I did follow the instruction as per requirement i installed CUDA 10.2, and TensorFlow - 1.15.4 and Python 3.6.9 and to make sure all installation went right i did run the run_all_tests.sh and it was successful with no errors, but still when i run the main script with train/ infer/ eval any of these modes the GPU is being assigned right but it does not use it or speed up the process (using a single 16gb v100 GPU and tried with horovod ), so is there anything else i need to do to make the GPU work?
I did follow the instruction as per requirement i installed CUDA 10.2, and TensorFlow - 1.15.4 and Python 3.6.9 and to make sure all installation went right i did run the run_all_tests.sh and it was successful with no errors, but still when i run the main script with train/ infer/ eval any of these modes the GPU is being assigned right but it does not use it or speed up the process (using a single 16gb v100 GPU and tried with horovod ), so is there anything else i need to do to make the GPU work?