cmake error 2

[num-doc]

Built target ORBEEZ 95%] Linking CUDA device code CMakeFiles/0RBEEZ.dir 95%] Linking CXX static library libORBEEZ.a 95%] Built target ORBEEZ gmake: *** [Makefile:156: all] Error 2

How should I solve ？The compilation process is not clear in the end which third-party library compilation failure. I need your help

[num-doc]

instant-ngp-kf error
Thirdparty/instant-ngp-kf/CMakeFiles/testbed.dir/build.make:99: testbed] Error 1 [CMakeFiles/Makefile2:1470: Thirdparty/instant-ngp-kf/CMakeFiles/testbed.dir/all] Error 2

[jeff999955]

@num-doc Hello, would you kindly provide the full log by cmake --build build --parallel $(nproc) 2>&1 | tee build.log so that we could help you out? Thank you!

[num-doc]

Thanks for your help, now recompiling using the exact same environment. Look forward to success。

[xmlyqing00]

Hi, I'm facing the same issue when compiling the project in the last step using cmake --build build --parallel $(nproc --all). It shows

I cannot locate the error even I put -v to see the verbose output. But when I remove the parallel compiling. The error seems to disappear. It's weird. I'm using 4090 and your Docker file. I upgraded the CUDA version to 12.2 in your dockerfile for architecture compatibility. And I also changed TCNN_CUDA_ARCHITECTURES=89. But the reconstruction results are pretty blur, both monocular and RGB-D inputs. Similar to another issue. Do you have any ideas? Thank you!

Best, Yongqing

[jeff999955]

@num-doc Glad to hear your success. Sorry that we did not focus on building the python environment, and in the future if you encounter no packages named "package_name", you could easily resolve it by pip install package_name, e.g. pip install scipy. Thank you!

[jeff999955]

@xmlyqing00 Hello, we are not sure if our code support 4000 series and they were not released when the repo did. Maybe we have to update the instant-ngp backend to support the latest graphic cards.

[xmlyqing00]

@xmlyqing00 Hello, we are not sure if our code support 4000 series and they were not released when the repo did. Maybe we have to update the instant-ngp backend to support the latest graphic cards.

Thank you. We tried RGB-D demo and it worked well on 4090 GPU. For monocular demo, we also tried on 3090 GPU but got similar results as 4090.

[Aurora-chevalier]

@num-doc Hello, would you kindly provide the full log by cmake --build build --parallel $(nproc) 2>&1 | tee build.log so that we could help you out? Thank you!

Thank you for your great work on open source! I'm having some issues with the reproduction process when I use the command "cmake -B build -DCMAKE_BUILD_TYPE=Release . cmake --build build --parallel $(nproc --all)", I get the following error： When I use the directive "cmake --build build --parallel $(nproc) 2>&1 | tee build.log", the following log file is obtained： build.log I would appreciate it if you could help me solve my problem.

[MarvinChung]

@num-doc Hello, would you kindly provide the full log by cmake --build build --parallel $(nproc) 2>&1 | tee build.log so that we could help you out? Thank you!

Thank you for your great work on open source! I'm having some issues with the reproduction process when I use the command "cmake -B build -DCMAKE_BUILD_TYPE=Release . cmake --build build --parallel $(nproc --all)", I get the following error： When I use the directive "cmake --build build --parallel $(nproc) 2>&1 | tee build.log", the following log file is obtained： build.log I would appreciate it if you could help me solve my problem.

It seems to be a CUDA problem, could you provide your GPU, your CUDA version and the log of running

cmake -B build -DCMAKE_BUILD_TYPE=Release .

It might be that your TCNN_CUDA_ARCHITECTURES is not correct, you could first check the compilation section in BUILD.md.

[Aurora-chevalier]

@num-doc Hello, would you kindly provide the full log by so that we could help you out? Thank you!cmake --build build --parallel $(nproc) 2>&1 | tee build.log

Thank you for your great work on open source! I'm having some issues with the reproduction process when I use the command "cmake -B build -DCMAKE_BUILD_TYPE=Release . cmake --build build --parallel $(nproc --all)", I get the following error： When I use the directive "cmake --build build --parallel $(nproc) 2>&1 | tee build.log", the following log file is obtained： build.log I would appreciate it if you could help me solve my problem.

It seems to be a CUDA problem, could you provide your GPU, your CUDA version and the log of running

cmake -B build -DCMAKE_BUILD_TYPE=Release .

It might be that your TCNN_CUDA_ARCHITECTURES is not correct, you could first check the compilation section in BUILD.md.

Thanks for the quick reply! My GPU is 3090 and CUDA version is 11.7 when I run the command "cmake -B build -DCMAKE_BUILD_TYPE=Release." The following logs were obtained: log.txt

[Aurora-chevalier]

@num-doc Hello, would you kindly provide the full log by cmake --build build --parallel $(nproc) 2>&1 | tee build.log so that we could help you out? Thank you!

Thank you for your great work on open source! I'm having some issues with the reproduction process when I use the command "cmake -B build -DCMAKE_BUILD_TYPE=Release . cmake --build build --parallel $(nproc --all)", I get the following error： When I use the directive "cmake --build build --parallel $(nproc) 2>&1 | tee build.log", the following log file is obtained： build.log I would appreciate it if you could help me solve my problem.

It seems to be a CUDA problem, could you provide your GPU, your CUDA version and the log of running

cmake -B build -DCMAKE_BUILD_TYPE=Release .

It might be that your TCNN_CUDA_ARCHITECTURES is not correct, you could first check the compilation section in BUILD.md. I've solved part of the problem, but there are still a couple of errors about "pybind11" at compile time, and the compiled log file is as follows, do you know how to deal with it. build.log ![Uploading 2023-12-30 171357.png…]()

[num-doc]

@Aurora-chevalier Thank you for your attention to my problem. This problem has been solved after I reduced the CUDA version. I guess there is a compatibility problem with CUDA version too high.

[Aurora-chevalier]

@Aurora-chevalier Thank you for your attention to my problem. This problem has been solved after I reduced the CUDA version. I guess there is a compatibility problem with CUDA version too high.

Is your CUDA version 11.6?

[num-doc]

@Aurora-chevalier 11.7

[Aurora-chevalier]

@Aurora-chevalier 11.7

My cuda version is also 11.7, but I get the following error when compiling： Do you know how to deal with it？

[Aurora-chevalier]

Thanks for the replies from all of you, I fixed this issue

[zxcasfddfzx]

Thanks for the replies from all of you, I fixed this issue

Could I ask how this issue was resolved?