Closed 01baftb closed 1 week ago
Have you installed the CUDA toolkit?
Have you installed the CUDA toolkit?
No, my PC doesn't have NVIDA CPU. Doesn't the DLL support CPU only?
Ahh, yes, there is the problem. The precompiled dll that I post requires the cuda libraries. Can you compile the code yourself? The instructions are here.
If you cannot compile this yourself, I could build you a library that does not require cuda, but you'll have to wait for me to do this until the weekend. Compiling yourself is pretty easy.
I am not sure how to build without GPU. The instructions says:All you need to do here is swap the CMake file. Rename cpu_CMakeLists.txt to CMakeLists.txt.
. But there is no _cpuCMakeLists.txt file in the repo.
It’s in the src folder.
Did you see the file in the src folder? Were you able to compile?
Did you see the file in the src folder? Were you able to compile?
Yes, I see it in src folder. I haven’t built yet. Will report back once I try.
I was able to build without GPU and got the demo scripts to work. It would be great if the releases included pre-built DLL for use without GPU. Or the existing built DLL handled situations where a user does not have a GPU.
Glad it worked!
We are definitely interested in making the build process easier, but this type of thing is not really in my skill set. I am hoping someone from the LEAP user community will step up and help us out with this.
I am going to close this issue, but feel free to open another one if you have further issues.
I used the manual install method (i.e. copying the libleapct.dll file instead of compiling the source) as instructed in the manual_install.py. I am unable to run any of the demo scripts after manual installation. I have confirmed that the files,
leapctype.py
,leaptorch.py
,leap_filter_sequence.py
,leap_preprocessing_algorithms.py
, andlibleapct.dll
are properly copied to my Python's virtual environment path after running the manual_install.py script:System info: Windows 10 x64 Python 3.11.9
Below are my steps to repeat: