rsanchezgarc/deepEMhancer

Use specific GPU

jianhaoc opened this issue · 1 comments

jianhaoc commented

Hi, I failed to use specific GPU to run deepEMhancer. GPU 0 is occupied and I would like to use GPU 1. How can I set it up? Below is what I have tried but failed. Thank you!

$ conda activate deepEMhancer_env
$ export PYTHONNOUSERSITE=True
$ TF_FORCE_GPU_ALLOW_GROWTH='true'
$ deepemhancer -p highRes -i cryosparc_P217_J1195_003_volume_map.mrc -o cryosparc_P217_J1195_003_volume_map-deepEMhancer-highRes.mrc
updating environment to select gpu: [0]
Using TensorFlow backend.
2021-02-15 14:27:29.617977: E tensorflow/stream_executor/cuda/cuda_driver.cc:828] failed to allocate 125.62M (131727360 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY: out of memory
2021-02-15 14:27:31.373303: F ./tensorflow/core/kernels/random_op_gpu.h:227] Non-OK-status: CudaLaunchKernel(FillPhiloxRandomKernelLaunch, num_blocks, block_size, 0, d.stream(), gen, data, size, dist) status: Internal: out of memory
Aborted (core dumped)

rsanchezgarc commented

Hi

Did you try the -g 1 option?

 deepemhancer -p highRes -i cryosparc_P217_J1195_003_volume_map.mrc -o cryosparc_P217_J1195_003_volume_map-deepEMhancer-highRes.mrc -g 1