Vina Docking speed using binary

[aashuph16221]

Hello,

I have installed the Vina 1.2.3 binary (vina_1.2.3_linux_x86_64) on an HPC system. I ran the program using 1 node with both 1 core and 16 cores, setting the exhaustiveness to 8, but observed almost the same runtime in both cases.

Is there a way to assign each ligand to a separate core for docking? For example, if I use 8 cores, can I configure it to dock 8 ligands simultaneously, one on each core?

Would compiling from source help improve the speed? I’ve tried increasing the exhaustiveness, but it hasn’t led to a significant improvement. My batch size is 10,000 ligands, and each node has 64 cores.

Thanks and regards, Aashish

[diogomart]

Hello,

I ran the program using 1 node with both 1 core and 16 cores, setting the exhaustiveness to 8, but observed almost the same runtime in both cases.

Was this using the HPC manager (e.g. slurm) or the --cpu option in Vina? The difference should have been noticeable. Some clusters allow you to use all the resources in a node even if you don't request them. Vina doesn't use more threads than the exhaustiveness value, so in your case it would spawn a maximum of 8 threads.

[aashuph16221]

I was using the Slurm option to allocate resources. From what I understand, setting the exhaustiveness to 8 should mean Vina will use at least 8 cores.

[diogomart]

No, it means it will spawn 8 threads or less.

[aashuph16221]

Thank you for the information.