- Posts: 54
gromacs single precision GPU
- Luca
-
Topic Author
- Visitor
9 years 1 month ago #4495
by Luca
gromacs single precision GPU was created by Luca
How much do i lose if i run a Martini MD simulation with single precision gromacs instead of double precision gromacs?
What if i run a double precision energy minimization (that need double precision) on a pc and then run the md simulation on a different pc with single precision gromacs but a GPU acceleration?
Because i have a long MD simulation to run but i dont know if a sigle precision MD simulation can give me a good result ( RMSD, etc... ).
thank you
What if i run a double precision energy minimization (that need double precision) on a pc and then run the md simulation on a different pc with single precision gromacs but a GPU acceleration?
Because i have a long MD simulation to run but i dont know if a sigle precision MD simulation can give me a good result ( RMSD, etc... ).
thank you
Please Log in or Create an account to join the conversation.
- helgi
-
- Offline
- Administrator
Less
More
9 years 1 month ago #4497
by helgi
Replied by helgi on topic gromacs single precision GPU
Due to the large size of the Martini beads you can safely use single precision gromacs for all Martini simulations. We have only done a few tests of Martini with GPU acceleration and those tests only gave minimal speedup so if you get this working please let us know.
Cheers,
- Helgi
Cheers,
- Helgi
Please Log in or Create an account to join the conversation.
- Luca
-
Topic Author
- Visitor
9 years 1 month ago #4498
by Luca
Replied by Luca on topic gromacs single precision GPU
Can you tell me how much faster is running a simulation with single precision GPU accelerated martini?
Because in Gromacs site there's a GPU acceleration article with charts on how much faster the md is with gpu in terms of nanoseconds x day and looks like i cana get double the speed.
I will try to run the same simulation with single GPU and double precision and see if the RMSDs differ, i think i can use some statistic to compare the two.
i'll let you know.
Because in Gromacs site there's a GPU acceleration article with charts on how much faster the md is with gpu in terms of nanoseconds x day and looks like i cana get double the speed.
I will try to run the same simulation with single GPU and double precision and see if the RMSDs differ, i think i can use some statistic to compare the two.
i'll let you know.
Please Log in or Create an account to join the conversation.
Time to create page: 0.098 seconds