Skip to content

Use round-to-negative division when computing output sizes for convolutions involving striding and dilation. #9640

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
resistor wants to merge 1 commit into from
Closed

Use round-to-negative division when computing output sizes for convolutions involving striding and dilation. #9640

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions aten/src/ATen/div_rtn.h
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,11 @@
#pragma once

// Integer division rounding to -Infinity
template<typename T>
static inline T div_rtn(T x, T y) {
int q = x/y;
int r = x%y;
if ((r!=0) && ((r<0) != (y<0))) --q;
return q;
}

6 changes: 4 additions & 2 deletions aten/src/THCUNN/generic/Col2Im.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/Col2Im.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(Col2Im_shapeCheck)(
THCState *state,
THCTensor *input,
Expand Down Expand Up @@ -31,8 +33,8 @@ static inline void THNN_(Col2Im_shapeCheck)(
}

int64_t inputLength = input->size(batch_dim + 2);
int64_t nBlocksH = 1 + (outputHeight + 2 * padH - dH * (kH - 1) - 1) / sH;
int64_t nBlocksW = 1 + ( outputWidth + 2 * padW - dW * (kW - 1) - 1) / sW;
int64_t nBlocksH = div_rtn<int64_t>(outputHeight + 2 * padH - dH * (kH - 1) - 1, sH) + 1;
int64_t nBlocksW = div_rtn<int64_t>(outputWidth + 2 * padW - dW * (kW - 1) - 1, sW) + 1;

if (inputLength != (nBlocksH * nBlocksW)) {
THError("Given output_size=(%d, %d), kernel_size=(%d, %d), "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THCUNN/generic/Im2Col.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/Im2Col.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(Im2Col_shapeCheck)(
THCState *state,
THCTensor *input,
Expand Down Expand Up @@ -29,8 +31,8 @@ static inline void THNN_(Im2Col_shapeCheck)(
int64_t nInputPlane = THCTensor_(size)(state, input, dim_batch + 1);
int64_t inputHeight = THCTensor_(size)(state, input, dim_batch + 2);
int64_t inputWidth = THCTensor_(size)(state, input, dim_batch + 3);
int64_t outputHeight = (inputHeight + 2 * padH - (dH * (kH - 1) + 1)) / sH + 1;
int64_t outputWidth = (inputWidth + 2 * padW - (dW * (kW - 1) + 1)) / sW + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2 * padH - (dH * (kH - 1) + 1), sH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2 * padW - (dW * (kW - 1) + 1), sW) + 1;

if (outputHeight < 1 || outputWidth < 1) {
THError("Given input with spatial size (%d, %d), kernel_size=(%d, %d), "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THCUNN/generic/SpatialConvolutionMM.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/SpatialConvolutionMM.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(SpatialConvolutionMM_shapeCheck)(
THCState *state,
THCTensor *input, THCTensor *gradOutput,
Expand Down Expand Up @@ -49,8 +51,8 @@ static inline void THNN_(SpatialConvolutionMM_shapeCheck)(
exactInputHeight, exactInputWidth, kH, kW);
}

int64_t outputHeight = (exactInputHeight - kH) / dH + 1;
int64_t outputWidth = (exactInputWidth - kW) / dW + 1;
int64_t outputHeight = div_rtn<int64_t>(exactInputHeight - kH, dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(exactInputWidth - kW, dW) + 1;

if (outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld). "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THCUNN/generic/SpatialDilatedConvolution.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/SpatialDilatedConvolution.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(SpatialDilatedConvolution_shapeCheck)(
THCState *state,
THCTensor *input, THCTensor *gradOutput,
Expand Down Expand Up @@ -44,8 +46,8 @@ static inline void THNN_(SpatialDilatedConvolution_shapeCheck)(
int64_t inputHeight = input->size(dimh);
int64_t inputWidth = input->size(dimw);

int64_t outputHeight = (inputHeight + 2*padH - (dilationH * (kH - 1) + 1)) / dH + 1;
int64_t outputWidth = (inputWidth + 2*padW - (dilationW * (kW - 1) + 1)) / dW + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2*padH - (dilationH * (kH - 1) + 1), dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2*padW - (dilationW * (kW - 1) + 1), dW) + 1;

if (outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld). "
Expand Down
8 changes: 5 additions & 3 deletions aten/src/THCUNN/generic/VolumetricConvolution.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/VolumetricConvolution.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(VolumetricConvolution_shapeCheck)
(THCState *state,
THCTensor *input,
Expand Down Expand Up @@ -83,9 +85,9 @@ static inline void THNN_(VolumetricConvolution_shapeCheck)
exactInputDepth,exactInputHeight,exactInputWidth,kT,kH,kW);
}

int64_t outputWidth = (exactInputDepth - kH) / dH + 1;
int64_t outputHeight = (exactInputHeight - kT) / dT + 1;
int64_t outputDepth = (exactInputWidth - kW) / dW + 1;
int64_t outputWidth = div_rtn<int64_t>(exactInputDepth - kH, dH) + 1;
int64_t outputHeight = div_rtn<int64_t>(exactInputHeight - kT, dT) + 1;
int64_t outputDepth = div_rtn<int64_t>(exactInputWidth - kW, dW) + 1;

if (outputWidth < 1 || outputHeight < 1 || outputDepth < 1)
{
Expand Down
8 changes: 5 additions & 3 deletions aten/src/THCUNN/generic/VolumetricDilatedConvolution.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define THC_GENERIC_FILE "generic/VolumetricDilatedConvolution.cu"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(VolumetricDilatedConvolution_shapeCheck)(
THCState *state,
THCTensor *input,
Expand Down Expand Up @@ -53,9 +55,9 @@ static inline void THNN_(VolumetricDilatedConvolution_shapeCheck)(
int64_t inputDepth = input->size(dimd);
int64_t inputHeight = input->size(dimh);
int64_t inputWidth = input->size(dimw);
int64_t outputDepth = (inputDepth + 2*padT - (dilationT * (kT - 1) + 1)) / dT + 1;
int64_t outputHeight = (inputHeight + 2*padH - (dilationH * (kH - 1) + 1)) / dH + 1;
int64_t outputWidth = (inputWidth + 2*padW - (dilationW * (kW - 1) + 1)) / dW + 1;
int64_t outputDepth = div_rtn<int64_t>(inputDepth + 2*padT - (dilationT * (kT - 1) + 1), dT) + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2*padH - (dilationH * (kH - 1) + 1), dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2*padW - (dilationW * (kW - 1) + 1), dW) + 1;

if (outputDepth < 1 || outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld x %ld). "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THNN/generic/Col2Im.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/Col2Im.c"
#else

#include <ATen/div_rtn.h>

// Note [im2col/col2im output padding]
// ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
// Our implementations of im2col and col2im take both the input height/width as
Expand Down Expand Up @@ -138,8 +140,8 @@ static inline void THNN_(Col2Im_shapeCheck)(
}

int64_t inputLength = input->size(batch_dim + 2);
int64_t nBlocksH = 1 + (outputHeight + 2 * padH - dH * (kH - 1) - 1) / sH;
int64_t nBlocksW = 1 + ( outputWidth + 2 * padW - dW * (kW - 1) - 1) / sW;
int64_t nBlocksH = div_rtn<int64_t>(outputHeight + 2 * padH - dH * (kH - 1) - 1, sH) + 1;
int64_t nBlocksW = div_rtn<int64_t>(outputWidth + 2 * padW - dW * (kW - 1) - 1, sW) + 1;

if (inputLength != (nBlocksH * nBlocksW)) {
THError("Given output_size=(%d, %d), kernel_size=(%d, %d), "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THNN/generic/Im2Col.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/Im2Col.c"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(Im2Col_shapeCheck)(
THNNState *state,
THTensor *input,
Expand All @@ -27,8 +29,8 @@ static inline void THNN_(Im2Col_shapeCheck)(
int64_t nInputPlane = THTensor_(size)(input, dim_batch + 1);
int64_t inputHeight = THTensor_(size)(input, dim_batch + 2);
int64_t inputWidth = THTensor_(size)(input, dim_batch + 3);
int64_t outputHeight = (inputHeight + 2 * padH - (dH * (kH - 1) + 1)) / sH + 1;
int64_t outputWidth = (inputWidth + 2 * padW - (dW * (kW - 1) + 1)) / sW + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2 * padH - (dH * (kH - 1) + 1), sH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2 * padW - (dW * (kW - 1) + 1), sW) + 1;
int64_t nOutputPlane = nInputPlane * kW * kH;
int64_t outputLength = outputHeight * outputWidth;

Expand Down
6 changes: 4 additions & 2 deletions aten/src/THNN/generic/SpatialConvolutionMM.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/SpatialConvolutionMM.c"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(SpatialConvolutionMM_shapeCheck)(
THTensor *input, THTensor *gradOutput,
THTensor *weight, THTensor *bias,
Expand Down Expand Up @@ -48,8 +50,8 @@ static inline void THNN_(SpatialConvolutionMM_shapeCheck)(
exactInputHeight, exactInputWidth, kH, kW);
}

int64_t outputHeight = (exactInputHeight - kH) / dH + 1;
int64_t outputWidth = (exactInputWidth - kW) / dW + 1;
int64_t outputHeight = div_rtn<int64_t>(exactInputHeight - kH, dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(exactInputWidth - kW, dW) + 1;

if (outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld). "
Expand Down
6 changes: 4 additions & 2 deletions aten/src/THNN/generic/SpatialDilatedConvolution.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/SpatialDilatedConvolution.c"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(SpatialDilatedConvolution_shapeCheck)(
THTensor *input, THTensor *gradOutput,
THTensor *weight, THTensor *bias,
Expand Down Expand Up @@ -43,8 +45,8 @@ static inline void THNN_(SpatialDilatedConvolution_shapeCheck)(
int64_t inputHeight = input->size(dimh);
int64_t inputWidth = input->size(dimw);

int64_t outputHeight = (inputHeight + 2*padH - (dilationH * (kH - 1) + 1)) / dH + 1;
int64_t outputWidth = (inputWidth + 2*padW - (dilationW * (kW - 1) + 1)) / dW + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2*padH - (dilationH * (kH - 1) + 1), dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2*padW - (dilationW * (kW - 1) + 1), dW) + 1;

if (outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld). "
Expand Down
8 changes: 5 additions & 3 deletions aten/src/THNN/generic/VolumetricConvolutionMM.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/VolumetricConvolutionMM.c"
#else

#include <ATen/div_rtn.h>

#define CONV3D_OMP_THRESHOLD 20

static void inline THNN_(VolumetricConvolutionMM_shapeCheck)(
Expand Down Expand Up @@ -76,9 +78,9 @@ static void inline THNN_(VolumetricConvolutionMM_shapeCheck)(
exactInputDepth, exactInputHeight, exactInputWidth, kT, kH, kW);
}

outputDepth = (exactInputDepth - kT) / dT + 1;
outputHeight = (exactInputHeight - kH) / dH + 1;
outputWidth = (exactInputWidth - kW) / dW + 1;
outputDepth = div_rtn<int64_t>(exactInputDepth - kT, dT) + 1;
outputHeight = div_rtn<int64_t>(exactInputHeight - kH, dH) + 1;
outputWidth = div_rtn<int64_t>(exactInputWidth - kW, dW) + 1;


if (outputDepth < 1 || outputWidth < 1 || outputHeight < 1) {
Expand Down
8 changes: 5 additions & 3 deletions aten/src/THNN/generic/VolumetricDilatedConvolution.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
#define TH_GENERIC_FILE "generic/VolumetricDilatedConvolution.c"
#else

#include <ATen/div_rtn.h>

static inline void THNN_(VolumetricDilatedConvolution_shapeCheck)(
THTensor *input, THTensor *gradOutput,
THTensor *weight, THTensor *bias,
Expand Down Expand Up @@ -47,9 +49,9 @@ static inline void THNN_(VolumetricDilatedConvolution_shapeCheck)(
int64_t inputDepth = input->size(dimd);
int64_t inputHeight = input->size(dimh);
int64_t inputWidth = input->size(dimw);
int64_t outputDepth = (inputDepth + 2*padT - (dilationT * (kT - 1) + 1)) / dT + 1;
int64_t outputHeight = (inputHeight + 2*padH - (dilationH * (kH - 1) + 1)) / dH + 1;
int64_t outputWidth = (inputWidth + 2*padW - (dilationW * (kW - 1) + 1)) / dW + 1;
int64_t outputDepth = div_rtn<int64_t>(inputDepth + 2*padT - (dilationT * (kT - 1) + 1), dT) + 1;
int64_t outputHeight = div_rtn<int64_t>(inputHeight + 2*padH - (dilationH * (kH - 1) + 1), dH) + 1;
int64_t outputWidth = div_rtn<int64_t>(inputWidth + 2*padW - (dilationW * (kW - 1) + 1), dW) + 1;

if (outputDepth < 1 || outputWidth < 1 || outputHeight < 1) {
THError("Given input size per channel: (%ld x %ld x %ld). "
Expand Down
10 changes: 10 additions & 0 deletions test/test_nn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -712,6 +712,16 @@ def test_no_grad(self):
self.assertFalse(output2.requires_grad)
self.assertRaises(RuntimeError, lambda: output2.backward(torch.ones(1, 5, 10, 10)))

def test_invalid_conv2d(self):
module = torch.nn.Conv2d(1, 1, kernel_size=3, dilation=2, stride=2)
input = torch.empty(1, 1, 4, 4)
self.assertRaises(RuntimeError, lambda: module(input))

def test_invalid_conv3d(self):
module = torch.nn.Conv3d(1, 1, kernel_size=3, dilation=2, stride=2)
input = torch.empty(1, 1, 4, 4, 4)
self.assertRaises(RuntimeError, lambda: module(input))

def _test_dropout(self, cls, input):
p = 0.2
input.fill_(1 - p)
Expand Down