I'm trying to use this repository to compute the optical flow between two 3D volumes for application in tomography. The two ndarrays are 128X128X128 pixels. However, I keep getting the following error:
LaunchError: cuLaunchKernel failed: too many resources requested for launch
The error seems to to be on line 58 of _call_smooth_cuda_gauss:
~/.local/lib/python3.6/site-packages/farneback3d/_filtering.py in _call_smooth_cuda_gauss(gpuimg, gpuresult, sigma, kernelsize, filter_mask)
56 np.float32(sigma * sigma),
57 np.int32(filter_nonzeros_only),
---> 58 grid=grid, block=block)
59 else:
60 convolve3d = mod.get_function("convolve3d_gauss_with_mask")
I can send the entire error log if it will be useful.
Could you please let me know what the problem could be.
Hi
I'm trying to use this repository to compute the optical flow between two 3D volumes for application in tomography. The two ndarrays are 128X128X128 pixels. However, I keep getting the following error: LaunchError: cuLaunchKernel failed: too many resources requested for launch
The error seems to to be on line 58 of _call_smooth_cuda_gauss: ~/.local/lib/python3.6/site-packages/farneback3d/_filtering.py in _call_smooth_cuda_gauss(gpuimg, gpuresult, sigma, kernelsize, filter_mask) 56 np.float32(sigma * sigma), 57 np.int32(filter_nonzeros_only), ---> 58 grid=grid, block=block) 59 else: 60 convolve3d = mod.get_function("convolve3d_gauss_with_mask")
I can send the entire error log if it will be useful.
Could you please let me know what the problem could be.
Thanks.