Fixing build-errors on Windows due to invalid types when calling THCCeilDiv.…

Fixing build-errors on Windows due to invalid types when calling THCCeilDiv. Adding type-cast to (long) to first parameter to fix the errors. (#409)

maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.cu

@@ -272,7 +272,7 @@ at::Tensor ROIAlign_forward_cuda(const at::Tensor& input,
   auto output_size = num_rois * pooled_height * pooled_width * channels;
   cudaStream_t stream = at::cuda::getCurrentCUDAStream();
 
-  dim3 grid(std::min(THCCeilDiv(output_size, 512L), 4096L));
+  dim3 grid(std::min(THCCeilDiv((long)output_size, 512L), 4096L));
   dim3 block(512);
 
   if (output.numel() == 0) {
@@ -317,7 +317,7 @@ at::Tensor ROIAlign_backward_cuda(const at::Tensor& grad,
 
   cudaStream_t stream = at::cuda::getCurrentCUDAStream();
 
-  dim3 grid(std::min(THCCeilDiv(grad.numel(), 512L), 4096L));
+  dim3 grid(std::min(THCCeilDiv((long)grad.numel(), 512L), 4096L));
   dim3 block(512);
 
   // handle possibly empty gradients

maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.cu

@@ -126,7 +126,7 @@ std::tuple<at::Tensor, at::Tensor> ROIPool_forward_cuda(const at::Tensor& input,
 
   cudaStream_t stream = at::cuda::getCurrentCUDAStream();
 
-  dim3 grid(std::min(THCCeilDiv(output_size, 512L), 4096L));
+  dim3 grid(std::min(THCCeilDiv((long)output_size, 512L), 4096L));
   dim3 block(512);
 
   if (output.numel() == 0) {
@@ -173,7 +173,7 @@ at::Tensor ROIPool_backward_cuda(const at::Tensor& grad,
 
   cudaStream_t stream = at::cuda::getCurrentCUDAStream();
 
-  dim3 grid(std::min(THCCeilDiv(grad.numel(), 512L), 4096L));
+  dim3 grid(std::min(THCCeilDiv((long)grad.numel(), 512L), 4096L));
   dim3 block(512);
 
   // handle possibly empty gradients