aidge/graph/OpArgs.hpp"
-#include "aidge/scheduler/Scheduler.hpp"
-#include "aidge/recipies/Recipies.hpp"
+#include "aidge/scheduler/SequentialScheduler.hpp"
+#include "aidge/recipes/Recipes.hpp"
 
 #include "aidge/backend/cuda.hpp"
 

unit_tests/Test_FCImpl.cpp

+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <array>
+
+#include <catch2/catch_test_macros.hpp>
+
+#include "Test_cuda.hpp"
+
+#include "aidge/data/Tensor.hpp"
+
+#include "aidge/backend/cpu.hpp"
+#include "aidge/backend/cuda.hpp"
+
+using namespace Aidge;
+
+TEST_CASE("[gpu/operator] FC(forward)", "[FC][GPU]") {
+    std::shared_ptr<Tensor> myWeights = std::make_shared<Tensor>(Array2D<float, 5, 75>{
+            {{1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,
+              5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,
+              9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12,
+              13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15},
+             {1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,
+              5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,
+              9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12,
+              13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15},
+             {1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,
+              5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,
+              9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12,
+              13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15},
+             {1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,
+              5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,
+              9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12,
+              13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15},
+             {1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,
+              5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,
+              9,  10, 11, 12, 13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12,
+              13, 14, 15, 1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15}}});
+    std::shared_ptr<Tensor> myBias = std::make_shared<Tensor>(Array1D<float, 5>{{1, 2, 3, 4, 5}});
+    std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array2D<float, 2, 5>{
+            {{23601, 23602, 23603, 23604, 23605}, {68601, 68602, 68603, 68604, 68605}}});
+    myWeights->setBackend("cuda");
+    myBias->setBackend("cuda");
+    std::shared_ptr<Node> myFC = FC(75, 5, false, "myfc");
+    auto op = std::static_pointer_cast<OperatorTensor>(myFC -> getOperator());
+    op -> associateInput(1, myWeights);
+    op -> associateInput(2, myBias);
+    SECTION("2D input") {
+        std::shared_ptr<Tensor> myInput = std::make_shared<Tensor>(Array2D<float, 2, 75>{
+                {{0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16, 17, 18,
+                  19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37,
+                  38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56,
+                  57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74},
+                 {75,  76,  77,  78,  79,  80,  81,  82,  83,  84,  85,  86,  87,  88,  89,
+                  90,  91,  92,  93,  94,  95,  96,  97,  98,  99,  100, 101, 102, 103, 104,
+                  105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119,
+                  120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134,
+                  135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149}}});
+        myInput->setBackend("cuda");
+        op->associateInput(0, myInput);
+        op -> setDataType(DataType::Float32);
+        op -> setBackend("cuda");
+        op->computeOutputDims();
+        myFC->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+    SECTION("4D input") {
+        std::shared_ptr<Tensor> myInput =
+                std::make_shared<Tensor>(Array4D<float, 2, 3, 5, 5>{{{{{0, 1, 2, 3, 4},
+                                                                     {5, 6, 7, 8, 9},
+                                                                     {10, 11, 12, 13, 14},
+                                                                     {15, 16, 17, 18, 19},
+                                                                     {20, 21, 22, 23, 24}},
+                                                                    {{25, 26, 27, 28, 29},
+                                                                     {30, 31, 32, 33, 34},
+                                                                     {35, 36, 37, 38, 39},
+                                                                     {40, 41, 42, 43, 44},
+                                                                     {45, 46, 47, 48, 49}},
+                                                                    {{50, 51, 52, 53, 54},
+                                                                     {55, 56, 57, 58, 59},
+                                                                     {60, 61, 62, 63, 64},
+                                                                     {65, 66, 67, 68, 69},
+                                                                     {70, 71, 72, 73, 74}}},
+                                                                   {{{75, 76, 77, 78, 79},
+                                                                     {80, 81, 82, 83, 84},
+                                                                     {85, 86, 87, 88, 89},
+                                                                     {90, 91, 92, 93, 94},
+                                                                     {95, 96, 97, 98, 99}},
+                                                                    {{100, 101, 102, 103, 104},
+                                                                     {105, 106, 107, 108, 109},
+                                                                     {110, 111, 112, 113, 114},
+                                                                     {115, 116, 117, 118, 119},
+                                                                     {120, 121, 122, 123, 124}},
+                                                                    {{125, 126, 127, 128, 129},
+                                                                     {130, 131, 132, 133, 134},
+                                                                     {135, 136, 137, 138, 139},
+                                                                     {140, 141, 142, 143, 144},
+                                                                     {145, 146, 147, 148, 149}}}}});
+        myInput->setBackend("cuda");
+        op->associateInput(0, myInput);
+        op -> setDataType(DataType::Float32);
+        op -> setBackend("cuda");
+        op->computeOutputDims();
+        myFC->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+}
\ No newline at end of file

unit_tests/Test_MaxPoolingImpl.cpp

+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <array>
+
+#include <catch2/catch_test_macros.hpp>
+
+#include "Test_cuda.hpp"
+
+#include "aidge/data/Tensor.hpp"
+
+#include "aidge/backend/cpu.hpp"
+#include "aidge/backend/cuda.hpp"
+
+using namespace Aidge;
+
+
+TEST_CASE("[cpu/operator] MaxPooling(forward)", "[MaxPooling][CPU]") {
+    std::shared_ptr<Tensor> myInput = std::make_shared<Tensor>(Array4D<float,2,2,5,5> { //NCHW
+        {
+            {
+                {{-0.3848,  0.2166, -0.4373,  0.6142,  0.5277},
+                 {0.7995,  0.3638, -1.4589, -1.0843,  1.0918},
+            	 {0.7147,  0.0936, -1.2902,  1.2037,  0.4874},
+                 {-0.5981,  2.1184, -0.9175,  1.3859,  0.3305},
+                 {-1.7700,  0.0563, -0.3914,  0.0538, -0.3955}},
+
+                {{-3.1409, -0.4554,  0.0524,  2.2291,  0.4859},
+                 {-0.7465, -0.6567, -2.3703, -0.6386, -1.4152},
+                 { 2.2329, -0.5850,  0.0700,  1.2838, -1.7363},
+                 { 0.2139,  0.0624, -1.0689, -0.8221, -0.8038},
+                 { 0.1886, -0.7840, -0.2313,  0.2651, -1.6244}}
+            },
+            {
+                {{ 0.4371,  1.6417,  0.9129,  0.6325,  0.5438},
+                 {-2.3552, -0.8850, -0.0232, -0.5462, -1.2011},
+                 {1.7653, -1.6668, -1.0814,  0.6182,  1.2071},
+                 {0.9541, -0.5133,  0.8664, -0.8892,  1.4585},
+                 {1.0220, -0.5107,  0.1829, -0.2301, -0.4268}},
+
+                {{ 1.0429,  0.6279, -0.2875,  0.7187, -0.1500},
+                 {1.6041,  2.9635,  1.4172, -0.7517,  0.5441},
+                 {-0.2276,  0.0857,  0.6776, -0.1389, -0.0614},
+                 {-0.1547, -0.3435,  0.0650, -0.5095, -1.8073},
+                 {1.7217,  0.3999, -0.5953,  1.0604, -0.4126}}
+            }
+        }
+    });
+    SECTION("Stride") {
+        std::shared_ptr<Node> myMaxPool = MaxPooling({2,2}, "mycdw", {2,2});
+        auto op = std::static_pointer_cast<OperatorTensor>(myMaxPool -> getOperator());
+
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array4D<float,2,2,2,2> {
+            {
+                {
+                    {{  0.7995,  0.6142},
+                     { 2.1184,  1.3859}},
+                    {{ -0.4554,  2.2291},
+                     {  2.2329,  1.2838}}
+                },
+                {
+                    {{1.6417,  0.9129},
+                     {1.7653,  0.8664}},
+                    {{2.9635,  1.4172},
+                     {0.0857,  0.6776}}
+                }
+            }
+        });
+        myMaxPool->getOperator()->associateInput(0,myInput);
+        myMaxPool->getOperator()->setDataType(DataType::Float32);
+        myMaxPool->getOperator()->setBackend("cuda");
+        op->computeOutputDims();
+        myMaxPool->forward();
+        
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+}
\ No newline at end of file

unit_tests/Test_ReLUImpl.cpp

+/********************************************************************************
+ * Copyright (c) 2023 CEA-List
+ *
+ * This program and the accompanying materials are made available under the
+ * terms of the Eclipse Public License 2.0 which is available at
+ * http://www.eclipse.org/legal/epl-2.0.
+ *
+ * SPDX-License-Identifier: EPL-2.0
+ *
+ ********************************************************************************/
+
+#include <array>
+
+#include <catch2/catch_test_macros.hpp>
+
+#include "Test_cuda.hpp"
+
+#include "aidge/data/Tensor.hpp"
+
+#include "aidge/backend/cpu.hpp"
+#include "aidge/backend/cuda.hpp"
+
+using namespace Aidge;
+
+
+TEST_CASE("[gpu/operator] ReLU(forward)", "[ReLU][GPU]") {
+    SECTION("1D Tensor") {
+        std::shared_ptr<Tensor> input0 = std::make_shared<Tensor>(Array1D<float,10> {
+            {0, 1, 2,-3, 4,-5,-6, 7, 8, 9}
+        });
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array1D<float,10> {
+            {0, 1, 2, 0, 4, 0, 0, 7, 8, 9}
+        });
+
+        std::shared_ptr<Node> myReLU = ReLU();
+        auto op = std::static_pointer_cast<OperatorTensor>(myReLU -> getOperator());
+        op->associateInput(0,input0);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myReLU->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+
+    SECTION("2D Tensor") {
+        std::shared_ptr<Tensor> input0 = std::make_shared<Tensor>(Array2D<float,2,10> {
+            {
+                { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+            }
+        });
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array2D<float,2,10> {
+            {
+                { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+            }
+        });
+
+        std::shared_ptr<Node> myReLU = ReLU();
+        auto op = std::static_pointer_cast<OperatorTensor>(myReLU -> getOperator());
+        op->associateInput(0,input0);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myReLU->forward();
+        
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+
+    SECTION("3D Tensor") {
+        std::shared_ptr<Tensor> input0 = std::make_shared<Tensor>(Array3D<float,2,2,10> {
+            {
+                {
+                    { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                    {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                },
+                {
+                    { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                    {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                }
+            }
+        });
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array3D<float,2,2,10> {
+            {
+                {
+                    { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                    { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                },
+                {
+                    { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                    { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                }
+            }
+        });
+
+        std::shared_ptr<Node> myReLU = ReLU();
+        auto op = std::static_pointer_cast<OperatorTensor>(myReLU -> getOperator());
+        op->associateInput(0,input0);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        myReLU->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+
+    SECTION("4D Tensor") {
+        std::shared_ptr<Tensor> input0 = std::make_shared<Tensor>(Array4D<float,2,2,2,10> {
+            {
+                {
+                    {
+                        { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                        {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                    },
+                    {
+                        { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                        {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                    }
+                },
+                {
+                    {
+                        { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                        {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                    },
+                    {
+                        { 0, 1, 2,-3, 4,-5,-6, 7, 8, 9},
+                        {-5, 4, 2,-3, 4,-5,-6, 7,-1,10}
+                    }
+                }
+            }
+        });
+        std::shared_ptr<Tensor> myOutput = std::make_shared<Tensor>(Array4D<float,2,2,2,10> {
+            {
+                {
+                    {
+                        { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                        { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                    },
+                    {
+                        { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                        { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                    }
+                },
+                {
+                    {
+                        { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                        { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                    },
+                    {
+                        { 0, 1, 2, 0, 4, 0, 0, 7, 8, 9},
+                        { 0, 4, 2, 0, 4, 0, 0, 7, 0,10}
+                    }
+                }
+            }
+        });
+
+        std::shared_ptr<Node> myReLU = ReLU();
+        auto op = std::static_pointer_cast<OperatorTensor>(myReLU -> getOperator());
+        op->associateInput(0,input0);
+        op->setDataType(DataType::Float32);
+        op->setBackend("cuda");
+        op->computeOutputDims();
+        op->forward();
+
+        float* computedOutput   = new float[myOutput->size()]();
+        cudaMemcpy(computedOutput, op->getOutput(0)->getImpl()->rawPtr(), sizeof(float) * myOutput->size(), cudaMemcpyDeviceToHost);
+
+        for(int i = 0; i < myOutput->size(); i++){
+            const float targetOutput = *(static_cast<float*>(myOutput->getImpl()->rawPtr()) + i);
+            REQUIRE(fabs(computedOutput[i] - targetOutput) < 1e-6);
+        }
+
+        delete[] computedOutput;
+    }
+}

version.txt

-0.0.1
\ No newline at end of file
+0.2.0
\ No newline at end of file