Hi,
I get the following error while running Alphafold 2.3.2:
2023-06-17 03:43:50.734422: W external/org_tensorflow/tensorflow/tsl/framework/bfc_allocator.cc:290] Allocator (GPU_0_bfc) ran out of memory trying to allocate 1.12GiB with freed_by_count=0.
Please find below my script:
#!/bin/bash
#SBATCH -p gpu
#SBATCH --gres=gpu:1g.5gb:11
#SBATCH --cpus-per-task=45
#SBATCH --mem=450G
#SBATCH -A protein_structure_prediction_abca4
#SBATCH --time 3-00:00:00
module load alphafold/2.3.2
mkdir -p /tmp/apereira_alphafold
srun run_alphafold.sh --fasta_paths=/shared/projects/protein_structure_prediction_abca4/sequences/hABCA4_201_AA_WT.fasta
--output_dir=/shared/projects/protein_structure_prediction_abca4/output
--model_preset=monomer
--db_preset=full_dbs
--data_dir=/shared/bank/alphafold2/current/2.3/
--uniref90_database_path=/shared/bank/alphafold2/current/2.3/uniref90/uniref90.fasta
--mgnify_database_path=/shared/bank/alphafold2/current/2.3/mgnify/mgy_clusters_2022_05.fa
--pdb70_database_path=/shared/bank/alphafold2/current/2.3/pdb70/pdb70
--template_mmcif_dir=/shared/bank/alphafold2/current/2.3/pdb_mmcif/mmcif_files
--max_template_date=2020-12-31
--obsolete_pdbs_path=/shared/bank/alphafold2/current/2.3/pdb_mmcif/obsolete.dat
--bfd_database_path=/shared/bank/alphafold2/current/2.3/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt
--uniclust30_database_path=/shared/bank/alphafold2/current/2.3/uniref30/UniRef30_2021_03
--use_gpu_relax=false
I cannot understand where the error comes from. Are the memory demands insufficient? Am I not using the GPU to its full capacity?
Thanks a lot for your help.
Best wishes,
Allwyn