Memory/dataset size limitations

[mpetkova]

Hello,

I'm running into memory issues when running the code on AWS instances with 8GB + of GPU memory. I am working with large datasets (they can be up to 200,000,000 points), so I tried testing gDel3D on 10,000,000 randomly-generated points. I got the following error, which was thrown when the code reached ~8GB of memory use, even on 16GB GPU instances:

cudaCheckError() with sync failed at /home/ubuntu/gDel3D/GDelFlipping/src/gDel3D/GpuDelaunay.cu:901 : an illegal memory access was encountered

Is there a hard limit to the number of points gDel3D can be run on, or are there any workarounds?

Thank you

[ghemann]

I agree, this is a problem for me as well. Any resolutions to this?

[JagnDC]

I also encountered the same problem, is there any solution?

[JagnDC]

I'm working with cuda9.0(1080ti). When I set pointnum=17000000, It will form an endless cycle.
Could someone solve the problem, please tell me ,thank u !