search

DeltaGroupNJUPT / Vina-GPU

A heterogeneous OpenCL implementation of AutoDock Vina
Apache License 2.0
83 stars 20 forks source link

exhaustiveness and threads relations #22

Closed igor611 closed 2 years ago

igor611 commented 2 years ago

Hi! I'm using CPU version of Vina (v 1.2.3) with exhaustiveness=8 and all available CPU on linux (12 cores on AMD 5990x). then I use Vina-GPU, how much "threads" i need to set to get the same (or related) docking accuracy? how exhaustiveness *cpus is related to threads?

Glinttsd commented 2 years ago

Hi, that's a good question.

First, in Vina, all exhaustiveness will be carried out by cpus that you set, the number of cpus does not affect the docking accuracy.

Second, in Vina-GPU, "thread=8000" should be enough to get related docking accuracy compared with "exhaustiveness=128" in Vina. However, we suggest using at least 1000 thread to get promising results.

Third, so, it is exhuastivensee that has positive correlation to thread (in terms of the docking accuracy), not exhaustiveness * cpus.

Hi! I'm using CPU version of Vina (v 1.2.3) with exhaustiveness=8 and all available CPU on linux (12 cores on AMD 5990x). then I use Vina-GPU, how much "threads" i need to set to get the same (or related) docking accuracy? how exhaustiveness *cpus is related to threads?

igor611 commented 2 years ago

Thank you a lot! I have a question, can I ask it here, or must start new issue in git? Then using vina-GPU at nvidia Tesla k80 I see, that only one GPU is working, the second is resting any way. This Tesla has 2 gpus. That to do? Must I compile vina-GPU with some parameters that force it to use GPU[1]? And then I will use 2 copies of Vina-GPU: one uses GPU[0] and the second one uses GPU [1]?

вт, 28 июн. 2022 г., 04:44 Tang Shidi @.***>:

Hi, that's a good question.

First, in Vina, all exhaustiveness will be carried out by cpus that you set, the number of cpus does not affect the docking accuracy.

Second, in Vina-GPU, "thread=8000" should be enough to get related docking accuracy compared with "exhaustiveness=128" in Vina. However, we suggest using at least 1000 thread to get promising results.

Third, so, it is exhuastivensee that has positive correlation to thread (in terms of the docking accuracy), not exhaustiveness * cpus.

Hi! I'm using CPU version of Vina (v 1.2.3) with exhaustiveness=8 and all available CPU on linux (12 cores on AMD 5990x). then I use Vina-GPU, how much "threads" i need to set to get the same (or related) docking accuracy? how exhaustiveness *cpus is related to threads?

— Reply to this email directly, view it on GitHub https://github.com/DeltaGroupNJUPT/Vina-GPU/issues/22#issuecomment-1168114712, or unsubscribe https://github.com/notifications/unsubscribe-auth/AUCJA3Z4AFFNTWMFM3R346TVRJKIDANCNFSM5Z7SQFRQ . You are receiving this because you authored the thread.Message ID: @.***>

Glinttsd commented 2 years ago

Thank you a lot! I have a question, can I ask it here, or must start new issue in git? Then using vina-GPU at nvidia Tesla k80 I see, that only one GPU is working, the second is resting any way. This Tesla has 2 gpus. That to do? Must I compile vina-GPU with some parameters that force it to use GPU[1]? And then I will use 2 copies of Vina-GPU: one uses GPU[0] and the second one uses GPU [1]? вт, 28 июн. 2022 г., 04:44 Tang Shidi @.>: … Hi, that's a good question. First, in Vina, all exhaustiveness will be carried out by cpus that you set, the number of cpus does not affect the docking accuracy. Second, in Vina-GPU, "thread=8000" should be enough to get related docking accuracy compared with "exhaustiveness=128" in Vina. However, we suggest using at least 1000 thread to get promising results. Third, so, it is exhuastivensee that has positive correlation to thread (in terms of the docking accuracy), not exhaustiveness cpus. Hi! I'm using CPU version of Vina (v 1.2.3) with exhaustiveness=8 and all available CPU on linux (12 cores on AMD 5990x). then I use Vina-GPU, how much "threads" i need to set to get the same (or related) docking accuracy? how exhaustiveness cpus is related to threads? — Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AUCJA3Z4AFFNTWMFM3R346TVRJKIDANCNFSM5Z7SQFRQ . You are receiving this because you authored the thread.Message ID: @.>

23 check this out : )