GRID K1, CUDA, Machine learning


I want to perform some Machine Learning tasks on a GRID K1. TensorFlow does not recognize my graphic card. I have Cuda 12.2. and some tasks gives the following error : Cuda compute capability 3.0. The minimum required Cuda capability is 3.5.

Does anyone know how to perform parallelism computing with Python on GRID K1

Thank you,

CUDA 12.x cannot be used with that GPU.

The last version of CUDA that supported that GPU was CUDA 10.x. You would also have to find a version of TF that supported a cc3.0 GPU. They have not built TF like that for quite some time.

if you do a google search on tensorflow compute capability 3.0 you will find relevant articles. I suspect this would be a difficult endeavor.