first push

RAFT/alt_cuda_corr/correlation.cpp

@@ -0,0 +1,54 @@
+#include <torch/extension.h>
+#include <vector>
+
+// CUDA forward declarations
+std::vector<torch::Tensor> corr_cuda_forward(
+    torch::Tensor fmap1,
+    torch::Tensor fmap2,
+    torch::Tensor coords,
+    int radius);
+
+std::vector<torch::Tensor> corr_cuda_backward(
+  torch::Tensor fmap1,
+  torch::Tensor fmap2,
+  torch::Tensor coords,
+  torch::Tensor corr_grad,
+  int radius);
+
+// C++ interface
+#define CHECK_CUDA(x) TORCH_CHECK(x.type().is_cuda(), #x " must be a CUDA tensor")
+#define CHECK_CONTIGUOUS(x) TORCH_CHECK(x.is_contiguous(), #x " must be contiguous")
+#define CHECK_INPUT(x) CHECK_CUDA(x); CHECK_CONTIGUOUS(x)
+
+std::vector<torch::Tensor> corr_forward(
+    torch::Tensor fmap1,
+    torch::Tensor fmap2,
+    torch::Tensor coords,
+    int radius) {
+  CHECK_INPUT(fmap1);
+  CHECK_INPUT(fmap2);
+  CHECK_INPUT(coords);
+
+  return corr_cuda_forward(fmap1, fmap2, coords, radius);
+}
+
+
+std::vector<torch::Tensor> corr_backward(
+    torch::Tensor fmap1,
+    torch::Tensor fmap2,
+    torch::Tensor coords,
+    torch::Tensor corr_grad,
+    int radius) {
+  CHECK_INPUT(fmap1);
+  CHECK_INPUT(fmap2);
+  CHECK_INPUT(coords);
+  CHECK_INPUT(corr_grad);
+
+  return corr_cuda_backward(fmap1, fmap2, coords, corr_grad, radius);
+}
+
+
+PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
+  m.def("forward", &corr_forward, "CORR forward");
+  m.def("backward", &corr_backward, "CORR backward");
+}