How to install tigre？

I use this “python setup.py install” to install tigre, but the error occured, as follow:
nvcc fatal : Unsupported gpu architecture 'compute_86'
error: command '/usr/local/cuda-11.0/bin/nvcc' failed with exit code 1
what can I do?
I try to add environment variable “ export TORCH_CUDA_ARCH_LIST="8.0" ” to ~/.bashrc, but nothing help.

Hi!
There was a tiny mistake in the code for CUDA 11.0 specifically. I just updated master, it should be fixed now.

sorry, I also have this question as follow:

nvcc fatal : Unsupported gpu architecture 'compute_86'
error: command '/usr/local/cuda-11.0/bin/nvcc' failed with exit code 1

@Love-Sunshine TIGRE isntallation was just updated yesterday, did you use the latest version?

These errors of unsupported architecture are minor bugs in the installer, so I can fix easy if I know which version exactly you are using.

Yes, I just download the latest version, but I still fail. The GPU uses NVIDIA TITAN Xp and NVIDIA RTX A5000, which are located on two devices respectively, and both devices' systems are Ubuntu 18.04. The version of CUDA is cuda_11.0.
I don't know if this information is sufficient.

@Love-Sunshine that was perfect info. I just updated setup.py, hopefully it works now.

Sorry, I have encountered a new issue as follows:
ValueError: flag not understood, only ---no_pinned_memory accepted.
Can you help me? Thanks a lot.

TIGRE/Python/setup.py

Lines 19 to 26 in 7510dff

    
           if len(sys.argv)>2: 
        
               if "--no_pinned_memory" in sys.argv[2:] : 
        
                   no_pinned=True 
        
                   sys.argv.pop(sys.argv.index("--no_pinned_memory")) 
        
               elif "--user" in sys.argv[2:]: 
        
                   pass 
        
               else: 
        
                   raise ValueError("flag not understood, only ---no_pinned_memory accepted")

Remove L. 23-26 from setup.py and retry.

Thank you very much for your help. I am now able to install it. But I still encountered a small problem. When I use jupyter notebook to run "d03_generateData.py", "tigre.plotproj(projections)" and "tigre.plotproj(projections - noise_projections)" cannot display the drawn graph. How can I solve this problem?

Sorry, I don't use jupyter for tigre and I don't know much about it.

The tigre.plotproj function draws the images on a plot area again and again. I'm not sure whether jupyter supprts that kind of displaying.
I guess you can run the script directly from the console:

python demos\d03_generateData.py

Thanks a lot, I will try it again.

	if len(sys.argv)>2:
	if "--no_pinned_memory" in sys.argv[2:] :
	no_pinned=True
	sys.argv.pop(sys.argv.index("--no_pinned_memory"))
	elif "--user" in sys.argv[2:]:
	pass
	else:
	raise ValueError("flag not understood, only ---no_pinned_memory accepted")