CUDA with PCL

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

CUDA with PCL

sant
Is it possible to use CUDA to accelerate pcl functionality using the GPU?I mean any functions I use now from pcl, like normal computation, feature extraction, etc
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: CUDA with PCL

Sérgio Agostinho

2017-06-27 11:23 GMT+01:00 sant <[hidden email]>:
Is it possible to use CUDA to accelerate pcl functionality using the GPU?I
mean any functions I use now from pcl, like normal computation, feature
extraction, etc



--
View this message in context: http://www.pcl-users.org/CUDA-with-PCL-tp4044667.html
Sent from the Point Cloud Library (PCL) Users mailing list mailing list archive at Nabble.com.
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: CUDA with PCL

Günther
Good morning everyone,

I am new to PCL and would like to shift what I have so far implemented using the CPU to the GPU side.
Unfortunately I am unsure:

- whether pcl::gpu and pcl::cuda namespaces interact transparently (I'd guess they don't)
- how to move existing data to the GPU (ie. for usage with pcl::gpu::EuclideanClusterExtraction)

The documentation tells me that pcl::cuda::PointCloudAOS "is the most efficient way to perform operations on x86 architectures" (http://docs.pointclouds.org/trunk/classpcl_1_1cuda_1_1_point_cloud_a_o_s.html).

Now.... when I look at pcl::gpu::EuclideanClusterExtraction (pcl::gpu namespace instead of pcl::cuda), I find the following method:

void setInput (CloudDevice input)

which takes a "CloudDevice" type as input, which is described in gpu_extract_clusters.h as being a pcl::gpu::Octree::PointCloud ...

So far I am just confused. Could anyone point me to an example which explains interaction between pcl::gpu and pcl::cuda?

Thanks in advance

Günther
Loading...