From b3f9c722892350b0f4248b074dfdb7c9fcdc4479 Mon Sep 17 00:00:00 2001
From: NAUD Maxence <maxence.naud@cea.fr>
Date: Tue, 14 Jan 2025 17:23:18 +0000
Subject: [PATCH] FIX: Conv[DepthWise] forward implementation for some cases
that were not tested in the new implementation
---
.../operator/ConvDepthWiseImpl_kernels.hpp | 34 +++++++++----------
.../backend/cpu/operator/ConvImpl_kernels.hpp | 12 +++----
2 files changed, 22 insertions(+), 24 deletions(-)
diff --git a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_kernels.hpp b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_kernels.hpp
index 46ae5987..906ea1ad 100644
--- a/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/ConvDepthWiseImpl_kernels.hpp
@@ -175,17 +175,17 @@ void ConvDepthWiseImpl2D_cpu_forward_kernel(const std::array<DimSize_t, 2>& stri
}
}
} else {
- for (std::size_t ox = 0, oIndex = 0; ox < oxSize; ++ox, oIndex+=oySize, iIndex-=strideDims[0]*inputDims[3]) {
+ for (std::size_t ox = 0, oIndex = 0; ox < oxSize; ++ox, oIndex+=oySize, iIndex+=(strideDims[0]-2)*inputDims[3]) {
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+0]*input[iIndex+oy]+weights[wIndex+1]*input[iIndex+oy+strideDims[0]]+weights[wIndex+2]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] = biasVal + weights[wIndex+0]*input[iIndex+oy*strideDims[1]]+weights[wIndex+1]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+2]*input[iIndex+oy*strideDims[1]+2];
}
- iIndex+=strideDims[0]*inputDims[3];
+ iIndex+=inputDims[3];
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+3]*input[iIndex+oy]+weights[wIndex+4]*input[iIndex+oy+strideDims[0]]+weights[wIndex+5]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] += weights[wIndex+3]*input[iIndex+oy*strideDims[1]]+weights[wIndex+4]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+5]*input[iIndex+oy*strideDims[1]+2];
}
- iIndex+=strideDims[0]*inputDims[3];
+ iIndex+=inputDims[3];
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+6]*input[iIndex+oy]+weights[wIndex+7]*input[iIndex+oy+strideDims[0]]+weights[wIndex+8]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] += weights[wIndex+6]*input[iIndex+oy*strideDims[1]]+weights[wIndex+7]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+8]*input[iIndex+oy*strideDims[1]+2];
}
}
}
@@ -193,25 +193,23 @@ void ConvDepthWiseImpl2D_cpu_forward_kernel(const std::array<DimSize_t, 2>& stri
}
}
} else if (dilated_kernel_x == 1 && dilated_kernel_y == 1) {
- std::size_t index = 0;
for (std::size_t batch = 0; batch < inputDims[0]; ++batch) {
for (std::size_t ch = 0; ch < inputDims[1]; ++ch) {
B biasVal = (biases != nullptr) ? biases[ch] : B(0);
- const std::size_t iIndex = (ch + batch*inputDims[1]) * inputDims[2] * inputDims[3];
+ std::size_t iIndex = (ch + batch*inputDims[1]) * inputDims[2] * inputDims[3];
const std::size_t wIndex = ch;
if (strideDims[0] == 1 && strideDims[1] == 1) {
- for (; index < iIndex + oxSize*oySize; ++index) {
- output[index] = biasVal + weights[wIndex] * input[index];
+ for (std::size_t i = iIndex; i < iIndex + oxSize*oySize; ++i) {
+ output[i] = biasVal + weights[wIndex] * input[i];
}
} else {
std::size_t oIndex = (ch + batch*inputDims[1]) * oxSize * oySize;
- for (std::size_t ox = 0; ox < oxSize; ++ox, oIndex+=oySize) {
- index = iIndex + strideDims[0]*inputDims[3];
+ for (std::size_t ox = 0; ox < oxSize; ++ox, oIndex+=oySize, iIndex+=strideDims[0]*inputDims[3]) {
for (std::size_t oy = 0, iy = 0; oy < oySize; ++oy, iy+=strideDims[1]) {
- output[oIndex + oy] += weights[wIndex]*input[index+iy];
+ output[oIndex + oy] = biasVal + weights[wIndex]*input[iIndex+iy];
}
}
}
@@ -234,16 +232,16 @@ void ConvDepthWiseImpl2D_cpu_forward_kernel(const std::array<DimSize_t, 2>& stri
const std::size_t ix = ox * strideDims[0];
const std::size_t iy = oy * strideDims[1];
- for (std::size_t sx = 0; sx*dilationDims[0] < dilated_kernel_x; ++sx) {
- for (std::size_t sy = 0; sy*dilationDims[1] < dilated_kernel_y; ++sy) {
- output[oIndexFull] += weights[wIndex + sx*kernelDims[1] + sy] *
- input[iIndex + static_cast<std::size_t>(ix + sx*dilationDims[0])*inputDims[3] + static_cast<std::size_t>(iy + sy*dilationDims[1])];
+ for (std::size_t kx = 0; kx*dilationDims[0] < dilated_kernel_x; ++kx) {
+ for (std::size_t ky = 0; ky*dilationDims[1] < dilated_kernel_y; ++ky) {
+ output[oIndexFull] += weights[wIndex + kx*kernelDims[1] + ky] *
+ input[iIndex + (ix + kx*dilationDims[0])*inputDims[3] + (iy + ky*dilationDims[1])];
}
}
}
}
+ output += outChannels_s;
}
- output += outChannels_s;
}
}
}
diff --git a/include/aidge/backend/cpu/operator/ConvImpl_kernels.hpp b/include/aidge/backend/cpu/operator/ConvImpl_kernels.hpp
index e3b709bf..1229d571 100644
--- a/include/aidge/backend/cpu/operator/ConvImpl_kernels.hpp
+++ b/include/aidge/backend/cpu/operator/ConvImpl_kernels.hpp
@@ -183,17 +183,17 @@ void ConvImpl2D_cpu_forward_kernel(const std::array<DimSize_t, 2>& strideDims,
}
}
} else {
- for (std::size_t ox = 0, oIndex = 0; ox < oxSize; ++ox, oIndex+=oySize, iIndex-=strideDims[0]*inputDims[3]) {
+ for (std::size_t ox = 0, oIndex = 0; ox < oxSize; ++ox, oIndex+=oySize, iIndex+=(strideDims[0]-2)*inputDims[3]) {
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+0]*input[iIndex+oy]+weights[wIndex+1]*input[iIndex+oy+strideDims[0]]+weights[wIndex+2]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] += weights[wIndex+0]*input[iIndex+oy*strideDims[1]]+weights[wIndex+1]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+2]*input[iIndex+oy*strideDims[1]+2];
}
- iIndex+=strideDims[0]*inputDims[3];
+ iIndex+=inputDims[3];
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+3]*input[iIndex+oy]+weights[wIndex+4]*input[iIndex+oy+strideDims[0]]+weights[wIndex+5]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] += weights[wIndex+3]*input[iIndex+oy*strideDims[1]]+weights[wIndex+4]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+5]*input[iIndex+oy*strideDims[1]+2];
}
- iIndex+=strideDims[0]*inputDims[3];
+ iIndex+=inputDims[3];
for (std::size_t oy = 0; oy < oySize; ++oy) {
- output[oIndex + oy] += weights[wIndex+6]*input[iIndex+oy]+weights[wIndex+7]*input[iIndex+oy+strideDims[0]]+weights[wIndex+8]*input[iIndex+oy+strideDims[0]*2];
+ output[oIndex + oy] += weights[wIndex+6]*input[iIndex+oy*strideDims[1]]+weights[wIndex+7]*input[iIndex+oy*strideDims[1]+1]+weights[wIndex+8]*input[iIndex+oy*strideDims[1]+2];
}
}
}
--
GitLab