Skip to content
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.

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

KinFu2 error: invalid configuration argument #53

Open
LostInCode404 opened this issue Mar 31, 2018 · 4 comments
Open

KinFu2 error: invalid configuration argument #53

LostInCode404 opened this issue Mar 31, 2018 · 4 comments

Comments

@LostInCode404
Copy link

LostInCode404 commented Mar 31, 2018

This error is in the kernel call extract_normals_kernel<<<grid, block>>>(en, output); in kfusion::device::extractNormals method in tsdf_volume.cu. This occurs because value of grid dimensions in arguments is (0,1,1). The first value is set to 0 after the divup because the argument const PtrSz<Point>& points to the extractNormals function is being passed as empty(i.e. points.size is 0). So, I tracked from beginning, i.e. demo.cpp to see where the problem is.

demo.cpp calls has_image = dynamic_fusion(depth_device_); which later calls volume_->compute_points(); and volume_->compute_normals(); here. The latter one will subsequently call the extractNormals method using value of cuda::DeviceArray<Point> *cloud_; which was set to empty in the former one here, being the cause of the problem. So, I think that the problem is that volume_->compute_points(); sets the cloud as empty by calling the function fetchCloud(*cloud_buffer_);.

Moving ahead, in fetchCloud() method, there is a check which checks if the cloud is empty, else it creates a new cloud. That is fine, I checked it and new cloud is being created which is not empty. But when this function returns, it returns DeviceArray<Point>((Point*)cloud_buffer.ptr(), size);, this is itself empty because it uses a size variable which is set to 0 in line size_t size = extractCloud(volume, aff, b);.

Further, in kfusion::device::extractCloud, there is a kernel call extract_kernel<<<grid, block>>>(fs, output); which goes successful but this is the culprit that returns size as 0. The line cudaMemcpyFromSymbol (&size, output_count, sizeof(size)) sets the value of size as 0. I think this is where the problem is.

I'm not completely sure whether the problem is where I'm pointing it out to be, but that's all that I could infer. How should I go about solving this problems? Any help is appreciated. Thanks.

Edit: Does it have something to do with the __CUDA_ARCH__ checks here? OR, checks on this line.

@lbq118
Copy link

lbq118 commented Mar 13, 2019

I also encountered the same problem. Is your problem solved?
Thanks

@ktsiolis91
Copy link

Hello, I've recently cloned this project and unfortunately I encountered the same issue. Did you figure it out?? I would really appreciate any help or guidance.

Thanks!!!

@LuyaooChen
Copy link

I also encountered the same problem...

1 similar comment
@jinzhenmu
Copy link

I also encountered the same problem...

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants