search

src-d / kmcuda

Large scale K-means and K-nn implementation on NVIDIA GPU / CUDA
Other
797 stars 146 forks source link

compute capability mismatch for device 0: wanted 6.1, have 7.5 #92

Open ahbon123 opened 5 years ago

ahbon123 commented 5 years ago

Hello, I have a similar issue, I have run command CUDA_ARCH=75 pip install git+https://github.com/src-d/kmcuda.git#subdirectory=src --user, but still not work.

Traceback (most recent call last):

  File "<ipython-input-7-fc4306ec9a9d>", line 11, in <module>
    centroids, assignments = kmeans_cuda(arr, 4, verbosity=1, seed=3)

ValueError: No such CUDA device exists

reassignments threshold: 100
compute capability mismatch for device 0: wanted 6.1, have 7.5
>>>> you may want to build kmcuda with -DCUDA_ARCH=75 (refer to "Building" in README.md)
tsologub commented 5 years ago

@ahbon123 have you solved this issue? any help on this?

tsologub commented 5 years ago

This issue is fixable with installing libKMCUDA via build instructions.

patzm commented 4 years ago

I think this is resolved with #97. However, this PR is not included in a tagged release just yet. Can you try

pip install "libKMCUDA @ git+https://github.com/src-d/kmcuda@7199d25289ea91a1c8bf2b45654407f4e12cdbff#egg=libKMCUDA&subdirectory=src"