Fix cuDNN error message when it's Conv2d

[xwang233]

Originally introduced in #45023. When I was doing test in the original PR, it was a Conv3d, so this problem was not discovered.

Arrays in ConvolutionParams have a fixed length of 3 or 5. This is because max_dim is set as a constexpr of 3, regardless of Conv2d or Conv3d. The current code will make some error message be weird. See below in the comments.

pytorch/aten/src/ATen/native/cudnn/Conv.cpp

Lines 212 to 226 in 9201c37

	struct ConvolutionParams
	{
	cudnnDataType_t dataType;
	int input_size[2 + max_dim];
	int input_stride[2 + max_dim];
	int weight_size[2 + max_dim];
	int padding[max_dim];
	int stride[max_dim];
	int dilation[max_dim];
	int64_t groups;
	bool deterministic;
	bool allow_tf32;
	// NB: transposed purposely omitted: transposed just swaps
	// forward and backward, so you can reuse the benchmark entry,
	};

[xwang233]

The weird error message before this PR.

Traceback (most recent call last):
  something
RuntimeError: cuDNN error: CUDNN_STATUS_INTERNAL_ERROR
You can try to repro this exception using the following code snippet. If that doesn't trigger the error, please include your original repro script when reporting this issue.

import torch
torch.backends.cuda.matmul.allow_tf32 = True
torch.backends.cudnn.benchmark = True
torch.backends.cudnn.deterministic = False
torch.backends.cudnn.allow_tf32 = True
data = torch.randn([128, 512, 7, 7], dtype=torch.float, device='cuda', requires_grad=True)
net = torch.nn.Conv2d(512, 2048, kernel_size=[1, 1], padding=[0, 0, 0], stride=[1, 1, 0], dilation=[1, 1, 0], groups=1)
net = net.cuda().float()
out = net(data)
out.backward(torch.randn_like(out))
torch.cuda.synchronize()

ConvolutionParams
    data_type = CUDNN_DATA_FLOAT
    padding = [0, 0, 0]
    stride = [1, 1, 0]
    dilation = [1, 1, 0]
    groups = 1
    deterministic = false
    allow_tf32 = true
input: TensorDescriptor 0x5630c484c770
    type = CUDNN_DATA_FLOAT
    nbDims = 4
    dimA = 128, 512, 7, 7,
    strideA = 25088, 49, 7, 1,
output: TensorDescriptor 0x7f38b8038cd0
    type = CUDNN_DATA_FLOAT
    nbDims = 4
    dimA = 128, 2048, 7, 7,
    strideA = 100352, 49, 7, 1,
weight: FilterDescriptor 0x7f38b8066d40
    type = CUDNN_DATA_FLOAT
    tensor_format = CUDNN_TENSOR_NCHW
    nbDims = 4
    dimA = 2048, 512, 1, 1,
Pointer addresses:
    input: 0x7f3999700000
    output: 0x7f399d440000
    weight: 0x7f3bb7d80000

Note these padding, stride, dilation are 3d. This PR fix the python repro script when the convolution is 2d.

The ConvolutionParams printed below is a bit hard to fix. However, they are mainly for developer debugging purpose. Hopefully, developers won't feel weird about it.

[xwang233]

cc @ptrblck

[codecov]

Codecov Report

Merging #45729 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master   #45729   +/-   ##
=======================================
  Coverage   68.60%   68.60%           
=======================================
  Files         410      410           
  Lines       52670    52670           
=======================================
  Hits        36132    36132           
  Misses      16538    16538           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9201c37...48d01f6. Read the comment docs.

[facebook-github-bot]

@ngimel has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

[facebook-github-bot]

@ngimel merged this pull request in 8619de8.