diff --git a/README.md b/README.md
index 74eb50826bf6f88a0ded363138adba04827390d0..865cb08a17ebf8638cb2ac56773a4f464860b8ae 100644
--- a/README.md
+++ b/README.md
@@ -14,9 +14,11 @@ So far be sure to have the correct requirements to use this library
## Pip installation
-You will need to install first the aidge_core library before installing aidge_cpu.
-Also, make sure that the install path was set before installing aidge_core library.
-Then run in your python environnement :
+You will need to install first the ``aidge_core`` library before installing ``aidge_backend_cpu``.
+
+If you have set a custom install path for the ``aidge_core`` library, make sure to use the same one here.
+
+Then run in your python environnement :
``` bash
pip install . -v
```
@@ -46,4 +48,4 @@ Important: this command can also be run with `make`.
To compile the CPU library with the python binding + the associated unitary tests, run
```
make cpu_with_pybind_tests
-```
\ No newline at end of file
+```
diff --git a/aidge_backend_cpu/unit_tests/test_recipies.py b/aidge_backend_cpu/unit_tests/test_recipies.py
new file mode 100644
index 0000000000000000000000000000000000000000..60949adf245f4f4a7ed316879fb307131f70739a
--- /dev/null
+++ b/aidge_backend_cpu/unit_tests/test_recipies.py
@@ -0,0 +1,77 @@
+"""
+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
+"""
+
+import unittest
+import aidge_core
+import aidge_backend_cpu
+
+from functools import reduce
+import numpy as np
+
+class test_recipies(unittest.TestCase):
+ def setUp(self):
+ pass
+
+ def tearDown(self):
+ pass
+
+ def test_fuse_batchnorm(self):
+ dims = [1, 1, 10, 10]
+ size = reduce((lambda x, y: x*y), dims)
+
+ input_data = np.arange(size).reshape(dims).astype(np.float32)
+ input_tensor = aidge_core.Tensor(input_data)
+
+ input_node = aidge_core.Producer(input_tensor, "X")
+ conv = aidge_core.Conv2D(1, 1, [3, 3], name="Conv0")
+ bn = aidge_core.BatchNorm2D(name="Add0")
+
+ graph_view = aidge_core.sequential([conv, bn])
+
+ # Add random values to conv and BatchNorm parameters
+ input_node.add_child(graph_view)
+ input_node.get_operator().set_datatype(aidge_core.DataType.Float32)
+ input_node.get_operator().set_backend("cpu")
+ graph_view.set_datatype(aidge_core.DataType.Float32)
+ graph_view.set_backend("cpu")
+
+ np_weights = np.arange(9).reshape([1, 1, 3, 3]).astype(np.float32)
+ np_bias = np.arange(1).reshape([1, 1]).astype(np.float32)
+
+ np_scale = np.array([0.05]).astype(np.float32)
+ np_shift = np.array([0.05]).astype(np.float32)
+ np_mean = np.array([0.05]).astype(np.float32)
+ np_var = np.array([0.05]).astype(np.float32)
+ conv.input(1)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_weights))
+ conv.input(2)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_bias))
+ bn.input(1)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_scale))
+ bn.input(2)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_shift))
+ bn.input(3)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_mean))
+ bn.input(4)[0].get_operator().set_output_tensor(aidge_core.Tensor(np_var))
+ scheduler0 = aidge_core.SequentialScheduler(graph_view)
+ scheduler0.forward()
+
+ for outNode in graph_view.get_output_nodes():
+ output_aidge0 = outNode.get_operator().output(0)
+
+ aidge_core.fuse_batchnorm(graph_view)
+ scheduler1 = aidge_core.SequentialScheduler(graph_view)
+ scheduler1.forward()
+
+ for outNode in graph_view.get_output_nodes():
+ output_aidge1 = outNode.get_operator().output(0)
+
+ self.assertTrue(aidge_core.approx_eq(output_aidge0, output_aidge1, 0.000001, 0.0001))
+
+if __name__ == '__main__':
+ unittest.main()
+
+
+
diff --git a/aidge_backend_cpu/unit_tests/test_tensor.py b/aidge_backend_cpu/unit_tests/test_tensor.py
index 1d12fc0cbadf71f04226a98e2e65984abc7e3254..438b6acd51791a52c9e308fb1aceaefb2a45fb29 100644
--- a/aidge_backend_cpu/unit_tests/test_tensor.py
+++ b/aidge_backend_cpu/unit_tests/test_tensor.py
@@ -45,5 +45,17 @@ class test_tensor(unittest.TestCase):
self.assertTrue(i_t == i_n) # TODO : May need to change this to a difference
for i,j in zip(t.dims(), np_array.shape):
self.assertEqual(i,j)
+
+ def test_get_set(self):
+ dims = [2,2,2]
+
+ np_array = np.arange(8).reshape(dims)
+ # Numpy -> Tensor
+ t = aidge_core.Tensor(np_array)
+ for i in range(8):
+ self.assertEqual(t[i], i)
+ t[i] = 5
+ self.assertEqual(t[i], 5)
+
if __name__ == '__main__':
unittest.main()
diff --git a/include/aidge/backend/cpu.hpp b/include/aidge/backend/cpu.hpp
index 336d549a71f625667e7e3d368819400396b893e1..1f45d700f6fc9f1d69682cb2de601979049c0af6 100644
--- a/include/aidge/backend/cpu.hpp
+++ b/include/aidge/backend/cpu.hpp
@@ -21,6 +21,7 @@
#include "aidge/backend/cpu/operator/ConvImpl.hpp"
#include "aidge/backend/cpu/operator/FCImpl.hpp"
#include "aidge/backend/cpu/operator/LeakyReLUImpl.hpp"
+#include "aidge/backend/cpu/operator/MatMulImpl.hpp"
#include "aidge/backend/cpu/operator/ProducerImpl.hpp"
#include "aidge/backend/cpu/operator/ReLUImpl.hpp"
#include "aidge/backend/cpu/operator/SoftmaxImpl.hpp"
diff --git a/include/aidge/backend/cpu/data/TensorImpl.hpp b/include/aidge/backend/cpu/data/TensorImpl.hpp
index dfcb8afa79c98438ae261a244aee94f4ede6c0b3..012ff5af1c15e73fe76114a23ec62f9ef023bce2 100644
--- a/include/aidge/backend/cpu/data/TensorImpl.hpp
+++ b/include/aidge/backend/cpu/data/TensorImpl.hpp
@@ -47,6 +47,10 @@ class TensorImpl_cpu : public TensorImpl {
return mData.data();
};
+ void* getRaw(std::size_t idx){
+ return static_cast<void*>(static_cast<T *>(rawPtr()) + idx);
+ };
+
virtual ~TensorImpl_cpu() = default;
void setRawPtr(void *ptr) override final {
diff --git a/include/aidge/backend/cpu/operator/AddImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/AddImpl_forward_kernels.hpp
index 490598599aedf24b26865ce6a1ddb3fe32044b1b..221e36dcfac44e21d1b1a35674ca21403b4b57ab 100644
--- a/include/aidge/backend/cpu/operator/AddImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/AddImpl_forward_kernels.hpp
@@ -20,7 +20,7 @@ namespace Aidge {
template <class I1, class O>
void AddImpl1I_cpu_forward_kernel(const std::size_t inputLength, const void* input1_, void* output_) {
- // FIXME: missing Add parameters as arguments
+ // FIXME: missing Add attributes as arguments
const I1* input1 = static_cast<const I1*>(input1_);
O* output = static_cast<O*>(output_);
@@ -32,7 +32,7 @@ void AddImpl1I_cpu_forward_kernel(const std::size_t inputLength, const void* inp
template <class I1, class I2, class O>
void AddImpl2I_cpu_forward_kernel(const std::size_t inputLength, const void* input1_, const void* input2_,
void* output_) {
- // FIXME: missing Add parameters as arguments
+ // FIXME: missing Add attributes as arguments
const I1* input1 = static_cast<const I1*>(input1_);
const I2* input2 = static_cast<const I2*>(input2_);
O* output = static_cast<O*>(output_);
@@ -45,7 +45,7 @@ void AddImpl2I_cpu_forward_kernel(const std::size_t inputLength, const void* inp
template <class I1, class I2, class I3, class O>
void AddImpl3I_cpu_forward_kernel(const std::size_t inputLength, const void* input1_, const void* input2_,
const void* input3_, void* output_) {
- // FIXME: missing Add parameters as arguments
+ // FIXME: missing Add attributes as arguments
const I1* input1 = static_cast<const I1*>(input1_);
const I2* input2 = static_cast<const I2*>(input2_);
const I3* input3 = static_cast<const I3*>(input3_);
diff --git a/include/aidge/backend/cpu/operator/AvgPoolingImpl.hpp b/include/aidge/backend/cpu/operator/AvgPoolingImpl.hpp
index 8373cb84a550efd8741a2dbc04c1e94ad37fe611..cfbcadfe6b719369618955a14c4cde5733ef6773 100644
--- a/include/aidge/backend/cpu/operator/AvgPoolingImpl.hpp
+++ b/include/aidge/backend/cpu/operator/AvgPoolingImpl.hpp
@@ -29,11 +29,11 @@ namespace Aidge {
class AvgPoolingImpl2DForward_cpu
: public Registrable<AvgPoolingImpl2DForward_cpu,
std::tuple<DataType, DataType>,
- void(const AvgPooling_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *, void *)> {};
+ void(const AvgPooling_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *, void *)> {};
class AvgPoolingImpl2DBackward_cpu
: public Registrable<AvgPoolingImpl2DBackward_cpu,
std::tuple<DataType, DataType>,
- void(const AvgPooling_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *, void *)> {};
+ void(const AvgPooling_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *, void *)> {};
class AvgPoolingImpl2D_cpu : public OperatorImpl {
private:
diff --git a/include/aidge/backend/cpu/operator/AvgPoolingImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/AvgPoolingImpl_forward_kernels.hpp
index 776e020f1a20056db345c8e845fd73bb31b4138b..60b4923bdc18674da52be9bd07d9947fb9790f0d 100644
--- a/include/aidge/backend/cpu/operator/AvgPoolingImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/AvgPoolingImpl_forward_kernels.hpp
@@ -26,51 +26,51 @@ namespace Aidge {
* @brief Forward kernel for 2D AvgPoolingolution on CPU backend.
* @tparam I Input data type.
* @tparam O Output data type.
- * @param params tuple of Parameters from the Operator
+ * @param params tuple of Attributes from the Operator
* @param dims Array of input dimensions.
* @param input_ const input Tensor.
* @param output_ Output Tensor.
*/
template <class I, class O>
-void AvgPoolingImpl2D_cpu_forward_kernel(const AvgPooling_Op<2>::Parameters ¶ms,
+void AvgPoolingImpl2D_cpu_forward_kernel(const AvgPooling_Op<2>::Attrs &attrs,
const std::array<DimSize_t, 4> &dims,
const void *input_,
void *output_) {
- // FIXME: missing convolution parameters as arguments
+ // FIXME: missing convolution attributes as arguments
const I *input = static_cast<const I *>(input_);
O *output = static_cast<O *>(output_);
// output H size
const std::size_t oxSize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<2>(params)[0] + std::get<2>(params)[2] - std::get<1>(params)[0] + std::get<0>(params)[0]) /
- static_cast<float>(std::get<0>(params)[0])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<2>(attrs)[0] + std::get<2>(attrs)[2] - std::get<1>(attrs)[0] + std::get<0>(attrs)[0]) /
+ static_cast<float>(std::get<0>(attrs)[0])));
// output W size
const std::size_t oySize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<2>(params)[1] + std::get<2>(params)[3] - std::get<1>(params)[1] + std::get<0>(params)[1]) /
- static_cast<float>(std::get<0>(params)[1])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<2>(attrs)[1] + std::get<2>(attrs)[3] - std::get<1>(attrs)[1] + std::get<0>(attrs)[1])/
+ static_cast<float>(std::get<0>(attrs)[1])));
// TODO: kernel computation
// output (batch, outCh, Xout, Yout)
// input (batch, ch, Xin, Yin)
// weight (outCh, ch, kernelX, kernelY)
- // does not take Dilation parameter into account
+ // does not take Dilation attribute into account
using signedsize = std::make_signed<std::size_t>::type;
for (std::size_t batch = 0; batch < dims[0]; ++batch) {
for (std::size_t ch = 0; ch < dims[1]; ++ch) {
const std::size_t oIndex = (ch + batch*dims[1]) * oxSize * oySize;
const std::size_t iIndex = (ch + batch*dims[1]) * dims[2] * dims[3];
for (std::size_t ox = 0; ox < oxSize; ++ox) {
- const signedsize difx = static_cast<signedsize>(std::get<2>(params)[0] - ox * std::get<0>(params)[0]);
+ const signedsize difx = static_cast<signedsize>(std::get<2>(attrs)[0] - ox * std::get<0>(attrs)[0]);
const std::size_t sxMin = static_cast<std::size_t>(std::max(difx, signedsize(0)));
- const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<1>(params)[0] ? std::get<1>(params)[0] : dims[2] + difx);
+ const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<1>(attrs)[0] ? std::get<1>(attrs)[0] : dims[2] + difx);
for (std::size_t oy = 0; oy < oySize; ++oy) {
- const signedsize dify = static_cast<signedsize>(std::get<2>(params)[1] - oy * std::get<0>(params)[1]);
+ const signedsize dify = static_cast<signedsize>(std::get<2>(attrs)[1] - oy * std::get<0>(attrs)[1]);
const std::size_t syMin = static_cast<std::size_t>(std::max(dify, signedsize(0)));
- const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<1>(params)[1] ? std::get<1>(params)[1] : dims[3] + dify);
+ const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<1>(attrs)[1] ? std::get<1>(attrs)[1] : dims[3] + dify);
const std::size_t oIndexFull = oIndex + ox*oySize + oy;
- const std::size_t ix = ox * std::get<0>(params)[0];
- const std::size_t iy = oy * std::get<0>(params)[1];
+ const std::size_t ix = ox * std::get<0>(attrs)[0];
+ const std::size_t iy = oy * std::get<0>(attrs)[1];
if (sxMin == 0 && syMin == 0 && sxMax == 3 && syMax == 3) {
output[oIndexFull] += static_cast<O>(
diff --git a/include/aidge/backend/cpu/operator/BatchNormImpl.hpp b/include/aidge/backend/cpu/operator/BatchNormImpl.hpp
index d9f25b4a8e38510f82fc5afe9ed4b656197a47d5..30557f6cbba05829b3cc9e17364ae4d933a568cf 100644
--- a/include/aidge/backend/cpu/operator/BatchNormImpl.hpp
+++ b/include/aidge/backend/cpu/operator/BatchNormImpl.hpp
@@ -29,7 +29,7 @@ namespace Aidge {
class BatchNormImpl2DForward_cpu
: public Registrable<BatchNormImpl2DForward_cpu,
std::tuple<DataType, DataType, DataType>,
- void(const BatchNorm_Op<2>::Parameters &,
+ void(const BatchNorm_Op<2>::Attrs &,
const std::array<DimSize_t, 4> &,
const void *,
const void *,
@@ -41,7 +41,7 @@ class BatchNormImpl2DForward_cpu
class BatchNormImpl2DBackward_cpu
: public Registrable<BatchNormImpl2DBackward_cpu,
std::tuple<DataType, DataType, DataType>,
- void(const BatchNorm_Op<2>::Parameters &,
+ void(const BatchNorm_Op<2>::Attrs &,
const std::array<DimSize_t, 4> &,
const void *,
const void *,
diff --git a/include/aidge/backend/cpu/operator/BatchNormImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/BatchNormImpl_forward_kernels.hpp
index eedb80bde60d65b53bac70cc33ca83eb4f0121e7..486829e782ae2173332a7efa6646bb7bba322252 100644
--- a/include/aidge/backend/cpu/operator/BatchNormImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/BatchNormImpl_forward_kernels.hpp
@@ -27,7 +27,7 @@ namespace Aidge {
* @tparam W Weight data type.
* @tparam B Bias data type.
* @tparam O Output data type.
- * @param params tuple of Parameters from the Operator
+ * @param params tuple of Attributes from the Operator
* @param dims Array of input dimensions.
* @param input_ const input Tensor.
* @param scale_ const scale Tensor.
@@ -37,9 +37,9 @@ namespace Aidge {
* @param output_ Output Tensor.
*/
template <class I, class P, class O>
-void BatchNormImpl2D_cpu_forward_kernel(const BatchNorm_Op<2>::Parameters ¶ms, const std::array<DimSize_t, 4> &dims,
+void BatchNormImpl2D_cpu_forward_kernel(const BatchNorm_Op<2>::Attrs &attrs, const std::array<DimSize_t, 4> &dims,
const void *input_, const void *scale_, const void *shift_, void *batchMean_, void *batchVar_, void *output_, const bool freeze) {
- // FIXME: missing convolution parameters as arguments
+ // FIXME: missing convolution attributes as arguments
const I *input = static_cast<const I *>(input_);
const P *scale = static_cast<const P *>(scale_);
const P *shift = static_cast<const P *>(shift_);
@@ -52,12 +52,12 @@ void BatchNormImpl2D_cpu_forward_kernel(const BatchNorm_Op<2>::Parameters ¶m
const DimSize_t featureMapSize = dims[2]*dims[3];
- if ((freeze == true) || (std::get<1>(params) == 0.0f)) {
+ if ((freeze == true) || (std::get<1>(attrs) == 0.0f)) {
for (std::size_t batch = 0; batch < nbBatch; ++batch) {
for (std::size_t ch = 0; ch < nbChannels; ++ch) {
const std::size_t ioIndex = (ch + batch*nbChannels) * featureMapSize;
std::fill(output + ioIndex, output + ioIndex + featureMapSize, shift[ch]);
- const P var = std::sqrt(batchVar[ch] + static_cast<P>(std::get<0>(params)));
+ const P var = std::sqrt(batchVar[ch] + static_cast<P>(std::get<0>(attrs)));
for (std::size_t feature = 0; feature<featureMapSize; ++feature) {
output[ioIndex + feature] += scale[ch] * (input[ioIndex + feature]-batchMean[ch]) / var;
@@ -81,10 +81,10 @@ void BatchNormImpl2D_cpu_forward_kernel(const BatchNorm_Op<2>::Parameters ¶m
const I inputMean = sum / static_cast<I>(nbDataPerChannel);
const I inputVar = sumSquare / static_cast<I>(nbDataPerChannel) - inputMean*inputMean;
- batchMean[ch] = batchMean[ch]*(1-std::get<1>(params)) + inputMean*std::get<1>(params);
- batchVar[ch] = batchVar[ch]*(1-std::get<1>(params)) + inputVar*(static_cast<I>(nbDataPerChannel)/static_cast<I>(nbDataPerChannel-1))*std::get<1>(params);
+ batchMean[ch] = batchMean[ch]*(1-std::get<1>(attrs)) + inputMean*std::get<1>(attrs);
+ batchVar[ch] = batchVar[ch]*(1-std::get<1>(attrs)) + inputVar*(static_cast<I>(nbDataPerChannel)/static_cast<I>(nbDataPerChannel-1))*std::get<1>(attrs);
- const P var = std::sqrt(inputVar + static_cast<P>(std::get<0>(params)));
+ const P var = std::sqrt(inputVar + static_cast<P>(std::get<0>(attrs)));
for (std::size_t batch = 0; batch < nbBatch; ++batch) {
const std::size_t ioIndex = (ch + batch*nbChannels) * featureMapSize;
for (std::size_t feature = 0; feature<featureMapSize; ++feature) {
diff --git a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl.hpp b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl.hpp
index 0d21c676d797b2fc4e95c4aea47674c8fca5eef4..2826b635590c5d19f34c8e4beee20fc8dba2183b 100644
--- a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl.hpp
+++ b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl.hpp
@@ -29,12 +29,12 @@ namespace Aidge {
class ConvDepthWiseImpl2DForward_cpu
: public Registrable<ConvDepthWiseImpl2DForward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const ConvDepthWise_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *,
+ void(const ConvDepthWise_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *,
const void *, const void *, void *)> {};
class ConvDepthWiseImpl2DBackward_cpu
: public Registrable<ConvDepthWiseImpl2DBackward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const ConvDepthWise_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *,
+ void(const ConvDepthWise_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *,
const void *, const void *, void *)> {};
class ConvDepthWiseImpl2D_cpu : public OperatorImpl {
diff --git a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_forward_kernels.hpp
index ee2d82e00376c5a2cc5a075565e35eb8885c021e..669bdbc898528b0f96a59dd3c6f8e438ae1291e4 100644
--- a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_forward_kernels.hpp
@@ -27,7 +27,7 @@ namespace Aidge {
* @tparam W Weight data type.
* @tparam B Bias data type.
* @tparam O Output data type.
- * @param params tuple of Parameters from the Operator
+ * @param params tuple of Attributes from the Operator
* @param dims Array of input dimensions.
* @param input_ const input Tensor.
* @param weights_ const weight Tensor.
@@ -35,9 +35,9 @@ namespace Aidge {
* @param output_ Output Tensor.
*/
template <class I, class W, class B, class O>
-void ConvDepthWiseImpl2D_cpu_forward_kernel(const ConvDepthWise_Op<2>::Parameters ¶ms, const std::array<DimSize_t, 4> &dims,
+void ConvDepthWiseImpl2D_cpu_forward_kernel(const ConvDepthWise_Op<2>::Attrs &attrs, const std::array<DimSize_t, 4> &dims,
const void *input_, const void *weights_, const void *biases_, void *output_) {
- // FIXME: missing convolution parameters as arguments
+ // FIXME: missing convolution attributes as arguments
const I *input = static_cast<const I *>(input_);
const W *weights = static_cast<const W *>(weights_);
const B *biases = static_cast<const B *>(biases_);
@@ -46,52 +46,52 @@ void ConvDepthWiseImpl2D_cpu_forward_kernel(const ConvDepthWise_Op<2>::Parameter
// output H size
const std::size_t oxSize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<4>(params)[0] + std::get<4>(params)[2] - std::get<3>(params)[0] + std::get<0>(params)[0]) /
- static_cast<float>(std::get<0>(params)[0])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<4>(attrs)[0] + std::get<4>(attrs)[2] - std::get<3>(attrs)[0] + std::get<0>(attrs)[0]) /
+ static_cast<float>(std::get<0>(attrs)[0])));
// output W size
const std::size_t oySize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<4>(params)[1] + std::get<4>(params)[3] - std::get<3>(params)[1] + std::get<0>(params)[1]) /
- static_cast<float>(std::get<0>(params)[1])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<4>(attrs)[1] + std::get<4>(attrs)[3] - std::get<3>(attrs)[1] + std::get<0>(attrs)[1]) /
+ static_cast<float>(std::get<0>(attrs)[1])));
// TODO: kernel computation
// output (batch, outCh, Xout, Yout)
// input (batch, ch, Xin, Yin)
// weight (outCh, ch, kernelX, kernelY)
- // does not take Dilation parameter into account
+ // does not take Dilation attribute into account
using signedsize = std::make_signed<std::size_t>::type;
for (std::size_t batch = 0; batch < dims[0]; ++batch) {
- for (std::size_t ch = 0; ch < std::get<2>(params); ++ch) {
- const std::size_t oIndex = (ch + batch*std::get<2>(params)) * oxSize * oySize;
+ for (std::size_t ch = 0; ch < std::get<2>(attrs); ++ch) {
+ const std::size_t oIndex = (ch + batch*std::get<2>(attrs)) * oxSize * oySize;
B biasVal = (biases != nullptr) ? biases[ch] : B(0);
std::fill(output + oIndex, output+(oIndex+oxSize*oySize), biasVal);
const std::size_t iIndex = (ch + batch*dims[1]) * dims[2] * dims[3];
- const std::size_t wIndex = ch * std::get<3>(params)[0] * std::get<3>(params)[1];
+ const std::size_t wIndex = ch * std::get<3>(attrs)[0] * std::get<3>(attrs)[1];
for (std::size_t ox = 0; ox < oxSize; ++ox) {
- const signedsize difx = static_cast<signedsize>(std::get<4>(params)[0] - ox * std::get<0>(params)[0]);
+ const signedsize difx = static_cast<signedsize>(std::get<4>(attrs)[0] - ox * std::get<0>(attrs)[0]);
const std::size_t sxMin = static_cast<std::size_t>(std::max(difx, signedsize(0)));
- const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<3>(params)[0] ? std::get<3>(params)[0] : dims[2] + difx);
+ const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<3>(attrs)[0] ? std::get<3>(attrs)[0] : dims[2] + difx);
for (std::size_t oy = 0; oy < oySize; ++oy) {
- const signedsize dify = static_cast<signedsize>(std::get<4>(params)[1] - oy * std::get<0>(params)[1]);
+ const signedsize dify = static_cast<signedsize>(std::get<4>(attrs)[1] - oy * std::get<0>(attrs)[1]);
const std::size_t syMin = static_cast<std::size_t>(std::max(dify, signedsize(0)));
- const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<3>(params)[1] ? std::get<3>(params)[1] : dims[3] + dify);
+ const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<3>(attrs)[1] ? std::get<3>(attrs)[1] : dims[3] + dify);
const std::size_t oIndexFull = oIndex + ox*oySize + oy;
- const signedsize ix = static_cast<signedsize>(ox * std::get<0>(params)[0]) - std::get<4>(params)[0];
- const signedsize iy = static_cast<signedsize>(oy * std::get<0>(params)[1]) - std::get<4>(params)[1];
+ const signedsize ix = static_cast<signedsize>(ox * std::get<0>(attrs)[0]) - std::get<4>(attrs)[0];
+ const signedsize iy = static_cast<signedsize>(oy * std::get<0>(attrs)[1]) - std::get<4>(attrs)[1];
if (sxMin == 0 && syMin == 0 && sxMax == 3 && syMax == 3) {
- output[oIndexFull] += (weights[wIndex + 0*std::get<3>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 0*std::get<3>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 0*std::get<3>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+2)] +
- weights[wIndex + 1*std::get<3>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 1*std::get<3>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 1*std::get<3>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+2)] +
- weights[wIndex + 2*std::get<3>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 2*std::get<3>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 2*std::get<3>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+2)]);
+ output[oIndexFull] += (weights[wIndex + 0*std::get<3>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 0*std::get<3>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 0*std::get<3>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+2)] +
+ weights[wIndex + 1*std::get<3>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 1*std::get<3>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 1*std::get<3>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+2)] +
+ weights[wIndex + 2*std::get<3>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 2*std::get<3>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 2*std::get<3>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+2)]);
} else {
for (std::size_t sx = sxMin; sx < sxMax; ++sx) {
for (std::size_t sy = syMin; sy < syMax; ++sy) {
- output[oIndexFull] += weights[wIndex + sx*std::get<3>(params)[1] + sy] *
+ output[oIndexFull] += weights[wIndex + sx*std::get<3>(attrs)[1] + sy] *
input[iIndex + static_cast<std::size_t>(ix+static_cast<signedsize>(sx))*dims[3] + static_cast<std::size_t>(iy+static_cast<signedsize>(sy))];
}
}
diff --git a/include/aidge/backend/cpu/operator/ConvImpl.hpp b/include/aidge/backend/cpu/operator/ConvImpl.hpp
index 1f3dffe43b966bc37887f267cc56760a899476f9..b9411fe0f1ac079d9857cc8f2178fc98fadc3a77 100644
--- a/include/aidge/backend/cpu/operator/ConvImpl.hpp
+++ b/include/aidge/backend/cpu/operator/ConvImpl.hpp
@@ -29,12 +29,12 @@ namespace Aidge {
class ConvImpl2DForward_cpu
: public Registrable<ConvImpl2DForward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const Conv_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *,
+ void(const Conv_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *,
const void *, const void *, void *)> {};
class ConvImpl2DBackward_cpu
: public Registrable<ConvImpl2DBackward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const Conv_Op<2>::Parameters &, const std::array<DimSize_t, 4> &, const void *,
+ void(const Conv_Op<2>::Attrs &, const std::array<DimSize_t, 4> &, const void *,
const void *, const void *, void *)> {};
class ConvImpl2D_cpu : public OperatorImpl {
diff --git a/include/aidge/backend/cpu/operator/ConvImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/ConvImpl_forward_kernels.hpp
index bc2f10099f42cba91be8d089b66dc176fdeb7c10..9d4d6dfdfcc114e47e478089c4d5a42c2bee0f28 100644
--- a/include/aidge/backend/cpu/operator/ConvImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/ConvImpl_forward_kernels.hpp
@@ -27,7 +27,7 @@ namespace Aidge {
* @tparam W Weight data type.
* @tparam B Bias data type.
* @tparam O Output data type.
- * @param params tuple of Parameters from the Operator
+ * @param params tuple of Attributes from the Operator
* @param dims Array of input dimensions.
* @param input_ const input Tensor.
* @param weights_ const weight Tensor.
@@ -35,9 +35,9 @@ namespace Aidge {
* @param output_ Output Tensor.
*/
template <class I, class W, class B, class O>
-void ConvImpl2D_cpu_forward_kernel(const Conv_Op<2>::Parameters ¶ms, const std::array<DimSize_t, 4> &dims,
+void ConvImpl2D_cpu_forward_kernel(const Conv_Op<2>::Attrs &attrs, const std::array<DimSize_t, 4> &dims,
const void *input_, const void *weights_, const void *biases_, void *output_) {
- // FIXME: missing convolution parameters as arguments
+ // FIXME: missing convolution attributes as arguments
const I *input = static_cast<const I *>(input_);
const W *weights = static_cast<const W *>(weights_);
const B *biases = static_cast<const B *>(biases_);
@@ -45,34 +45,34 @@ void ConvImpl2D_cpu_forward_kernel(const Conv_Op<2>::Parameters ¶ms, const s
/*
// output H size
const std::size_t oxSize =
- static_cast<std::size_t>(static_cast<float>(dims[0] - std::get<4>(params)[0] + std::get<0>(params)[0]) /
- static_cast<float>(std::get<0>(params)[0]));
+ static_cast<std::size_t>(static_cast<float>(dims[0] - std::get<4>(attrs)[0] + std::get<0>(attrs)[0]) /
+ static_cast<float>(std::get<0>(attrs)[0]));
// output W size
const std::size_t oySize =
- static_cast<std::size_t>(static_cast<float>(dims[1] - std::get<4>(params)[1] + std::get<0>(params)[1]) /
- static_cast<float>(std::get<0>(params)[1]));
+ static_cast<std::size_t>(static_cast<float>(dims[1] - std::get<4>(attrs)[1] + std::get<0>(attrs)[1]) /
+ static_cast<float>(std::get<0>(attrs)[1]));
// TODO: kernel computation
// output (Xout, Yout, outCh, batch)
// input (Xin, Yin, inCh, batch)
// weight (kernelX, kernelY, inCh, outCh)
- // does not take Dilation parameter into account
+ // does not take Dilation attribute into account
for (std::size_t ox = 0; ox < oxSize; ++ox) {
for (std::size_t oy = 0; oy < oySize; ++oy) {
- const std::size_t ix = ox * std::get<0>(params)[0];
- const std::size_t iy = oy * std::get<0>(params)[1];
+ const std::size_t ix = ox * std::get<0>(attrs)[0];
+ const std::size_t iy = oy * std::get<0>(attrs)[1];
- for (std::size_t outCh = 0; outCh < std::get<3>(params); ++outCh) {
- const std::size_t oIndex = dims[3] * (outCh + std::get<3>(params) * (oy + oySize * ox));
+ for (std::size_t outCh = 0; outCh < std::get<3>(attrs); ++outCh) {
+ const std::size_t oIndex = dims[3] * (outCh + std::get<3>(attrs) * (oy + oySize * ox));
B biasVal = (biases != nullptr) ? biases[outCh] : B(0);
for (std::size_t batch = 0; batch < dims[3]; ++batch) {
output[oIndex + batch] = biasVal;
}
for (std::size_t inCh = 0; inCh < dims[2]; ++inCh) {
- for (std::size_t sx = 0; sx < std::get<4>(params)[0]; ++sx) {
- for (std::size_t sy = 0; sy < std::get<4>(params)[1]; ++sy) {
+ for (std::size_t sx = 0; sx < std::get<4>(attrs)[0]; ++sx) {
+ for (std::size_t sy = 0; sy < std::get<4>(attrs)[1]; ++sy) {
const std::size_t wIndex =
- outCh + std::get<3>(params) * (inCh + dims[2] * (sy + std::get<4>(params)[1] * sx));
+ outCh + std::get<3>(attrs) * (inCh + dims[2] * (sy + std::get<4>(attrs)[1] * sx));
std::size_t iIndex = dims[3] * (inCh + dims[2] * ((iy + sy) + dims[1] * (ix + sx)));
for (std::size_t batch = 0; batch < dims[3]; ++batch) {
output[oIndex + batch] += weights[wIndex] * input[iIndex + batch];
@@ -88,53 +88,53 @@ void ConvImpl2D_cpu_forward_kernel(const Conv_Op<2>::Parameters ¶ms, const s
// output H size
const std::size_t oxSize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<5>(params)[0] + std::get<5>(params)[2] - std::get<4>(params)[0] + std::get<0>(params)[0]) /
- static_cast<float>(std::get<0>(params)[0])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[2] + std::get<5>(attrs)[0] + std::get<5>(attrs)[2] - std::get<4>(attrs)[0] + std::get<0>(attrs)[0]) /
+ static_cast<float>(std::get<0>(attrs)[0])));
// output W size
const std::size_t oySize =
- static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<5>(params)[1] + std::get<5>(params)[3] - std::get<4>(params)[1] + std::get<0>(params)[1]) /
- static_cast<float>(std::get<0>(params)[1])));
+ static_cast<std::size_t>(std::floor(static_cast<float>(dims[3] + std::get<5>(attrs)[1] + std::get<5>(attrs)[3] - std::get<4>(attrs)[1] + std::get<0>(attrs)[1]) /
+ static_cast<float>(std::get<0>(attrs)[1])));
// TODO: kernel computation
// output (batch, outCh, Xout, Yout)
// input (batch, inCh, Xin, Yin)
// weight (outCh, inCh, kernelX, kernelY)
- // does not take Dilation parameter into account
+ // does not take Dilation attribute into account
using signedsize = std::make_signed<std::size_t>::type;
for (std::size_t batch = 0; batch < dims[0]; ++batch) {
- for (std::size_t outCh = 0; outCh < std::get<3>(params); ++outCh) {
- const std::size_t oIndex = (outCh + batch*std::get<3>(params)) * oxSize * oySize;
+ for (std::size_t outCh = 0; outCh < std::get<3>(attrs); ++outCh) {
+ const std::size_t oIndex = (outCh + batch*std::get<3>(attrs)) * oxSize * oySize;
B biasVal = (biases != nullptr) ? biases[outCh] : B(0);
std::fill(output + oIndex, output+(oIndex+oxSize*oySize), biasVal);
for (std::size_t inCh = 0; inCh < dims[1]; ++inCh) {
const std::size_t iIndex = (inCh + batch*dims[1]) * dims[2] * dims[3];
- const std::size_t wIndex = (inCh + outCh*dims[1]) * std::get<4>(params)[0] * std::get<4>(params)[1];
+ const std::size_t wIndex = (inCh + outCh*dims[1]) * std::get<4>(attrs)[0] * std::get<4>(attrs)[1];
for (std::size_t ox = 0; ox < oxSize; ++ox) {
- const signedsize difx = static_cast<signedsize>(std::get<5>(params)[0] - ox * std::get<0>(params)[0]);
+ const signedsize difx = static_cast<signedsize>(std::get<5>(attrs)[0] - ox * std::get<0>(attrs)[0]);
const std::size_t sxMin = static_cast<std::size_t>(std::max(difx, signedsize(0)));
- const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<4>(params)[0] ? std::get<4>(params)[0] : dims[2] + difx);
+ const std::size_t sxMax = (static_cast<signedsize>(dims[2]) + difx) < 0 ? 0 : ((dims[2] + difx) > std::get<4>(attrs)[0] ? std::get<4>(attrs)[0] : dims[2] + difx);
for (std::size_t oy = 0; oy < oySize; ++oy) {
- const signedsize dify = static_cast<signedsize>(std::get<5>(params)[1] - oy * std::get<0>(params)[1]);
+ const signedsize dify = static_cast<signedsize>(std::get<5>(attrs)[1] - oy * std::get<0>(attrs)[1]);
const std::size_t syMin = static_cast<std::size_t>(std::max(dify, signedsize(0)));
- const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<4>(params)[1] ? std::get<4>(params)[1] : dims[3] + dify);
+ const std::size_t syMax = (static_cast<signedsize>(dims[3]) + dify) < 0 ? 0 : ((dims[3] + dify) > std::get<4>(attrs)[1] ? std::get<4>(attrs)[1] : dims[3] + dify);
const std::size_t oIndexFull = oIndex + ox*oySize + oy;
- const signedsize ix = static_cast<signedsize>(ox * std::get<0>(params)[0]) - std::get<5>(params)[0];
- const signedsize iy = static_cast<signedsize>(oy * std::get<0>(params)[1]) - std::get<5>(params)[1];
+ const signedsize ix = static_cast<signedsize>(ox * std::get<0>(attrs)[0]) - std::get<5>(attrs)[0];
+ const signedsize iy = static_cast<signedsize>(oy * std::get<0>(attrs)[1]) - std::get<5>(attrs)[1];
if (sxMin == 0 && syMin == 0 && sxMax == 3 && syMax == 3) {
- output[oIndexFull] += (weights[wIndex + 0*std::get<4>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 0*std::get<4>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 0*std::get<4>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+2)] +
- weights[wIndex + 1*std::get<4>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 1*std::get<4>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 1*std::get<4>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+2)] +
- weights[wIndex + 2*std::get<4>(params)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+0)] +
- weights[wIndex + 2*std::get<4>(params)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+1)] +
- weights[wIndex + 2*std::get<4>(params)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+2)]);
+ output[oIndexFull] += (weights[wIndex + 0*std::get<4>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 0*std::get<4>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 0*std::get<4>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+0)*dims[3] + static_cast<std::size_t>(iy+2)] +
+ weights[wIndex + 1*std::get<4>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 1*std::get<4>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 1*std::get<4>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+1)*dims[3] + static_cast<std::size_t>(iy+2)] +
+ weights[wIndex + 2*std::get<4>(attrs)[1] + 0] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+0)] +
+ weights[wIndex + 2*std::get<4>(attrs)[1] + 1] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+1)] +
+ weights[wIndex + 2*std::get<4>(attrs)[1] + 2] * input[iIndex + static_cast<std::size_t>(ix+2)*dims[3] + static_cast<std::size_t>(iy+2)]);
} else {
for (std::size_t sx = sxMin; sx < sxMax; ++sx) {
for (std::size_t sy = syMin; sy < syMax; ++sy) {
- output[oIndexFull] += weights[wIndex + sx*std::get<4>(params)[1] + sy] *
+ output[oIndexFull] += weights[wIndex + sx*std::get<4>(attrs)[1] + sy] *
input[iIndex + static_cast<std::size_t>(ix+static_cast<signedsize>(sx))*dims[3] + static_cast<std::size_t>(iy+static_cast<signedsize>(sy))];
}
}
diff --git a/include/aidge/backend/cpu/operator/FCImpl.hpp b/include/aidge/backend/cpu/operator/FCImpl.hpp
index c69cc0b08a58877108c78d6f12c29e9089c2f665..1dfa40439dbba9cdd4fe3436fea30f771678c1ff 100644
--- a/include/aidge/backend/cpu/operator/FCImpl.hpp
+++ b/include/aidge/backend/cpu/operator/FCImpl.hpp
@@ -26,11 +26,11 @@ namespace Aidge {
// compute kernel registry for forward and backward
class FCImplForward_cpu : public Registrable<FCImplForward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const FC_Op::Parameters &, const DimSize_t, const DimSize_t,
+ void(const FC_Op::Attrs &, const DimSize_t, const DimSize_t,
const void *, const void *, const void *, void *)> {};
class FCImplBackward_cpu : public Registrable<FCImplBackward_cpu,
std::tuple<DataType, DataType, DataType, DataType>,
- void(const FC_Op::Parameters &, const DimSize_t, const DimSize_t,
+ void(const FC_Op::Attrs &, const DimSize_t, const DimSize_t,
const void *, const void *, const void *, void *)> {};
class FCImpl_cpu : public OperatorImpl {
diff --git a/include/aidge/backend/cpu/operator/FCImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/FCImpl_forward_kernels.hpp
index d6acb7dfea3415a8d67384745e16ecdd8bf06324..91e2558a7ef1079cbc9fb11f78fab53ef4246149 100644
--- a/include/aidge/backend/cpu/operator/FCImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/FCImpl_forward_kernels.hpp
@@ -19,17 +19,17 @@
namespace Aidge {
// template <class I, class W, class B, class O>
-// void FCImpl_cpu_forward_kernel(const FC_Op::Parameters& params, const std::array<DimSize_t, 4>& dims,
+// void FCImpl_cpu_forward_kernel(const FC_Op::Attrs& attrs, const std::array<DimSize_t, 4>& dims,
// const void* input_, const void* weights_, const void* biases_, void* output_) {
-// // FIXME: missing FC parameters as arguments
+// // FIXME: missing FC attributes as arguments
// const I* input = static_cast<const I*>(input_);
// const W* weights = static_cast<const W*>(weights_);
// const B* biases = static_cast<const B*>(biases_);
// O* output = static_cast<O*>(output_);
-// for (std::size_t outIdx = 0; outIdx < std::get<0>(params); ++outIdx) {
+// for (std::size_t outIdx = 0; outIdx < std::get<0>(attrs); ++outIdx) {
// std::size_t oIndex = outIdx * dims[3];
-// const B bias = std::get<1>(params) ? B(0) : biases[outIdx];
+// const B bias = std::get<1>(attrs) ? B(0) : biases[outIdx];
// for (std::size_t batch = 0; batch < dims[3]; ++batch) {
// output[oIndex + batch] = bias;
// }
@@ -39,10 +39,10 @@ namespace Aidge {
// for (std::size_t iy = 0; iy < dims[1]; ++iy) {
// for (std::size_t inCh = 0; inCh < dims[2]; ++inCh) {
// const std::size_t iIndex = dims[3] * (inCh + dims[2] * (iy + dims[1] * ix));
-// for (std::size_t outCh = 0; outCh < std::get<0>(params); ++outCh) {
+// for (std::size_t outCh = 0; outCh < std::get<0>(attrs); ++outCh) {
// const std::size_t oIndex = dims[3] * outCh;
-// const std::size_t wIndex = (inCh + dims[2] * (iy + dims[1] * ix)) * std::get<0>(params) +
-// outCh; // (iIndex*std::get<0>(params) + oIndex)/dims[3];
+// const std::size_t wIndex = (inCh + dims[2] * (iy + dims[1] * ix)) * std::get<0>(attrs) +
+// outCh; // (iIndex*std::get<0>(attrs) + oIndex)/dims[3];
// for (std::size_t batch = 0; batch < dims[3]; ++batch) {
// output[oIndex + batch] += weights[wIndex] * input[iIndex + batch];
// }
@@ -53,9 +53,9 @@ namespace Aidge {
// }
// template <class I, class W, class B, class O>
-// void FCImpl_cpu_forward_kernel(const FC_Op::Parameters& params, const std::array<DimSize_t, 2>& dims,
+// void FCImpl_cpu_forward_kernel(const FC_Op::Attrs& attrs, const std::array<DimSize_t, 2>& dims,
// const void* input_, const void* weights_, const void* biases_, void* output_) {
-// // FIXME: missing FC parameters as arguments
+// // FIXME: missing FC attributes as arguments
// const I* input = static_cast<const I*>(input_);
// const W* weights = static_cast<const W*>(weights_);
// const B* biases = static_cast<const B*>(biases_);
@@ -63,9 +63,9 @@ namespace Aidge {
// // let's have I.dims() = [N, C, H, W] instead of [H, W, C, N]
-// for (std::size_t outIdx = 0; outIdx < std::get<0>(params); ++outIdx) {
+// for (std::size_t outIdx = 0; outIdx < std::get<0>(attrs); ++outIdx) {
// std::size_t oIndex = outIdx * dims[0];
-// const B bias = std::get<1>(params) ? B(0) : biases[outIdx];
+// const B bias = std::get<1>(attrs) ? B(0) : biases[outIdx];
// for (std::size_t batch = 0; batch < dims[0]; ++batch) {
// output[oIndex + batch] = bias;
// }
@@ -74,8 +74,8 @@ namespace Aidge {
// for (std::size_t batch = 0; batch < dims[0]; ++batch) {
// const std::size_t oIndex = dims[1] * batch;
// for (std::size_t i = 0; i < dims[1]; ++i) {
-// for (std::size_t outCh = 0; outCh < std::get<0>(params); ++outCh) {
-// std::size_t wIndex = i * std::get<0>(params) + outCh; // (iIndex*std::get<0>(params) + oIndex)/dims[3];
+// for (std::size_t outCh = 0; outCh < std::get<0>(attrs); ++outCh) {
+// std::size_t wIndex = i * std::get<0>(attrs) + outCh; // (iIndex*std::get<0>(attrs) + oIndex)/dims[3];
// output[oIndex + outCh] += weights[wIndex] * input[i + batch];
// }
// }
@@ -83,29 +83,29 @@ namespace Aidge {
// }
template <class I, class W, class B, class O>
-void FCImpl_cpu_forward_kernel(const FC_Op::Parameters& params, const DimSize_t batchSize, const DimSize_t oneInputSize,
+void FCImpl_cpu_forward_kernel(const FC_Op::Attrs& attrs, const DimSize_t batchSize, const DimSize_t oneInputSize,
const void* input_, const void* weights_, const void* biases_, void* output_) {
- // FIXME: missing FC parameters as arguments
+ // FIXME: missing FC attributes as arguments
const I* input = static_cast<const I*>(input_);
const W* weights = static_cast<const W*>(weights_);
const B* biases = static_cast<const B*>(biases_);
O* output = static_cast<O*>(output_);
- if (std::get<1>(params)) {
- std::fill(output, output+(batchSize*std::get<0>(params)), B(0));
+ if (std::get<1>(attrs)) {
+ std::fill(output, output+(batchSize*std::get<0>(attrs)), B(0));
}
else {
for (std::size_t batch = 0; batch < batchSize; ++batch) {
- std::copy(biases, biases+std::get<0>(params), output+(batch*std::get<0>(params)));
+ std::copy(biases, biases+std::get<0>(attrs), output+(batch*std::get<0>(attrs)));
}
}
for (std::size_t batch = 0; batch < batchSize; ++batch) {
- for (std::size_t out = 0; out < std::get<0>(params); ++out) {
- output[out + batch*std::get<0>(params)] = std::inner_product(input + batch*oneInputSize,
+ for (std::size_t out = 0; out < std::get<0>(attrs); ++out) {
+ output[out + batch*std::get<0>(attrs)] = std::inner_product(input + batch*oneInputSize,
input + (batch + 1)*oneInputSize,
weights + out*oneInputSize,
- output[out + batch*std::get<0>(params)]);
+ output[out + batch*std::get<0>(attrs)]);
}
}
}
diff --git a/include/aidge/backend/cpu/operator/LeakyReLUImpl.hpp b/include/aidge/backend/cpu/operator/LeakyReLUImpl.hpp
index abe167bea16de01f861beb9701f747d39f265d9d..386ef999fddbda184edee88723d213f53ff62ded 100644
--- a/include/aidge/backend/cpu/operator/LeakyReLUImpl.hpp
+++ b/include/aidge/backend/cpu/operator/LeakyReLUImpl.hpp
@@ -24,10 +24,10 @@ namespace Aidge {
// compute kernel registry for forward and backward
class LeakyReLUImplForward_cpu
- : public Registrable<LeakyReLUImplForward_cpu, std::tuple<DataType, DataType>, void(const LeakyReLU_Op::Parameters&, std::size_t, const void*, void*)> {
+ : public Registrable<LeakyReLUImplForward_cpu, std::tuple<DataType, DataType>, void(const LeakyReLU_Op::Attrs&, std::size_t, const void*, void*)> {
};
class LeakyReLUImplBackward_cpu
- : public Registrable<LeakyReLUImplBackward_cpu, std::tuple<DataType, DataType>, void(const LeakyReLU_Op::Parameters&, std::size_t, const void*, void*)> {
+ : public Registrable<LeakyReLUImplBackward_cpu, std::tuple<DataType, DataType>, void(const LeakyReLU_Op::Attrs&, std::size_t, const void*, void*)> {
};
class LeakyReLUImpl_cpu : public OperatorImpl {
diff --git a/include/aidge/backend/cpu/operator/LeakyReLUImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/LeakyReLUImpl_forward_kernels.hpp
index ff9a8ac6a8f968f244429b330401d794f16fac01..761b9579c3c3dc187e4b0fac24812fa77f916e65 100644
--- a/include/aidge/backend/cpu/operator/LeakyReLUImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/LeakyReLUImpl_forward_kernels.hpp
@@ -18,14 +18,14 @@
namespace Aidge {
template <class I, class O>
-void LeakyReLUImpl_cpu_forward_kernel(const LeakyReLU_Op::Parameters& params,
+void LeakyReLUImpl_cpu_forward_kernel(const LeakyReLU_Op::Attrs& attrs,
std::size_t inputLenght,
const void* input_,
void* output_) {
const I* input = static_cast<const I*>(input_);
O* output = static_cast<O*>(output_);
- I negativeSlope = static_cast<I>(std::get<0>(params));
+ I negativeSlope = static_cast<I>(std::get<0>(attrs));
for (std::size_t i = 0; i < inputLenght; ++i) {
output[i] = input[i] >= 0 ? input[i] : input[i] * negativeSlope;
diff --git a/include/aidge/backend/cpu/operator/MatMulImpl.hpp b/include/aidge/backend/cpu/operator/MatMulImpl.hpp
new file mode 100644
index 0000000000000000000000000000000000000000..bf8e31efd253ee8855f3473ef0b4a60c59a04b5f
--- /dev/null
+++ b/include/aidge/backend/cpu/operator/MatMulImpl.hpp
@@ -0,0 +1,76 @@
+/********************************************************************************
+ * 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
+ *
+ ********************************************************************************/
+
+#ifndef AIDGE_CPU_OPERATOR_MATMULIMPL_H_
+#define AIDGE_CPU_OPERATOR_MATMULIMPL_H_
+
+#include <array>
+#include <memory>
+#include <vector>
+
+#include "aidge/backend/OperatorImpl.hpp"
+#include "aidge/operator/MatMul.hpp"
+#include "aidge/utils/Registrar.hpp"
+#include "aidge/utils/Types.h"
+
+namespace Aidge {
+// class MatMul_Op;
+
+// compute kernel registry for forward and backward
+class MatMulImplForward_cpu
+ : public Registrable<MatMulImplForward_cpu, std::tuple<DataType, DataType, DataType>,
+ void(const MatMul_Op::Attrs &, const DimSize_t, const DimSize_t,
+ const void *, const void *, void *)> {};
+class MatMulImplBackward_cpu
+ : public Registrable<MatMulImplBackward_cpu, std::tuple<DataType, DataType, DataType>,
+ void(const MatMul_Op::Attrs &, const DimSize_t, const DimSize_t,
+ const void *, const void *, void *)> {};
+
+class MatMulImpl_cpu : public OperatorImpl {
+private:
+ const MatMul_Op &mOp;
+ std::array<NbElts_t, 2> mNbConsumedData;
+ std::array<NbElts_t, 1> mNbProducedData;
+
+public:
+ MatMulImpl_cpu(const MatMul_Op &op)
+ : mOp(op),
+ mNbConsumedData({0, 0}),
+ mNbProducedData({0})
+ {
+ // ctor
+ }
+
+ static std::unique_ptr<MatMulImpl_cpu> create(const MatMul_Op &op)
+ {
+ return std::make_unique<MatMulImpl_cpu>(op);
+ }
+
+public:
+ NbElts_t getNbRequiredData(const IOIndex_t inputIdx) const override final;
+ NbElts_t getNbRequiredProtected(const IOIndex_t inputIdx) const override final;
+ NbElts_t getRequiredMemory(const IOIndex_t /*outputIdx*/,
+ const std::vector<DimSize_t> & /*inputsSize*/) const override final;
+ NbElts_t getNbConsumedData(const IOIndex_t inputIdx) const override final;
+ NbElts_t getNbProducedData(const IOIndex_t outputIdx) const override final;
+
+ void updateConsummerProducer() override final;
+
+ void forward();
+ void backward();
+};
+
+namespace {
+static Registrar<MatMul_Op> registrarMatMulImpl_cpu("cpu", Aidge::MatMulImpl_cpu::create);
+}
+} // namespace Aidge
+
+#endif /* AIDGE_CPU_OPERATOR_MATMULIMPL_H_ */
diff --git a/include/aidge/backend/cpu/operator/MatMulImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/MatMulImpl_forward_kernels.hpp
new file mode 100644
index 0000000000000000000000000000000000000000..bc52779eff274379a853ea84fb839c9486652433
--- /dev/null
+++ b/include/aidge/backend/cpu/operator/MatMulImpl_forward_kernels.hpp
@@ -0,0 +1,58 @@
+/********************************************************************************
+ * 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
+ *
+ ********************************************************************************/
+
+#ifndef AIDGE_CPU_OPERATOR_MATMULIMPL_FORWARD_KERNEL_H_
+#define AIDGE_CPU_OPERATOR_MATMULIMPL_FORWARD_KERNEL_H_
+
+#include "aidge/utils/Registrar.hpp"
+#include <algorithm>
+
+#include "aidge/backend/cpu/operator/MatMulImpl.hpp"
+
+namespace Aidge {
+
+template <class I, class W, class O>
+void MatMulImpl_cpu_forward_kernel(const MatMul_Op::Attrs& attrs, const DimSize_t batchSize, const DimSize_t oneInputSize,
+ const void* input_, const void* weights_, void* output_) {
+ // FIXME: missing MatMul parameters as arguments
+ const I* input = static_cast<const I*>(input_);
+ const W* weights = static_cast<const W*>(weights_);
+ O* output = static_cast<O*>(output_);
+
+
+ std::fill(output, output+(batchSize*std::get<0>(attrs)), O(0));
+
+ for (std::size_t batch = 0; batch < batchSize; ++batch) {
+ for (std::size_t out = 0; out < std::get<0>(attrs); ++out) {
+ output[out + batch*std::get<0>(attrs)] = std::inner_product(input + batch*oneInputSize,
+ input + (batch + 1)*oneInputSize,
+ weights + out*oneInputSize,
+ output[out + batch*std::get<0>(attrs)]);
+ }
+ }
+}
+
+
+namespace {
+static Registrar<MatMulImplForward_cpu> registrarMatMulImpl2DForward_cpu_Float32(
+ {DataType::Float32, DataType::Float32, DataType::Float32},
+ Aidge::MatMulImpl_cpu_forward_kernel<float, float, float>);
+static Registrar<MatMulImplForward_cpu> registrarMatMulImpl2DForward_cpu_Int32(
+ {DataType::Int32, DataType::Int32, DataType::Int32},
+ Aidge::MatMulImpl_cpu_forward_kernel<int, int, int>);
+static Registrar<MatMulImplForward_cpu> registrarMatMulImpl2DForward_cpu_Float64(
+ {DataType::Float64, DataType::Float64, DataType::Float64},
+ Aidge::MatMulImpl_cpu_forward_kernel<double, double, double>);
+} // namespace
+
+} // namespace Aidge
+
+#endif /* AIDGE_CPU_OPERATOR_MATMULIMPL_FORWARD_KERNEL_H_ */
diff --git a/include/aidge/backend/cpu/operator/ScalingImpl.hpp b/include/aidge/backend/cpu/operator/ScalingImpl.hpp
index cb11b3016b9f694cc518f20a62ea143a94a58afe..37549349b9f5ffbf443d976135db05b4cec209b7 100644
--- a/include/aidge/backend/cpu/operator/ScalingImpl.hpp
+++ b/include/aidge/backend/cpu/operator/ScalingImpl.hpp
@@ -18,16 +18,17 @@
#include "aidge/utils/Types.h"
#include <memory>
#include <vector>
+#include <array>
namespace Aidge {
// class Scaling_Op;
// compute kernel registry for forward and backward
class ScalingImplForward_cpu
- : public Registrable<ScalingImplForward_cpu, std::tuple<DataType, DataType>, void(const Scaling_Op::Parameters&, std::size_t, const void*, void*)> {
+ : public Registrable<ScalingImplForward_cpu, std::tuple<DataType, DataType>, void(const Scaling_Op::Attrs&, std::size_t, const void*, void*)> {
};
class ScalingImplBackward_cpu
- : public Registrable<ScalingImplBackward_cpu, std::tuple<DataType, DataType>, void(const Scaling_Op::Parameters&, std::size_t, const void*, void*)> {
+ : public Registrable<ScalingImplBackward_cpu, std::tuple<DataType, DataType>, void(const Scaling_Op::Attrs&, std::size_t, const void*, void*)> {
};
class ScalingImpl_cpu : public OperatorImpl {
@@ -47,7 +48,7 @@ class ScalingImpl_cpu : public OperatorImpl {
public:
NbElts_t getNbRequiredData(const IOIndex_t inputIdx) const override final;
NbElts_t getNbRequiredProtected(const IOIndex_t inputIdx) const override final;
- NbElts_t getRequiredMemory(__attribute__((unused)) const IOIndex_t outputIdx, __attribute__((unused)) const std::vector<DimSize_t>& inputsSize) const override final;
+ NbElts_t getRequiredMemory(const IOIndex_t outputIdx, const std::vector<DimSize_t>& inputsSize) const override final;
NbElts_t getNbConsumedData(const IOIndex_t inputIdx) const override final;
NbElts_t getNbProducedData(const IOIndex_t outputIdx) const override final;
diff --git a/include/aidge/backend/cpu/operator/ScalingImpl_forward_kernels.hpp b/include/aidge/backend/cpu/operator/ScalingImpl_forward_kernels.hpp
index c5b06290ee04ecf9759f418cd26d83e889fcc84e..8fe13bce3a4c470d77b083603d3b889a46fda71f 100644
--- a/include/aidge/backend/cpu/operator/ScalingImpl_forward_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/ScalingImpl_forward_kernels.hpp
@@ -18,14 +18,14 @@
namespace Aidge {
template <class I, class O>
-void ScalingImpl_cpu_forward_kernel(const Scaling_Op::Parameters& params,
+void ScalingImpl_cpu_forward_kernel(const Scaling_Op::Attrs& attrs,
std::size_t inputLenght,
const void* input_,
void* output_) {
const I* input = static_cast<const I*>(input_);
O* output = static_cast<O*>(output_);
- I scalingFactor = static_cast<I>(std::get<0>(params));
+ const I& scalingFactor = static_cast<const I&>(std::get<0>(attrs));
for (std::size_t i = 0; i < inputLenght; ++i) {
output[i] = input[i] * scalingFactor;
diff --git a/setup.py b/setup.py
index 0b0f66e9132d66cdb6385d7f8c6c69ae0cc5d0e3..16305afdfdfa5de2e328460d9e96c77eb96a9d98 100644
--- a/setup.py
+++ b/setup.py
@@ -62,11 +62,11 @@ class CMakeBuild(build_ext):
os.chdir(str(build_temp))
- # Impose to use the executable of the python
+ # Impose to use the executable of the python
# used to launch setup.py to setup PythonInterp
param_py = "-DPYTHON_EXECUTABLE=" + sys.executable
-
- install_path = f"{build_temp}/install" if "AIDGE_INSTALL" not in os.environ else os.environ["AIDGE_INSTALL"]
+
+ install_path = os.path.join(sys.prefix, "lib", "libAidge") if "AIDGE_INSTALL" not in os.environ else os.environ["AIDGE_INSTALL"]
self.spawn(['cmake', str(cwd), param_py, '-DTEST=OFF', f'-DCMAKE_INSTALL_PREFIX:PATH={install_path}'])
if not self.dry_run:
@@ -83,11 +83,11 @@ class CMakeBuild(build_ext):
for file in files:
if file.endswith('.so') and (root != str(aidge_package.absolute())):
currentFile=os.path.join(root, file)
- shutil.copy(currentFile, str(aidge_package.absolute()))
+ shutil.copy(currentFile, str(aidge_package.absolute()))
# Copy version.txt in aidge_package
os.chdir(os.path.dirname(__file__))
- shutil.copy("version.txt", str(aidge_package.absolute()))
+ shutil.copy("version.txt", str(aidge_package.absolute()))
if __name__ == '__main__':
diff --git a/src/operator/AddImpl.cpp b/src/operator/AddImpl.cpp
index d3da42185237a59146af17199e34a00dbebd6d96..be7923339308073c26b60ee0349a44037769765a 100644
--- a/src/operator/AddImpl.cpp
+++ b/src/operator/AddImpl.cpp
@@ -99,6 +99,7 @@ Aidge::NbElts_t Aidge::AddImpl_cpu<2>::getNbRequiredProtected(const Aidge::IOInd
Aidge::NbElts_t Aidge::AddImpl_cpu<2>::getRequiredMemory(const Aidge::IOIndex_t outputIdx, const std::vector<Aidge::DimSize_t>& /*inputsSize*/) const {
// Requires the whole tensors, regardless of available data on inputs
assert(outputIdx == 0 && "operator has only one output");
+ (void) outputIdx; // avoid unused warning
const auto& outputDims = std::static_pointer_cast<Tensor>(mOp.getOutput(0))->dims();
return std::accumulate(outputDims.begin(), outputDims.end(),
@@ -167,6 +168,7 @@ Aidge::NbElts_t Aidge::AddImpl_cpu<3>::getNbRequiredProtected(const Aidge::IOInd
Aidge::NbElts_t Aidge::AddImpl_cpu<3>::getRequiredMemory(const Aidge::IOIndex_t outputIdx, const std::vector<Aidge::DimSize_t>& /*inputsSize*/) const {
// Requires the whole tensors, regardless of available data on inputs
assert(outputIdx == 0 && "operator has only one output");
+ (void) outputIdx; // avoid unused warning
const auto& outputDims = std::static_pointer_cast<Tensor>(mOp.getOutput(0))->dims();
return std::accumulate(outputDims.begin(), outputDims.end(),
diff --git a/src/operator/AvgPoolingImpl.cpp b/src/operator/AvgPoolingImpl.cpp
index 6c434a5c38853a1dee66db5be95b6b1bfdde8162..b1f82bbb4323a402d698d772966409e1a8f7224b 100644
--- a/src/operator/AvgPoolingImpl.cpp
+++ b/src/operator/AvgPoolingImpl.cpp
@@ -70,7 +70,7 @@ void Aidge::AvgPoolingImpl2D_cpu::forward() {
Registrar<AvgPoolingImpl2DForward_cpu>::create({mOp.getInput(0)->dataType(), mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(),
+ kernelFunc(mOp.getStaticAttributes(),
mOp.getInput(0)->dims<4>(),
mOp.getInput(0)->getImpl()->rawPtr(),
mOp.getOutput(0)->getImpl()->rawPtr());
diff --git a/src/operator/BatchNormImpl.cpp b/src/operator/BatchNormImpl.cpp
index a0d4d032ded9ede1b2dba307aa967af330167d25..90ee2b7a2361166109568e317a1788137150a8d1 100644
--- a/src/operator/BatchNormImpl.cpp
+++ b/src/operator/BatchNormImpl.cpp
@@ -76,7 +76,7 @@ void Aidge::BatchNormImpl2D_cpu::forward() {
mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(),
+ kernelFunc(mOp.getStaticAttributes(),
mOp.getInput(0)->dims<4>(),
mOp.getInput(0)->getImpl()->rawPtr(),
mOp.getInput(1)->getImpl()->rawPtr(),
diff --git a/src/operator/ConvDepthWiseImpl.cpp b/src/operator/ConvDepthWiseImpl.cpp
index 3e920cf68366b82bce8df29c8aea0c838e6a1364..7801f64ef46ced22d95af47b8b0e8cc9888a81da 100644
--- a/src/operator/ConvDepthWiseImpl.cpp
+++ b/src/operator/ConvDepthWiseImpl.cpp
@@ -77,7 +77,7 @@ void Aidge::ConvDepthWiseImpl2D_cpu::forward() {
mOp.getInput(2)->dataType(), mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(), std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
+ kernelFunc(mOp.getStaticAttributes(), std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
mOp.getInput(0)->getImpl()->rawPtr(), mOp.getInput(1)->getImpl()->rawPtr(),
mOp.getInput(2)->getImpl()->rawPtr(), mOp.getOutput(0)->getImpl()->rawPtr());
}
diff --git a/src/operator/ConvImpl.cpp b/src/operator/ConvImpl.cpp
index b4ddf80929923a9c2c5998ac8614ebb0d3afe000..edab4432fd5792f27ea158f265641855532d6d0b 100644
--- a/src/operator/ConvImpl.cpp
+++ b/src/operator/ConvImpl.cpp
@@ -75,7 +75,7 @@ void Aidge::ConvImpl2D_cpu::forward() {
mOp.getInput(2)->dataType(), mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(), std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
+ kernelFunc(mOp.getStaticAttributes(), std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
mOp.getInput(0)->getImpl()->rawPtr(), mOp.getInput(1)->getImpl()->rawPtr(),
mOp.getInput(2)->getImpl()->rawPtr(), mOp.getOutput(0)->getImpl()->rawPtr());
diff --git a/src/operator/FCImpl.cpp b/src/operator/FCImpl.cpp
index 086902be0ab1c2027a8c62c143bc27921e5e9e1b..3cf1ccf6e951ea05521ef67c99a3e628e0f620f5 100644
--- a/src/operator/FCImpl.cpp
+++ b/src/operator/FCImpl.cpp
@@ -98,7 +98,7 @@ void Aidge::FCImpl_cpu::forward()
// Call kernel
// if (mOp.getInput(0)->nbDims() == 4) {
// kernelFunc(
- // mOp.getParams(),
+ // mOp.getStaticAttributes(),
// std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
// mOp.getInput(0)->getImpl()->rawPtr(),
// mOp.mInputs[1]->getImpl()->rawPtr(),
@@ -107,7 +107,7 @@ void Aidge::FCImpl_cpu::forward()
// }
// else
kernelFunc(
- mOp.getParams(),
+ mOp.getStaticAttributes(),
mOp.getInput(0)->dims()[0],
mOp.getInput(0)->sizeM1(),
mOp.getInput(0)->getImpl()->rawPtr(),
diff --git a/src/operator/LeakyReLUImpl.cpp b/src/operator/LeakyReLUImpl.cpp
index f6a44d381081c7c7f1dcbbf02d91212168cc07aa..316d3641bb960ed8850a94f40186b77cc8522b58 100644
--- a/src/operator/LeakyReLUImpl.cpp
+++ b/src/operator/LeakyReLUImpl.cpp
@@ -65,7 +65,7 @@ void Aidge::LeakyReLUImpl_cpu::forward() {
mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(),
+ kernelFunc(mOp.getStaticAttributes(),
std::static_pointer_cast<Tensor>(mOp.getInput(0))->size(),
mOp.getInput(0)->getImpl()->rawPtr(),
mOp.getOutput(0)->getImpl()->rawPtr());
diff --git a/src/operator/MatMulImpl.cpp b/src/operator/MatMulImpl.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..972e1f0fc87ad00afe670d77afc8617137076a08
--- /dev/null
+++ b/src/operator/MatMulImpl.cpp
@@ -0,0 +1,121 @@
+/********************************************************************************
+ * 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 <cassert>
+#include <chrono> // std::chrono::milliseconds
+#include <numeric> // std::accumulate
+#include <thread> // std::this_thread::sleep_for
+#include <vector>
+
+#include "aidge/operator/MatMul.hpp"
+#include "aidge/utils/Types.h"
+
+#include "aidge/backend/cpu/operator/MatMulImpl.hpp"
+#include "aidge/backend/cpu/operator/MatMulImpl_forward_kernels.hpp"
+
+Aidge::NbElts_t Aidge::MatMulImpl_cpu::getNbRequiredData(const Aidge::IOIndex_t inputIdx) const
+{
+ assert(mOp.getInput(inputIdx) && "requires valid input");
+
+ // Requires the whole tensors
+ const auto &inputDims
+ = std::static_pointer_cast<Tensor>(mOp.getInput(inputIdx))->dims();
+
+ return std::accumulate(
+ inputDims.begin(),
+ inputDims.end(),
+ Aidge::NbElts_t(1),
+ std::multiplies<Aidge::NbElts_t>());
+}
+
+Aidge::NbElts_t
+ Aidge::MatMulImpl_cpu::getNbRequiredProtected(const Aidge::IOIndex_t /*inputIdx*/) const
+{
+ // for the direct convolution algorithm, convolutions can be in-place, if
+ // there is no padding!
+ return 0;
+}
+
+Aidge::NbElts_t Aidge::MatMulImpl_cpu::getRequiredMemory(
+ const IOIndex_t outputIdx, const std::vector<DimSize_t> &/*inputsSize*/) const
+{
+ // Requires the whole tensors, regardless of available data on inputs
+ assert(outputIdx == 0 && "operator has only one output");
+ (void) outputIdx;
+
+ const auto &outputDims = std::static_pointer_cast<Tensor>(mOp.getOutput(0))->dims();
+ return std::accumulate(
+ outputDims.begin(),
+ outputDims.end(),
+ static_cast<NbElts_t>(1),
+ std::multiplies<NbElts_t>());
+}
+
+Aidge::NbElts_t Aidge::MatMulImpl_cpu::getNbConsumedData(Aidge::IOIndex_t inputIdx) const
+{
+ assert((inputIdx != gk_IODefaultIndex) && (inputIdx < mNbConsumedData.size()));
+ return mNbConsumedData[static_cast<std::size_t>(inputIdx)];
+}
+
+Aidge::NbElts_t Aidge::MatMulImpl_cpu::getNbProducedData(Aidge::IOIndex_t outputIdx) const
+{
+ assert(static_cast<std::size_t>(outputIdx) < mNbProducedData.size());
+ return mNbProducedData[static_cast<std::size_t>(outputIdx)];
+}
+
+void Aidge::MatMulImpl_cpu::updateConsummerProducer(){
+ // Update producer-consumer data
+ for (IOIndex_t inputIdx = 0; static_cast<std::size_t>(inputIdx) < mNbConsumedData.size(); ++inputIdx)
+ mNbConsumedData[inputIdx]
+ += getNbRequiredData(static_cast<std::size_t>(inputIdx)); // each input is consumed by the minimum
+ // amount for a forward pass
+
+ mNbProducedData[0] += getRequiredMemory(0, {});
+}
+
+void Aidge::MatMulImpl_cpu::forward()
+{
+ // FIXME: uncomment the following code once memory handling will work
+ assert(mOp.getInput(0) && "missing input #0");
+ assert(mOp.mInputs[1] && "missing input #1");
+
+ // Find the correct kernel type
+ auto kernelFunc = Registrar<MatMulImplForward_cpu>::create(
+ {mOp.getInput(0)->dataType(),
+ mOp.mInputs[1]->dataType(),
+ mOp.getOutput(0)->dataType()});
+
+ // Call kernel
+ // if (mOp.getInput(0)->nbDims() == 4) {
+ // kernelFunc(
+ // mOp.getStaticAttributes(),
+ // std::static_pointer_cast<Tensor>(mOp.getInput(0))->dims<4>(),
+ // mOp.getInput(0)->getImpl()->rawPtr(),
+ // mOp.mInputs[1]->getImpl()->rawPtr(),
+ // mOp.mInputs[2]->getImpl()->rawPtr(),
+ // mOp.getOutput(0)->getImpl()->rawPtr());
+ // }
+ // else
+ kernelFunc(
+ mOp.getStaticAttributes(),
+ mOp.getInput(0)->dims()[0],
+ mOp.getInput(0)->sizeM1(),
+ mOp.getInput(0)->getImpl()->rawPtr(),
+ mOp.mInputs[1]->getImpl()->rawPtr(),
+ mOp.getOutput(0)->getImpl()->rawPtr());
+
+
+}
+
+void Aidge::MatMulImpl_cpu::backward()
+{
+ printf("Not implemented yet.\n");
+}
diff --git a/src/operator/ScalingImpl.cpp b/src/operator/ScalingImpl.cpp
index 0770a0ccc1434d03fc26b07c425053cd7c09bee6..84cd6ee33a8316a24bae472c74c039dabe0afba3 100644
--- a/src/operator/ScalingImpl.cpp
+++ b/src/operator/ScalingImpl.cpp
@@ -10,16 +10,14 @@
********************************************************************************/
#include <cassert>
-#include <chrono> // std::chrono::milliseconds
-#include <numeric> // std::accumulate
-#include <thread> // std::this_thread::sleep_for
+#include <numeric> // std::accumulate
+#include <functional> // std::multiplies
#include "aidge/operator/Scaling.hpp"
#include "aidge/backend/cpu/operator/ScalingImpl.hpp"
#include "aidge/backend/cpu/operator/ScalingImpl_forward_kernels.hpp"
#include "aidge/utils/Types.h"
-#include <numeric>
#include <vector>
// FIXME: replace whole Tensor with minimum needed data quantity
@@ -38,7 +36,9 @@ Aidge::NbElts_t Aidge::ScalingImpl_cpu::getNbRequiredProtected(const Aidge::IOIn
return 0;
}
-Aidge::NbElts_t Aidge::ScalingImpl_cpu::getRequiredMemory(__attribute__((unused)) const Aidge::IOIndex_t outputIdx, __attribute__((unused)) const std::vector<Aidge::DimSize_t> &inputsSize) const {
+Aidge::NbElts_t Aidge::ScalingImpl_cpu::getRequiredMemory(const Aidge::IOIndex_t outputIdx, const std::vector<Aidge::DimSize_t> &inputsSize) const {
+ (void) outputIdx;
+ (void) inputsSize;
const auto& outputDims = mOp.getOutput(0)->dims();
return std::accumulate(outputDims.begin(), outputDims.end(),
static_cast<NbElts_t>(1), std::multiplies<NbElts_t>());
@@ -68,7 +68,7 @@ void Aidge::ScalingImpl_cpu::forward() {
mOp.getOutput(0)->dataType()});
// Call kernel
- kernelFunc(mOp.getParams(),
+ kernelFunc(mOp.getStaticAttributes(),
std::static_pointer_cast<Tensor>(mOp.getInput(0))->size(),
mOp.getInput(0)->getImpl()->rawPtr(),
mOp.getOutput(0)->getImpl()->rawPtr());
diff --git a/unit_tests/Test_TensorImpl.cpp b/unit_tests/data/Test_TensorImpl.cpp
similarity index 78%
rename from unit_tests/Test_TensorImpl.cpp
rename to unit_tests/data/Test_TensorImpl.cpp
index d28505f7b6f2961e581cadee778bdb16364353de..6c75c4dc19ff1b646308858ad262441d43390122 100644
--- a/unit_tests/Test_TensorImpl.cpp
+++ b/unit_tests/data/Test_TensorImpl.cpp
@@ -41,12 +41,12 @@ TEST_CASE("Tensor creation") {
}
SECTION("get function") {
- REQUIRE(x.get<int>(std::array<std::size_t, 3>({0, 0, 0})) == 1);
- REQUIRE(x.get<int>(std::array<std::size_t, 3>({0, 0, 1})) == 2);
- REQUIRE(x.get<int>(std::array<std::size_t, 3>({0, 1, 1})) == 4);
- REQUIRE(x.get<int>(std::array<std::size_t, 3>({1, 1, 0})) == 7);
- x.get<int>(std::array<std::size_t, 3>({1, 1, 1})) = 36;
- REQUIRE(x.get<int>(std::array<std::size_t, 3>({1, 1, 1})) == 36);
+ REQUIRE(x.get<int>({0, 0, 0}) == 1);
+ REQUIRE(x.get<int>({0, 0, 1}) == 2);
+ REQUIRE(x.get<int>({0, 1, 1}) == 4);
+ REQUIRE(x.get<int>({1, 1, 0}) == 7);
+ x.get<int>({1, 1, 1}) = 36;
+ REQUIRE(x.get<int>({1, 1, 1}) == 36);
}
SECTION("Pretty printing for debug") { REQUIRE_NOTHROW(x.print()); }
diff --git a/unit_tests/operator/Test_AddImpl.cpp b/unit_tests/operator/Test_AddImpl.cpp
index e24d7ac6bd97586ebdeddce5ccb75807ddf530f0..18d98d169ddcb74310c5153d7c2c95103c395bb7 100644
--- a/unit_tests/operator/Test_AddImpl.cpp
+++ b/unit_tests/operator/Test_AddImpl.cpp
@@ -19,25 +19,25 @@
using namespace Aidge;
TEST_CASE("[cpu/operator] Add(forward)") {
- std::shared_ptr<Tensor> input1 = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
- {
- {
- {{20, 47},{21, 48},{22, 49}},
- {{23, 50},{24, 51},{25, 52}},
- {{26, 53},{27, 54},{28, 55}}
- },
- {
- {{29, 56},{30, 57},{31, 58}},
- {{32, 59},{33, 60},{34, 61}},
- {{35, 62},{36, 63},{37, 64}}
- },
- {
- {{38, 65},{39, 66},{40, 67}},
- {{41, 68},{42, 69},{43, 70}},
- {{44, 71},{45, 72},{46, 73}}
- }
- }
- });
+ std::shared_ptr<Tensor> input1 = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
+ { //
+ { //
+ {{20, 47},{21, 48},{22, 49}}, //
+ {{23, 50},{24, 51},{25, 52}}, //
+ {{26, 53},{27, 54},{28, 55}} //
+ }, //
+ { //
+ {{29, 56},{30, 57},{31, 58}}, //
+ {{32, 59},{33, 60},{34, 61}}, //
+ {{35, 62},{36, 63},{37, 64}} //
+ }, //
+ { //
+ {{38, 65},{39, 66},{40, 67}}, //
+ {{41, 68},{42, 69},{43, 70}}, //
+ {{44, 71},{45, 72},{46, 73}} //
+ } //
+ } //
+ }); //
SECTION("One input") {
std::shared_ptr<Node> myAdd = Add<1>();
@@ -51,7 +51,7 @@ TEST_CASE("[cpu/operator] Add(forward)") {
}
SECTION("Two inputs") {
- std::shared_ptr<Tensor> expectedOutput = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
+ std::shared_ptr<Tensor> expectedOutput = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
{
{
{{40, 94},{42, 96},{44, 98}},
@@ -81,9 +81,9 @@ TEST_CASE("[cpu/operator] Add(forward)") {
REQUIRE(*std::static_pointer_cast<Tensor>(myAdd->getOperator()->getOutput(0)) == *expectedOutput);
}
-
+
SECTION("Three inputs") {
- std::shared_ptr<Tensor> expectedOutput = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
+ std::shared_ptr<Tensor> expectedOutput = std::make_shared<Tensor>(Array4D<int,3,3,3,2> {
{
{
{{ 60, 141},{ 63, 144},{ 66, 147}},
diff --git a/unit_tests/operator/Test_LeakyReLUImpl.cpp b/unit_tests/operator/Test_LeakyReLUImpl.cpp
index 7096962e196c2ace4abf2b0b14aca8dfa37d3441..d5bd91ff75404a7b928c8919c64e06315b78206f 100644
--- a/unit_tests/operator/Test_LeakyReLUImpl.cpp
+++ b/unit_tests/operator/Test_LeakyReLUImpl.cpp
@@ -153,7 +153,7 @@ TEST_CASE("[cpu/operator] LeakyReLU(forward)") {
REQUIRE(*myLeakyReLU->getOperator()->getOutput(0) == *expectedOutput);
}
- SECTION("Test construction parameter: negative_slop") {
+ SECTION("Test construction attribute: negative_slop") {
std::shared_ptr<Tensor> input0 = std::make_shared<Tensor>(Array1D<float,10> {
{0.0f, 1.0f, 2.0f,-3.0f, 4.0f,-5.0f,-6.0f, 7.0f, 8.0f, 9.0f}
});
diff --git a/unit_tests/operator/Test_MatMulImpl.cpp b/unit_tests/operator/Test_MatMulImpl.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..0da01b3287043e07e5b967df8882960cfb814f8f
--- /dev/null
+++ b/unit_tests/operator/Test_MatMulImpl.cpp
@@ -0,0 +1,108 @@
+/********************************************************************************
+ * 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 <catch2/catch_test_macros.hpp>
+#include <memory>
+
+#include "aidge/data/Tensor.hpp"
+#include "aidge/operator/MatMul.hpp"
+
+#include "aidge/backend/cpu/operator/MatMulImpl.hpp"
+
+using namespace Aidge;
+
+TEST_CASE("[cpu/operator] MatMul(forward)", "[MatMul]") {
+ // Test MatMul forward with batch size = 2 and feature size = 75
+ std::shared_ptr<Tensor> myWeights = std::make_shared<Tensor>(Array2D<int, 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> myOutput = std::make_shared<Tensor>(Array2D<int, 2, 5>{
+ {{23600, 23600, 23600, 23600, 23600}, {68600, 68600, 68600, 68600, 68600}}});
+
+ std::shared_ptr<Node> myMatMul = MatMul(5, "mymatmul");
+ myMatMul->getOperator()->setDatatype(DataType::Int32);
+ myMatMul->getOperator()->setBackend("cpu");
+ myMatMul->getOperator()->associateInput(1, myWeights);
+
+ SECTION("2D input") {
+ std::shared_ptr<Tensor> myInput = std::make_shared<Tensor>(Array2D<int, 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}}});
+ myMatMul->getOperator()->associateInput(0, myInput);
+ myMatMul->getOperator()->computeOutputDims();
+ myMatMul->forward();
+ REQUIRE(*std::static_pointer_cast<Tensor>(myMatMul->getOperator()->getOutput(0)) == *myOutput);
+ }
+ SECTION("4D input") {
+ std::shared_ptr<Tensor> myInput =
+ std::make_shared<Tensor>(Array4D<int, 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}}}}});
+ myMatMul->getOperator()->associateInput(0, myInput);
+ myMatMul->getOperator()->computeOutputDims();
+ myMatMul->forward();
+ REQUIRE(*std::static_pointer_cast<Tensor>(myMatMul->getOperator()->getOutput(0)) == *myOutput);
+ }
+
+ // std::cout << static_cast<Tensor>((*myMatMul->getOperator())["weight"])[0][0][0][0] << std::endl;
+}
\ No newline at end of file
diff --git a/unit_tests/Test_Scheduler.cpp b/unit_tests/scheduler/Test_Scheduler.cpp
similarity index 100%
rename from unit_tests/Test_Scheduler.cpp
rename to unit_tests/scheduler/Test_Scheduler.cpp