cmake . -B build - Compiling Neural Graphic Primitives - cmake

Question

I trying compile NGP but after cmake . -B build I get this lines:
I've installed
Visual Studio 2019
CUDA Toolkit 11.6
CMake 3.22
Python 3.10
Optix 7.6
C:\NGP\instant-ngp>cmake . -B build
-- Building for: Visual Studio 16 2019
-- Selecting Windows SDK version 10.0.19041.0 to target Windows 10.0.19043.
-- The C compiler identification is MSVC 19.29.30147.0
-- The CXX compiler identification is MSVC 19.29.30147.0
-- The CUDA compiler identification is NVIDIA 11.6.124
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.29.30133/bin/Hostx64/x64/cl.exe - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2019/Community/VC/Tools/MSVC/14.29.30133/bin/Hostx64/x64/cl.exe - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- Detecting CUDA compiler ABI info
-- Detecting CUDA compiler ABI info - done
-- Check for working CUDA compiler: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.6/bin/nvcc.exe - skipped
-- Detecting CUDA compile features
-- Detecting CUDA compile features - done
-- **Obtained CUDA architectures automatically from installed GPUs**
**CMake Error: The source directory "SOURCES/CMakeFiles/CMakeTmp" does not exist.**
Specify --help for usage, or press the help button on the CMake GUI.
CMake Error at dependencies/tiny-cuda-nn/CMakeLists.txt:107 (try_run):
Failed to configure test project build system.
Call Stack (most recent call first):
dependencies/tiny-cuda-nn/CMakeLists.txt:146 (TCNN_AUTODETECT_CUDA_ARCHITECTURES)
-- Configuring incomplete, errors occurred!
See also "C:/NGP/instant-ngp/build/CMakeFiles/CMakeOutput.log".
I am not coder I just want to try NERF. It ss possible download somewhere compiled NGP or it doesn't work that way?

Answer 1

EDIT: Noticed that OP is using the -S flag implicitly the issue seems to be elsewhere, hence the updated answer:
UPDATED ANSWER
Please bare in mind that with the limited information at my disposal. This is again only a wild guess, based on what I tried.
After re-examining the output of CMake I've noticed that the issue is within one of the dependencies i.e.
CMake Error at dependencies/tiny-cuda-nn/CMakeLists.txt:107 (try_run):
I was not able to completely replicate the problem, however I do believe I know how to debug these issues. My wild guess here is that either:
The individual dependencies were incorrectly pulled or
The CMake files were poorly generated due to some issue and you didn't clean them.
I would recommend trying the following:
Delete the build folder, and run cmake -S. -Bbuild again. And if that fails you can try the second option.
Remove the git repository (folder called instant-ngp) and clone it again using exactly this command:
git clone --recursive https://github.com/nvlabs/instant-ngp
Please note that the --recursive option is very important due to the fact that you need to pull the dependant repositories as well.
Now if all of that fails. You still have one more option and that is to go to the folder dependencies/tiny-cuda-nn/ and generate the CMake files yourself, however I do not recommend that due to the fact that if you have limited experience you may not know what exactly to do.
I've also checked the variables you wanted to set, and you may do that by running this in Powershell
$Env:CMAKE_CUDA_ARCHITECTURES=YOUR_VALUE_HERE (in my case it is 75 because I have a RTX2080). However bare in mind that
You shouldn't need it, because the architecture is discovered for you
You need to delete the build folder in order to clean up the cached CMake files
Hope it helps!