Skip to content

Fix prelu_backward TensorIterator split #36134

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
wants to merge 7 commits into from
Closed

Fix prelu_backward TensorIterator split #36134

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
4588625
Fix prelu_backward TensorIterator split
zasdfgbnm Apr 7, 2020
6aa19b9
save
zasdfgbnm Apr 7, 2020
e333b12
save
zasdfgbnm Apr 7, 2020
4f1630d
Merge branch 'master' of github.com:pytorch/pytorch into zasdfgbnm-pa…
zasdfgbnm Apr 17, 2020
8a8d5a9
save
zasdfgbnm Apr 17, 2020
352992d
fix
zasdfgbnm Apr 17, 2020
dcdbc28
Merge branch 'master' into zasdfgbnm-patch-6
zasdfgbnm Apr 22, 2020
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
23 changes: 8 additions & 15 deletions aten/src/ATen/native/cuda/Activation.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -171,7 +171,7 @@ template<typename scalar_t>
void launch_prelu_cuda_backward_share_weights_kernel(TensorIterator &iter, const scalar_t* weight_data) {
if (!iter.can_use_32bit_indexing()) {
for (auto& sub_iter : iter.with_32bit_indexing()) {
launch_prelu_cuda_backward_share_weights_kernel(iter, weight_data);
launch_prelu_cuda_backward_share_weights_kernel(sub_iter, weight_data);
}
return;
}
Expand All @@ -180,6 +180,8 @@ void launch_prelu_cuda_backward_share_weights_kernel(TensorIterator &iter, const
if (numel == 0) {
return;
}

TORCH_INTERNAL_ASSERT_DEBUG_ONLY(iter.can_use_32bit_indexing());

scalar_t *input_grad_data = static_cast<scalar_t *>(iter.data_ptr(0));
scalar_t *weight_grad_collector_data = static_cast<scalar_t *>(iter.data_ptr(1));
Expand All @@ -189,20 +191,11 @@ void launch_prelu_cuda_backward_share_weights_kernel(TensorIterator &iter, const
int64_t grid = (numel + block_work_size - 1) / block_work_size;
auto stream = at::cuda::getCurrentCUDAStream();

if (iter.is_contiguous()) {
prelu_cuda_backward_share_weights_kernel<scalar_t><<<grid, num_threads, 0, stream>>>(
numel, input_data, grad_out_data, input_grad_data, weight_grad_collector_data, weight_data,
TrivialOffsetCalculator<2>(), TrivialOffsetCalculator<2>()
);
} else {
std::array<const int64_t*, 2> out_strides = {iter.strides(0).data(), iter.strides(1).data()};
std::array<const int64_t*, 2> inp_strides = {iter.strides(2).data(), iter.strides(3).data()};
prelu_cuda_backward_share_weights_kernel<scalar_t><<<grid, num_threads, 0, stream>>>(
numel, input_data, grad_out_data, input_grad_data, weight_grad_collector_data, weight_data,
OffsetCalculator<2>(iter.ndim(), iter.shape().data(), inp_strides.data()),
OffsetCalculator<2>(iter.ndim(), iter.shape().data(), out_strides.data())
);
}
TORCH_INTERNAL_ASSERT(iter.is_contiguous());
Copy link
Collaborator Author

@zasdfgbnm zasdfgbnm Apr 22, 2020

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Caller of launch_prelu_cuda_backward_share_weights_kernel already guarantees contiguous.

prelu_cuda_backward_share_weights_kernel<scalar_t><<<grid, num_threads, 0, stream>>>(
numel, input_data, grad_out_data, input_grad_data, weight_grad_collector_data, weight_data,
TrivialOffsetCalculator<2>(), TrivialOffsetCalculator<2>()
);
}

template <typename scalar_t>
Expand Down
8 changes: 8 additions & 0 deletions test/test_nn.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9177,6 +9177,14 @@ def test_convTranspose_empty(self, device):
with torch.backends.cudnn.flags(enabled=False):
self._test_module_empty_input(mod, inp, check_size=False)

@onlyCUDA
@largeCUDATensorTest('16GB')
def test_prelu_backward_32bit_indexing(self, device):
m = torch.nn.PReLU().cuda().half()
input_ = torch.ones((1024, 1024, 1024, 2), dtype=torch.half, device=device)
output = m(input_)
output.backward(input_)

def test_linear_empty(self, device):
mod = torch.nn.Linear(7, 7).to(device)
inp = torch.randn(0, 7, device=device)
Expand Down