My name is Tao Zhu. I’m a graduate student at Shanghai University, and interested in parallel computation. When I first get in touch with 3DTK project, I find this project use GPU to accelerate k nearest neighbor search. However, after I use ‘ccmake’ to open the CUDA setting(the fig1 below)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
WITH_CUDA ON
CUDA_64_BIT_DEVICE_CODE ON
CUDA_ATTACH_VS_BUILD_RULE_TO_C OFF
CUDA_BUILD_CUBIN ON
CUDA_BUILD_EMULATION OFF
CUDA_HOST_COMPILATION_CPP ON
CUDA_INSTALL_PREFIX /usr/local/cuda-8.0
CUDA_NVCC_FLAGS
CUDA_SDK_INSTALL_PREFIX CUDA_SDK_INSTALL_PREFIX-NOTFOUND
CUDA_VERBOSE_BUILD OFF
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Fig1 open CUDA settings
and successfully using ‘make’ instruction to compile the project, I don’t know how to use CUDA with the instruction “Slam6D”. I have checked the “Slam6D -h” , but I don’t find the specific option for the CUDA in knn search(fig2).
Hello,
My name is Tao Zhu. I’m a graduate student at Shanghai University, and interested in parallel computation. When I first get in touch with 3DTK project, I find this project use GPU to accelerate k nearest neighbor search. However, after I use ‘ccmake’ to open the CUDA setting(the fig1 below)
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
WITH_CUDA ON
CUDA_64_BIT_DEVICE_CODE ON
CUDA_ATTACH_VS_BUILD_RULE_TO_C OFF
CUDA_BUILD_CUBIN ON
CUDA_BUILD_EMULATION OFF
CUDA_HOST_COMPILATION_CPP ON
CUDA_INSTALL_PREFIX /usr/local/cuda-8.0
CUDA_NVCC_FLAGS
CUDA_SDK_INSTALL_PREFIX CUDA_SDK_INSTALL_PREFIX-NOTFOUND
CUDA_VERBOSE_BUILD OFF
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Fig1 open CUDA settings
and successfully using ‘make’ instruction to compile the project, I don’t know how to use CUDA with the instruction “Slam6D”. I have checked the “Slam6D -h” , but I don’t find the specific option for the CUDA in knn search(fig2).
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
-t [ --nns_method ] arg (=0) selects the Nearest Neighbor Search
Algorithm
0 = simple k-d tree
1 = cached k-d tree
2 = ANNTree
3 = BOCTree
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Fig2 knn search Slam6D -h
After all, I don’t know how to use CUDA code in the project. Hope someone could help me figure out this problem:)